Port forward to the named service ports.

When adding a service via the wizard, resolve the service's targetPort to the actual pod container port instead of using the service port directly.
Problem: Service port 80 → Pod port 8000, but kportal was trying to forward to port 80 on the pod.
Solution: Look up the pod's actual containerPort when the service uses a named targetPort (like http), and use that for port-forwarding.

.goreleaser.yaml

@@ -19,7 +19,7 @@ builds:
       - arm64
     ldflags:
       - -s -w
-      - -X main.version={{.Version}}
+      - -X main.appVersion={{.Version}}
 
 archives:
   - id: kportal

.kportal.yaml

@@ -1,6 +1,27 @@
 # Example kportal configuration
 # Copy this file to your project and customize as needed
 
+# Optional: Health check configuration
+# These settings control how kportal monitors connection health and detects stale connections
+healthCheck:
+  interval: "3s"              # How often to check connection health (default: 3s)
+  timeout: "2s"               # Timeout for health check operations (default: 2s)
+  method: "data-transfer"     # Health check method: "tcp-dial" or "data-transfer" (default: data-transfer)
+                              # - tcp-dial: Simple TCP connection test (fast, less reliable)
+                              # - data-transfer: Attempts to read data (slower, more reliable)
+  maxConnectionAge: "25m"     # Maximum connection age before proactive reconnect (default: 25m)
+                              # Helps avoid Kubernetes API server timeouts (typically 30m)
+  maxIdleTime: "10m"          # Maximum idle time before marking as stale (default: 10m)
+                              # Connections with no data transfer are marked stale
+
+# Optional: Reliability configuration
+# These settings improve connection stability for long-running transfers
+reliability:
+  tcpKeepalive: "30s"         # TCP keepalive interval for OS-level connection monitoring (default: 30s)
+  dialTimeout: "30s"          # Connection dial timeout (default: 30s)
+  retryOnStale: true          # Automatically reconnect when stale connections detected (default: true)
+  watchdogPeriod: "30s"       # Goroutine watchdog check interval to detect hung workers (default: 30s)
+
 contexts:
   # Production context
   - name: production

Makefile

@@ -37,7 +37,7 @@ GOFMT=$(GOCMD) fmt
 
 # Build flags
 BUILD_FLAGS=-buildvcs=false
-LDFLAGS=-ldflags="-s -w -X main.version=$(VERSION)"
+LDFLAGS=-ldflags="-s -w -X main.appVersion=$(VERSION)"
 
 all: fmt vet staticcheck test build
 

README.md

@@ -9,33 +9,25 @@
 </p>
 
 <p align="center">
-  <strong>Modern Kubernetes port-forward manager with interactive terminal UI</strong>
+  <strong>Kubernetes port-forward manager with interactive terminal UI</strong>
 </p>
 
-kportal simplifies managing multiple Kubernetes port-forwards with an elegant, interactive terminal interface. Built with [Bubble Tea](https://github.com/charmbracelet/bubbletea), it provides real-time status updates, automatic reconnection, and hot-reload configuration support.
+kportal manages multiple Kubernetes port-forwards with an interactive terminal interface. It provides real-time status updates, automatic reconnection, hot-reload configuration, and mDNS hostname publishing.
 
 ![kportal Screenshot](docs/kportal-screenshot.png)
 
 ## ✨ Features
 
-- 🎯 **Interactive TUI** - Beautiful terminal interface with keyboard navigation (↑↓/jk, Space to toggle, q to quit)
-- ➕ **Live Add** - Add new port-forwards on-the-fly without editing config files or restarting
-- ✏️ **Live Edit** - Modify existing port-forwards (ports, resources, aliases) in real-time
-- 🗑️ **Live Delete** - Remove port-forwards instantly from the running session
-- 🔄 **Auto-Reconnect** - Automatic retry with exponential backoff on connection failures (max 10s)
-- ⚡ **Hot-Reload** - Update configuration without restarting - changes applied automatically
-- 🏥 **Health Checks** - Real-time port forward status monitoring with 5-second intervals
-- 🎨 **Multi-Context** - Support for multiple Kubernetes contexts and namespaces
-- 📦 **Batch Management** - Manage all port-forwards from a single configuration file
-- 🔌 **Toggle Forwards** - Enable/disable individual port-forwards on the fly with Space key
-- 🚀 **Grace Period** - Smart 10-second grace period to avoid false "Error" status on startup
-- 📊 **Status Display** - Clear visual indicators: Active (●), Starting (○), Reconnecting (◐), Error (✗)
-- 🔍 **Error Reporting** - Detailed error messages displayed below the table
-- 🔄 **Pod Restart Handling** - Detects and reconnects to pods when they restart
-- 🏷️ **Label Selector Support** - Dynamically target pods using label selectors
-- 📋 **Prefix Matching** - Automatically find and reconnect to pods with name prefixes
-- 🚫 **Port Conflict Detection** - Validates port availability before starting with detailed PID info
-- 🎭 **Alias Support** - Cleaner, more readable display names for your forwards
+- **Interactive TUI** - Terminal interface with keyboard navigation
+- **Live management** - Add, edit, and delete port-forwards without restarting
+- **Auto-reconnect** - Exponential backoff retry on connection failures
+- **Hot-reload** - Configuration changes applied automatically
+- **Health monitoring** - Multiple check methods with stale connection detection
+- **Multi-context** - Support for multiple Kubernetes contexts and namespaces
+- **Pod restart handling** - Automatic reconnection when pods restart
+- **Label selectors** - Dynamic pod targeting using label selectors
+- **Port conflict detection** - Validates port availability with PID information
+- **mDNS hostnames** - Access forwards via `.local` hostnames
 
 ## 📦 Installation
 
@@ -45,7 +37,7 @@ kportal simplifies managing multiple Kubernetes port-forwards with an elegant, i
 brew install lukaszraczylo/brew-taps/kportal
 ```
 
-### Quick Install Script
+### Quick Install
 
 ```bash
 curl -fsSL https://raw.githubusercontent.com/lukaszraczylo/kportal/main/install.sh | bash
@@ -53,24 +45,19 @@ curl -fsSL https://raw.githubusercontent.com/lukaszraczylo/kportal/main/install.
 
 ### Manual Download
 
-Download the latest binary for your platform from the [releases page](https://github.com/lukaszraczylo/kportal/releases):
-
-- **macOS**: `kportal-{version}-darwin-{amd64|arm64}.tar.gz`
-- **Linux**: `kportal-{version}-linux-{amd64|arm64}.tar.gz`
-- **Windows**: `kportal-{version}-windows-{amd64|arm64}.zip`
+Download binaries from the [releases page](https://github.com/lukaszraczylo/kportal/releases).
 
 ### Build from Source
 
 ```bash
 git clone https://github.com/lukaszraczylo/kportal.git
 cd kportal
-make build
-make install
+make build && make install
 ```
 
 ## 🚀 Quick Start
 
-1. **Create a configuration file** (`.kportal.yaml`):
+Create `.kportal.yaml`:
 
 ```yaml
 contexts:
@@ -84,32 +71,32 @@ contexts:
             localPort: 5432
             alias: prod-db
 
-      - name: frontend
-        forwards:
           - resource: service/redis
             protocol: tcp
             port: 6379
-            localPort: 6380
-            alias: prod-redis
+            localPort: 6379
 ```
 
-2. **Run kportal**:
+Run:
 
 ```bash
 kportal
 ```
 
-3. **Navigate the interface**:
-   - `↑↓` or `j/k` - Navigate through forwards
-   - `Space` or `Enter` - Toggle forward on/off
-   - `a` - Add new port-forward interactively
-   - `e` - Edit selected port-forward
-   - `d` - Delete selected port-forward
-   - `q` - Quit application
+### Keyboard Controls
+
+| Key | Action |
+|-----|--------|
+| `↑↓` / `j/k` | Navigate |
+| `Space` / `Enter` | Toggle forward |
+| `a` | Add forward |
+| `e` | Edit forward |
+| `d` | Delete forward |
+| `q` | Quit |
 
 ## 📖 Configuration
 
-### Simple Configuration
+### Basic Structure
 
 ```yaml
 contexts:
@@ -117,551 +104,207 @@ contexts:
     namespaces:
       - name: <namespace-name>
         forwards:
-          - resource: <resource-type>/<resource-name>
+          - resource: <type>/<name>
             protocol: tcp
             port: <remote-port>
             localPort: <local-port>
-            alias: <friendly-name>  # Optional
+            alias: <display-name>      # optional
+            selector: <label-selector> # optional
 ```
 
-### Advanced Configuration
+### Forward Options
 
-```yaml
-contexts:
-  # Production cluster
-  - name: prod-us-west
-    namespaces:
-      - name: databases
-        forwards:
-          # Direct pod connection with prefix matching
-          - resource: pod/postgres-primary
-            protocol: tcp
-            port: 5432
-            localPort: 5432
-            alias: prod-postgres
-
-          # Service connection
-          - resource: service/redis-master
-            protocol: tcp
-            port: 6379
-            localPort: 6379
-            alias: prod-redis
-
-          # Pod with label selector
-          - resource: pod
-            selector: app=mongodb
-            protocol: tcp
-            port: 27017
-            localPort: 27017
-            alias: mongo
-
-      - name: applications
-        forwards:
-          - resource: deployment/api-server
-            protocol: tcp
-            port: 8080
-            localPort: 8080
-            alias: api
-
-  # Development cluster
-  - name: dev-local
-    namespaces:
-      - name: default
-        forwards:
-          - resource: service/grafana
-            protocol: tcp
-            port: 3000
-            localPort: 3000
-            alias: grafana-dashboard
-```
-
-### Configuration Options
-
-| Field | Type | Required | Description |
-|-------|------|----------|-------------|
-| `resource` | string | Yes | Kubernetes resource with type prefix (e.g., `service/name`, `pod/name`) |
-| `protocol` | string | Yes | Connection protocol (typically `tcp`) |
-| `port` | int | Yes | Remote port on the Kubernetes resource |
-| `localPort` | int | Yes | Local port to forward to |
-| `alias` | string | No | Friendly name for display (defaults to resource name) |
-| `selector` | string | No | Label selector for dynamic pod selection (e.g., `app=nginx,env=prod`) |
+| Field | Required | Description |
+|-------|----------|-------------|
+| `resource` | Yes | Resource type and name (e.g., `service/postgres`, `pod/my-app`) |
+| `protocol` | Yes | Protocol (`tcp`) |
+| `port` | Yes | Remote port |
+| `localPort` | Yes | Local port |
+| `alias` | No | Display name and mDNS hostname |
+| `selector` | No | Label selector for pod resolution |
 
 ### Resource Formats
 
-- **Pod by name**: `pod/pod-name` or just `pod-name`
-- **Pod by prefix**: `pod/my-app` (matches `my-app-xyz789`, `my-app-abc123`, etc.)
-- **Pod by selector**: Set `resource: pod` and use `selector: app=nginx`
-- **Service**: `service/service-name` or `svc/service-name`
-- **Deployment**: `deployment/deployment-name` or `deploy/deployment-name`
+| Format | Description |
+|--------|-------------|
+| `service/name` | Service forwarding |
+| `pod/name` | Direct pod by name |
+| `pod/prefix` | Pod by prefix (matches `prefix-*`) |
+| `pod` + `selector` | Pod by label selector |
+| `deployment/name` | Deployment |
 
-## 🎮 Usage
+### Health Check Configuration
 
-### Interactive Mode (Default)
+```yaml
+healthCheck:
+  interval: "3s"           # Check frequency
+  timeout: "2s"            # Check timeout
+  method: "data-transfer"  # tcp-dial or data-transfer
+  maxConnectionAge: "25m"  # Reconnect before k8s timeout
+  maxIdleTime: "10m"       # Detect idle connections
+
+reliability:
+  tcpKeepalive: "30s"
+  dialTimeout: "30s"
+  retryOnStale: true
+```
+
+Health check methods:
+- `tcp-dial` - Fast TCP connection test
+- `data-transfer` - Verifies tunnel functionality by attempting data read
+
+Connection age reconnection only triggers when the connection is also idle, preventing interruption of active transfers like database dumps.
+
+### mDNS Hostnames
+
+Enable mDNS to access forwards via `.local` hostnames:
+
+```yaml
+mdns:
+  enabled: true
+
+contexts:
+  - name: production
+    namespaces:
+      - name: default
+        forwards:
+          - resource: service/postgres
+            port: 5432
+            localPort: 5432
+            alias: prod-db  # Accessible via prod-db.local:5432
+```
+
+- Explicit `alias` becomes `<alias>.local`
+- Without alias, hostname is generated from resource name (`service/redis` → `redis.local`)
+- Works on macOS (Bonjour) and Linux (avahi-daemon)
+
+Verify registration:
+```bash
+dns-sd -B _kportal._tcp local       # macOS
+avahi-browse -t _kportal._tcp       # Linux
+```
+
+## Usage
+
+### Interactive Mode
 
 ```bash
 kportal
 ```
 
-Starts the interactive TUI where you can:
-- View all configured port-forwards in a table
-- See real-time status updates (Active, Starting, Reconnecting, Error)
-- Toggle forwards on/off with Space key
-- View detailed error messages at the bottom of the screen
-
 ### Verbose Mode
 
 ```bash
 kportal -v
 ```
 
-Runs in verbose mode with:
-- Detailed logging to stdout
-- Periodic status table updates every 2 seconds
-- Full error traces
-- No interactive controls (for automation/debugging)
-
 ### Validate Configuration
 
 ```bash
 kportal --check
 ```
 
-Validates your configuration file without starting any forwards:
-- Checks YAML syntax
-- Validates all required fields
-- Detects duplicate local ports
-- Shows validation errors with line numbers
-
-### Custom Configuration File
+### Custom Config File
 
 ```bash
 kportal -c /path/to/config.yaml
 ```
 
-### Version Information
+## Status Indicators
+
+| Indicator | Description |
+|-----------|-------------|
+| `● Active` | Connection healthy |
+| `○ Starting` | Initial connection (10s grace period) |
+| `◐ Reconnecting` | Reconnecting after failure |
+| `✗ Error` | Connection failed |
+| `○ Disabled` | Manually disabled |
+
+## Advanced Features
+
+### Hot-Reload
+
+Configuration changes are applied automatically. Manual reload:
 
 ```bash
-kportal --version
-# Output: kportal version 0.1.5
+kill -HUP $(pgrep kportal)
 ```
 
-## 🔄 kftray Migration
+### Port Conflict Detection
 
-Migrate from kftray JSON configuration:
+kportal validates port availability at startup and during hot-reload, showing which process is using conflicting ports.
+
+### Retry Strategy
+
+Exponential backoff: 1s → 2s → 4s → 8s → 10s (max). Retries continue indefinitely until connection succeeds.
+
+## Migration from kftray
 
 ```bash
 kportal --convert configs.json --convert-output .kportal.yaml
 ```
 
-**Example conversion:**
+## Signal Handling
 
-kftray JSON:
-```json
-[
-  {
-    "service": "postgres",
-    "namespace": "default",
-    "local_port": 5432,
-    "remote_port": 5432,
-    "context": "production",
-    "workload_type": "service",
-    "protocol": "tcp",
-    "alias": "prod-db"
-  }
-]
-```
-
-Converts to kportal YAML:
-```yaml
-contexts:
-  - name: production
-    namespaces:
-      - name: default
-        forwards:
-          - resource: service/postgres
-            protocol: tcp
-            port: 5432
-            localPort: 5432
-            alias: prod-db
-```
-
-## 🎨 Status Indicators
-
-| Indicator | Status | Description |
-|-----------|--------|-------------|
-| `● Active` | 🟢 Green | Port-forward is active and healthy |
-| `○ Starting` | 🟡 Yellow | Initial connection in progress (10s grace period) |
-| `◐ Reconnecting` | 🟡 Yellow | Attempting to reconnect after failure |
-| `✗ Error` | 🔴 Red | Connection failed - see error details below table |
-| `○ Disabled` | ⚪ Gray | Port-forward manually disabled via Space key |
-
-## 🛠️ Advanced Features
-
-### Hot-Reload
-
-kportal automatically watches for configuration file changes and reloads:
-
-```bash
-# Edit your config while kportal is running
-vim .kportal.yaml
-
-# Changes are applied automatically within seconds:
-# - New forwards are started
-# - Removed forwards are stopped
-# - Existing forwards continue running unchanged
-```
-
-Supports manual reload via `SIGHUP`:
-```bash
-kill -HUP $(pgrep kportal)
-```
-
-### Health Checks
-
-Built-in health monitoring system:
-- **Check interval**: Every 5 seconds
-- **Timeout**: 2 seconds per check
-- **Grace period**: 10 seconds for new connections
-- **Automatic updates**: Real-time status changes in UI
-- **Error tracking**: Detailed error messages for failed connections
-
-### Error Display
-
-When connections fail, errors are displayed below the table:
-
-```
-Errors:
-  • prod-postgres: dial tcp 127.0.0.1:5432: connect: connection refused
-  • prod-redis: i/o timeout after 2.0s
-```
-
-Errors automatically clear when:
-- Connection becomes healthy
-- Forward is disabled
-- Forward is removed
-
-### Port Conflict Detection
-
-kportal checks for port conflicts at multiple stages:
-
-**At startup:**
-```
-Port conflicts detected:
-  Port 8080:
-    • Requested by: api-server (context: prod, namespace: default)
-    • Currently used by: PID 1234 (chrome)
-```
-
-**During hot-reload:**
-- Only validates new ports being added
-- Skips currently managed ports
-- Rejects configuration if conflicts found
-
-### Pod Restart Handling
-
-When a pod restarts:
-1. Port-forward connection breaks
-2. kportal immediately re-resolves the resource:
-   - For prefix matches: Finds newest pod with matching prefix
-   - For selectors: Re-queries pods with matching labels
-3. Reconnects to new pod
-4. Logs the switch: `Switched to new pod: old-pod-abc → new-pod-xyz`
-
-### Retry Strategy
-
-Exponential backoff with maximum interval:
-- **Intervals**: 1s → 2s → 4s → 8s → 10s (max)
-- **Infinite retries**: Continues until connection succeeds
-- **Independent**: Each forward has its own retry logic
-- **Grace period**: First 10 seconds show "Starting" instead of "Error"
-
-## 🔧 Development
-
-### Prerequisites
-
-- Go 1.23 or higher
-- Access to a Kubernetes cluster
-- kubectl configured with contexts
-
-### Building
-
-```bash
-# Build binary
-make build
-
-# Run tests
-make test
-
-# Run all checks (fmt, vet, staticcheck, test)
-make all
-
-# Check current version
-make version
-
-# Install locally
-make install
-
-# Install system-wide
-sudo make install-system
-
-# Clean build artifacts
-make clean
-```
-
-### Project Structure
-
-```
-kportal/
-├── cmd/kportal/              # Main application entry point
-├── internal/
-│   ├── config/               # Configuration parsing and validation
-│   ├── forward/              # Port-forward manager and workers
-│   │   ├── manager.go        # Orchestrates all forwards
-│   │   ├── worker.go         # Individual forward worker
-│   │   └── port_checker.go   # Port conflict detection
-│   ├── healthcheck/          # Health monitoring system
-│   │   └── checker.go        # Port health checking
-│   ├── k8s/                  # Kubernetes client wrapper
-│   │   ├── client.go         # K8s client management
-│   │   ├── port_forward.go   # Port-forward implementation
-│   │   └── resolver.go       # Resource resolution
-│   ├── retry/                # Retry logic with backoff
-│   │   └── backoff.go        # Exponential backoff
-│   ├── ui/                   # Terminal UI implementations
-│   │   ├── bubbletea_ui.go   # Interactive TUI (Bubble Tea)
-│   │   └── table_ui.go       # Simple table for verbose mode
-│   └── converter/            # kftray JSON converter
-├── Formula/                  # Homebrew formula
-├── .github/workflows/        # CI/CD pipelines
-│   └── release.yml           # Release automation
-├── install.sh                # Installation script
-├── semver.yaml               # Semantic version config
-├── Makefile                  # Build automation
-└── README.md                 # This file
-```
-
-## 📝 Examples
-
-### Database Access
-
-```yaml
-contexts:
-  production:
-    namespaces:
-      databases:
-        - resource: postgres-primary
-          port: 5432
-          local_port: 5432
-          alias: prod-db
-```
-
-Connect with:
-```bash
-kportal  # Start in another terminal
-psql -h localhost -p 5432 -U postgres
-```
-
-### Multiple Services
-
-```yaml
-contexts:
-  dev:
-    namespaces:
-      default:
-        - resource: api
-          port: 8080
-          local_port: 8080
-        - resource: frontend
-          port: 3000
-          local_port: 3000
-        - resource: redis
-          port: 6379
-          local_port: 6379
-```
-
-Access:
-- API: `http://localhost:8080`
-- Frontend: `http://localhost:3000`
-- Redis: `redis-cli -p 6379`
-
-### Cross-Context Setup
-
-```yaml
-contexts:
-  prod-us:
-    namespaces:
-      backend:
-        - resource: api
-          port: 8080
-          local_port: 8080
-          alias: prod-us-api
-
-  prod-eu:
-    namespaces:
-      backend:
-        - resource: api
-          port: 8080
-          local_port: 8081  # Different local port
-          alias: prod-eu-api
-```
-
-Compare APIs across regions simultaneously.
-
-### Debug Multiple Pod Versions
-
-```yaml
-contexts:
-  production:
-    namespaces:
-      default:
-        # Version 1
-        - resource: pod
-          selector: app=myapp,version=v1
-          port: 8080
-          local_port: 8080
-          alias: app-v1
-
-        # Version 2
-        - resource: pod
-          selector: app=myapp,version=v2
-          port: 8080
-          local_port: 8081
-          alias: app-v2
-
-        # Debug port for v2
-        - resource: pod
-          selector: app=myapp,version=v2
-          port: 6060
-          local_port: 6060
-          alias: app-v2-pprof
-```
+- `Ctrl+C` / `SIGTERM` - Graceful shutdown
+- `SIGHUP` - Reload configuration
 
 ## 🐛 Troubleshooting
 
 ### Port Already in Use
 
-**Problem**: `Port 8080: already in use by PID 1234 (chrome)`
-
-**Solutions**:
 ```bash
-# Find the process
-lsof -i :8080
-
-# Kill the process
-kill 1234
-
-# Or use a different local port in config
-local_port: 8081
+lsof -i :<port>
+kill <pid>
 ```
 
 ### Connection Refused
 
-**Problem**: `dial tcp 127.0.0.1:8080: connect: connection refused`
-
-**Common causes**:
-1. **Pod not ready yet** - Wait for status to change from "Starting" → "Active" (10s grace period)
-2. **Wrong port number** - Verify the pod/service actually exposes that port
-3. **Service not exposed** - Check with `kubectl get svc` and `kubectl describe svc <name>`
-
-**Debug**:
-```bash
-# Check pod status
-kubectl get pods -n <namespace>
-
-# Check if port is exposed
-kubectl describe pod <pod-name> -n <namespace>
-
-# Check service endpoints
-kubectl get endpoints <service-name> -n <namespace>
-```
+1. Verify pod is running: `kubectl get pods -n <namespace>`
+2. Verify port is correct: `kubectl describe pod <pod>`
+3. Check service endpoints: `kubectl get endpoints <service>`
 
 ### Context Not Found
 
-**Problem**: `context "prod" not found in kubeconfig`
-
-**Solution**:
 ```bash
-# List available contexts
 kubectl config get-contexts
-
-# Verify context name matches
-kubectl config current-context
-
-# Update your config to use the correct context name
 ```
 
-### Health Check Errors During Startup
+## 🔧 Development
 
-**Problem**: Seeing "Error" status immediately after starting
+### Prerequisites
 
-**This is normal!** kportal has a 10-second grace period. If the connection is still failing after 10 seconds, check:
-- Pod is running: `kubectl get pods`
-- Port is correct in config
-- Network connectivity to cluster
+- Go 1.23+
+- Kubernetes cluster access
+- kubectl configured
 
-### Logs Covering UI
+### Build
 
-**Problem**: Kubernetes client logs appearing over the interactive UI
-
-**This is fixed in v0.1.5+**. The interactive mode now completely suppresses all logs including:
-- Standard Go `log` package
-- Kubernetes `klog` output
-- Any stderr/stdout leakage
-
-If you still see logs, please file an issue!
-
-## 🤝 Contributing
-
-Contributions are welcome! Please see [CONTRIBUTING.md](CONTRIBUTING.md) for guidelines.
-
-1. Fork the repository
-2. Create a feature branch (`git checkout -b feature/amazing-feature`)
-3. Make your changes and add tests
-4. Run checks: `make all`
-5. Commit your changes (follow [semantic commit messages](#semantic-versioning))
-6. Push to the branch (`git push origin feature/amazing-feature`)
-7. Open a Pull Request
-
-### Semantic Versioning
-
-This project uses [semver-gen](https://github.com/lukaszraczylo/semver-generator) for automatic semantic version generation based on git commit messages.
-
-**Version Keywords:**
-- **Patch** (0.0.X): `fix`, `bugfix`, `hotfix`, `patch`, `docs`, `test`, `refactor`
-- **Minor** (0.X.0): `feat`, `feature`, `add`, `enhance`, `update`, `improve`
-- **Major** (X.0.0): `breaking`, `major`, `BREAKING CHANGE`
-
-Example commits:
 ```bash
-git commit -m "feat: add health check grace period"  # Bumps minor version
-git commit -m "fix: resolve port conflict detection" # Bumps patch version
-git commit -m "breaking: change config file format"  # Bumps major version
+make build    # Build binary
+make test     # Run tests
+make all      # fmt, vet, staticcheck, test
+make install  # Install locally
 ```
 
-## 📄 License
+## Contributing
 
-This project is licensed under the MIT License - see the [LICENSE](LICENSE) file for details.
+See [CONTRIBUTING.md](CONTRIBUTING.md) for guidelines.
 
-## 🙏 Acknowledgments
+## License
 
-- Built with [Bubble Tea](https://github.com/charmbracelet/bubbletea) by Charm - An awesome framework for building terminal UIs
-- Styled with [Lipgloss](https://github.com/charmbracelet/lipgloss) - Terminal styling library
-- Inspired by [kftray](https://github.com/hcavarsan/kftray) - Original GUI port-forward manager
-- Uses [client-go](https://github.com/kubernetes/client-go) for Kubernetes integration
-- Version management by [semver-gen](https://github.com/lukaszraczylo/semver-generator)
+MIT License - see [LICENSE](LICENSE).
 
-## 📚 Documentation
+## Acknowledgments
+
+- [Bubble Tea](https://github.com/charmbracelet/bubbletea) - Terminal UI framework
+- [Lipgloss](https://github.com/charmbracelet/lipgloss) - Terminal styling
+- [client-go](https://github.com/kubernetes/client-go) - Kubernetes client
+- [kftray](https://github.com/hcavarsan/kftray) - Inspiration
+
+## Links
 
 - [Website](https://lukaszraczylo.github.io/kportal)
-- [Issue Tracker](https://github.com/lukaszraczylo/kportal/issues)
+- [Issues](https://github.com/lukaszraczylo/kportal/issues)
 - [Releases](https://github.com/lukaszraczylo/kportal/releases)
 - [Changelog](CHANGELOG.md)
-
-## Signal Handling
-
-- `Ctrl+C` / `SIGTERM`: Graceful shutdown (closes all forwards)
-- `SIGHUP`: Reload configuration file
-
----
-
-Made with ❤️ by [Lukasz Raczylo](https://github.com/lukaszraczylo)

RELEASE_SETUP.md

@@ -1,320 +1,126 @@
-# Release Infrastructure Setup Summary
+# Release Infrastructure
 
-This document summarizes all the release infrastructure that has been set up for kportal.
+Documentation for kportal's release automation and distribution.
 
-## ✅ Completed Setup
-
-### 1. GitHub Actions CI/CD Pipeline
+## 🔄 CI/CD Pipeline
 
 **File**: `.github/workflows/release.yml`
 
-**Features**:
-- Multi-platform binary builds (Linux, macOS, Windows - amd64 & arm64)
-- Automatic release creation on version tags
-- Binary archiving (tar.gz for Unix, zip for Windows)
-- SHA256 checksum generation
-- Automated Homebrew formula updates
-- Release notes generation
+The pipeline builds multi-platform binaries, creates GitHub releases, and updates Homebrew on version tags.
+
+### Trigger a Release
 
-**How to trigger**:
 ```bash
-# Commit with semantic versioning keywords
-git commit -m "feat: add new feature"
-
-# Tag the release
 git tag -a v0.2.0 -m "Release v0.2.0"
-
-# Push tags
 git push origin v0.2.0
 ```
 
-The pipeline will automatically:
+The pipeline will:
 1. Build binaries for all platforms
-2. Create GitHub release with binaries
+2. Create GitHub release with binaries and checksums
 3. Update Homebrew tap formula
-4. Generate release notes
 
-### 2. Installation Methods
+## 📦 Installation Methods
 
-#### A. Homebrew Formula
+### Homebrew
 
 **File**: `Formula/kportal.rb`
 
-**Installation command**:
 ```bash
 brew install lukaszraczylo/tap/kportal
 ```
 
-**Note**: Formula is automatically updated by CI/CD pipeline. You'll need to create a separate tap repository:
-1. Create repo: `https://github.com/lukaszraczylo/brew-taps`
-2. Add Formula/kportal.rb to that repo
-3. Set `HOMEBREW_TAP_TOKEN` secret in GitHub repository settings
+Formula is automatically updated by CI/CD. Requires:
+- Tap repository: `https://github.com/lukaszraczylo/brew-taps`
+- Secret: `HOMEBREW_TAP_TOKEN` with `repo` scope
 
-#### B. Quick Install Script
+### Install Script
 
 **File**: `install.sh`
 
-**Features**:
-- Auto-detects OS and architecture
-- Downloads appropriate binary
-- Extracts and installs to /usr/local/bin
-- Verifies installation
-- Colorful output with emoji indicators
-
-**Installation command**:
 ```bash
 curl -fsSL https://raw.githubusercontent.com/lukaszraczylo/kportal/main/install.sh | bash
 ```
 
-#### C. Manual Download
+Auto-detects OS/architecture and installs to `/usr/local/bin`.
 
-Users can download binaries directly from GitHub releases:
-```
-https://github.com/lukaszraczylo/kportal/releases
-```
+### Manual Download
 
-### 3. Documentation
+Download from [releases page](https://github.com/lukaszraczylo/kportal/releases).
 
-#### A. Comprehensive README.md
+## Platform Support
 
-**File**: `README.md`
+| OS | Architecture | Format |
+|----|--------------|--------|
+| Linux | amd64, arm64 | tar.gz |
+| macOS | amd64, arm64 | tar.gz |
+| Windows | amd64, arm64 | zip |
 
-**Contents**:
-- Feature showcase with emojis
-- Multiple installation methods
-- Quick start guide
-- Configuration examples
-- Usage instructions
-- Advanced features documentation
-- Troubleshooting guide
-- Contributing guidelines
+## 🚀 Release Process
 
-#### B. GitHub Pages Website
-
-**File**: `docs/index.html`
-
-**Features**:
-- Modern, responsive design with TailwindCSS
-- Hero section with clear CTA
-- Feature showcase cards
-- Installation guide
-- Configuration examples with syntax highlighting
-- Documentation links
-- Mobile-friendly
-
-**URL** (once enabled): `https://lukaszraczylo.github.io/kportal`
-
-**To enable**:
-1. Go to GitHub repository settings
-2. Pages section
-3. Source: Deploy from a branch
-4. Branch: main
-5. Folder: /docs
-
-### 4. Supporting Files
-
-#### CHANGELOG.md
-**File**: `CHANGELOG.md`
-
-Tracks all changes following Keep a Changelog format. Update this file with each release.
-
-#### CONTRIBUTING.md
-**File**: `CONTRIBUTING.md`
-
-Guidelines for:
-- Bug reporting
-- Feature requests
-- Pull request process
-- Commit message format
-- Development setup
-- Testing guidelines
-
-## 🚀 Release Workflow
-
-### Standard Release Process
-
-1. **Develop features**
+1. **Make changes and test**
    ```bash
-   git checkout -b feature/my-feature
-   # Make changes
-   make test
-   make all
+   make test && make all
    ```
 
-2. **Commit with semantic messages**
-   ```bash
-   git commit -m "feat: add amazing feature"
-   git commit -m "fix: resolve bug in health check"
-   ```
+2. **Update CHANGELOG.md**
 
-3. **Update CHANGELOG.md**
-   ```markdown
-   ## [0.2.0] - 2025-11-24
-   
-   ### Added
-   - Amazing new feature
-   
-   ### Fixed
-   - Bug in health check
-   ```
-
-4. **Tag the release**
+3. **Tag and push**
    ```bash
    git tag -a v0.2.0 -m "Release v0.2.0"
    git push origin main
    git push origin v0.2.0
    ```
 
-5. **CI/CD automatically**:
-   - Builds all binaries
-   - Creates GitHub release
-   - Updates Homebrew formula
-   - Attaches binaries and checksums
+## Version Bumping
 
-### Version Bumping (Semantic Versioning)
+Version determined by commit message keywords:
 
-Version is automatically determined by semver-gen from commit messages:
+| Bump | Keywords |
+|------|----------|
+| Patch (0.0.X) | `fix`, `bugfix`, `docs`, `test`, `refactor` |
+| Minor (0.X.0) | `feat`, `feature`, `add`, `enhance`, `update` |
+| Major (X.0.0) | `breaking`, `major`, `BREAKING CHANGE` |
 
-- **Patch** (0.0.X): `fix`, `bugfix`, `hotfix`, `patch`, `docs`, `test`, `refactor`
-- **Minor** (0.X.0): `feat`, `feature`, `add`, `enhance`, `update`, `improve`
-- **Major** (X.0.0): `breaking`, `major`, `BREAKING CHANGE`
+## Required Secrets
 
-## 📦 Platform Support
+| Secret | Purpose |
+|--------|---------|
+| `GITHUB_TOKEN` | Provided by GitHub Actions |
+| `HOMEBREW_TAP_TOKEN` | Personal access token with `repo` scope |
 
-### Supported Platforms
-
-| OS      | Architecture | Archive Format |
-|---------|-------------|----------------|
-| Linux   | amd64       | tar.gz         |
-| Linux   | arm64       | tar.gz         |
-| macOS   | amd64       | tar.gz         |
-| macOS   | arm64       | tar.gz         |
-| Windows | amd64       | zip            |
-| Windows | arm64       | zip            |
-
-## 🔒 Required GitHub Secrets
-
-For full automation, set these secrets in your GitHub repository:
-
-1. **GITHUB_TOKEN** - Automatically provided by GitHub Actions
-2. **HOMEBREW_TAP_TOKEN** - Personal access token for updating Homebrew tap
-   - Create at: https://github.com/settings/tokens
-   - Permissions needed: `repo` scope
-   - Add to repository secrets
-
-## 📝 Next Steps
+## ⚙️ Initial Setup
 
 ### 1. Enable GitHub Pages
-- Repository Settings → Pages → Source: main branch, /docs folder
 
-### 2. Create Homebrew Tap Repository
+Repository Settings → Pages → Source: main branch, /docs folder
+
+### 2. Create Homebrew Tap
+
 ```bash
-# Create new repository
 gh repo create lukaszraczylo/brew-taps --public
-
-# Clone and set up
-git clone https://github.com/lukaszraczylo/brew-taps
 cd brew-taps
-cp ../kportal/Formula/kportal.rb ./Formula/
-git add Formula/kportal.rb
-git commit -m "Initial formula for kportal"
-git push origin main
+mkdir Formula
+# Formula will be auto-updated by CI
 ```
 
-### 3. Add GitHub Token to Secrets
-- Repository Settings → Secrets and variables → Actions
-- New repository secret
+### 3. Add Token Secret
+
+Repository Settings → Secrets → Actions → New secret:
 - Name: `HOMEBREW_TAP_TOKEN`
-- Value: Your personal access token
-
-### 4. Create First Release
-```bash
-cd kportal
-git add .
-git commit -m "feat: initial release setup"
-git push origin main
-git tag -a v0.1.5 -m "Release v0.1.5"
-git push origin v0.1.5
-```
-
-### 5. Test Installation Methods
-
-After first release, test:
-```bash
-# Homebrew (once tap is set up)
-brew install lukaszraczylo/tap/kportal
-
-# Quick install script
-curl -fsSL https://raw.githubusercontent.com/lukaszraczylo/kportal/main/install.sh | bash
-
-# Manual download
-# Visit releases page and download binary
-```
-
-## 🎨 Customization
-
-### Update Website Colors
-
-Edit `docs/index.html`:
-```javascript
-tailwind.config = {
-    theme: {
-        extend: {
-            colors: {
-                primary: '#3b82f6',    // Blue
-                secondary: '#8b5cf6',   // Purple
-                dark: '#0f172a',        // Dark slate
-            }
-        }
-    }
-}
-```
-
-### Update Release Notes Template
-
-Edit `.github/workflows/release.yml` in the "Generate release notes" step.
-
-## 📊 Monitoring
-
-After releases, monitor:
-- GitHub Actions workflow runs
-- GitHub Releases page
-- Homebrew tap repository commits
-- Download statistics on releases page
+- Value: Personal access token with `repo` scope
 
 ## 🐛 Troubleshooting
 
 ### Release workflow fails
 - Check GitHub Actions logs
-- Verify all required secrets are set
-- Ensure tag follows v\d+.\d+.\d+ format
+- Verify secrets are configured
+- Ensure tag follows `v\d+.\d+.\d+` format
 
-### Homebrew formula not updating
-- Verify HOMEBREW_TAP_TOKEN is valid
+### Homebrew not updating
+- Verify `HOMEBREW_TAP_TOKEN` is valid
 - Check tap repository permissions
-- Review release workflow logs
 
 ### Install script fails
-- Test locally with different OS/arch combinations
-- Check release binary naming matches script expectations
-- Verify binaries are attached to release
-
-## ✅ Checklist for First Release
-
-- [ ] All code committed and pushed
-- [ ] GitHub Pages enabled
-- [ ] Homebrew tap repository created
-- [ ] HOMEBREW_TAP_TOKEN secret set
-- [ ] CHANGELOG.md updated
-- [ ] Version tag created and pushed
-- [ ] Release workflow completed successfully
-- [ ] Binaries attached to release
-- [ ] Homebrew formula updated
-- [ ] Install script tested
-- [ ] Documentation website live
-- [ ] README.md installation links work
-
----
-
-**Documentation last updated**: 2025-11-23
-**Setup completed for**: kportal v0.1.5
+- Verify release binaries are attached
+- Check binary naming matches script expectations

WIZARD_USAGE.md

@@ -1,171 +1,104 @@
-# Interactive Add/Remove Wizards
+# Interactive Wizards
 
-kportal now includes interactive wizards for adding and removing port forwards directly from the running UI!
+kportal includes wizards for adding and removing port forwards from the running UI.
 
-## Quick Start
+## ⌨️ Quick Reference
 
-Run kportal normally:
-```bash
-./kportal
-```
+| Key | Action |
+|-----|--------|
+| `a` | Add new forward |
+| `d` | Delete forwards |
 
-From the main view:
-- Press **`n`** to add a new port forward
-- Press **`d`** to delete existing port forwards
+## ➕ Add Forward Wizard
 
-## Add Forward Wizard (`n` key)
+Press `a` from the main view to start the wizard.
 
-The wizard guides you through 7 steps to add a new forward:
+### Steps
 
-### Step 1: Select Context
-Choose from available Kubernetes contexts in your kubeconfig.
+1. **Context** - Select Kubernetes context
+2. **Namespace** - Select namespace
+3. **Resource Type** - Choose pod (prefix), pod (selector), or service
+4. **Resource** - Enter prefix, selector, or select service
+5. **Remote Port** - Enter port on the resource
+6. **Local Port** - Enter local port (validates availability)
+7. **Confirm** - Review and optionally add an alias
 
-### Step 2: Select Namespace
-Pick the namespace where your resource lives.
+### Navigation
 
-### Step 3: Select Resource Type
-Three options:
-- **Pod (by name prefix)** - Forward to a specific pod by prefix matching
-- **Pod (by label selector)** - Forward to pods matching labels (survives restarts)
-- **Service** - Most stable, load-balanced option
+| Key | Action |
+|-----|--------|
+| `↑↓` / `j/k` | Navigate options |
+| `Enter` | Confirm and proceed |
+| `Esc` | Go back / Cancel |
+| `Ctrl+C` | Cancel immediately |
 
-### Step 4: Enter Resource
-- **Pod prefix**: Type a prefix like `nginx-` to match pods
-- **Label selector**: Enter labels like `app=nginx,env=prod`
-- **Service**: Select from a list of services
+## 🗑️ Delete Forward Wizard
 
-The wizard shows real-time validation and matching resources!
+Press `d` from the main view.
 
-### Step 5: Remote Port
-Enter the port number on the remote resource. The wizard displays detected ports from running containers.
+### Navigation
 
-### Step 6: Local Port
-Enter the local port to bind to. The wizard checks availability in real-time.
+| Key | Action |
+|-----|--------|
+| `↑↓` / `j/k` | Navigate |
+| `Space` | Toggle selection |
+| `a` | Select all |
+| `n` | Deselect all |
+| `Enter` | Confirm deletion |
+| `Esc` | Cancel |
 
-### Step 7: Confirmation
-Review your configuration and optionally add an alias (friendly name). Confirm to save!
+## 🎯 Resource Selection
 
-### Navigation Keys
+### Pod by Prefix
 
-- **`↑`/`↓`** or **`j`/`k`** - Navigate options
-- **`Enter`** - Confirm and proceed to next step
-- **`Esc`** - Go back one step (or cancel on first step)
-- **`Ctrl+C`** - Hard cancel and return to main view
-- **`Backspace`** - Delete characters in text fields
-
-## Remove Forward Wizard (`d` key)
-
-Multi-select interface for removing forwards:
-
-1. **Select forwards**: Use arrow keys to navigate, `Space` to toggle selection
-2. **Confirm removal**: Press `Enter` and confirm your choice
-
-### Navigation Keys
-
-- **`↑`/`↓`** or **`j`/`k`** - Navigate forwards
-- **`Space`** - Toggle selection of current forward
-- **`a`** - Select all forwards
-- **`n`** - Deselect all forwards
-- **`Enter`** - Proceed to confirmation
-- **`Esc`** - Cancel and return to main view
-- **`Ctrl+C`** - Hard cancel
-
-## Auto Hot-Reload
-
-When you save a forward via the wizard:
-1. The wizard writes to `.kportal.yaml` atomically
-2. The file watcher detects the change (~100ms)
-3. The manager reloads and starts the new forward
-4. The UI updates automatically
-
-No restart needed!
-
-## Error Handling
-
-The wizards handle errors gracefully:
-
-- **Cluster unreachable**: Shows error but allows manual entry
-- **Port conflicts**: Displays which process is using the port
-- **Invalid selectors**: Shows validation errors in real-time
-- **Duplicate ports**: Prevents adding forwards with conflicting ports
-
-## Tips
-
-### Pod Prefix Matching
-When using pod prefix, you can type just the app name:
+Enter app name prefix to match pods:
 - `nginx` matches `nginx-deployment-abc123`
 - `postgres` matches `postgres-statefulset-0`
 
-### Label Selectors
-Use standard Kubernetes label syntax:
-- `app=nginx` - Single label
-- `app=nginx,env=prod` - Multiple labels (comma-separated)
-- Real-time validation shows matching pods as you type!
+### Pod by Selector
 
-### Aliases
-Use aliases for cleaner UI display:
-- Instead of: `production/default/pod/nginx-deployment-abc123:80→8080`
-- Shows as: `my-nginx:80→8080`
+Use Kubernetes label syntax:
+- `app=nginx`
+- `app=nginx,env=prod`
 
-### Quick Selection
-In list views, you can use `j`/`k` (Vim-style) or arrow keys for navigation.
+Matching pods are shown in real-time.
 
-## Example Workflow
+### Service
 
-Adding a forward for a PostgreSQL database:
+Select from discovered services in the namespace.
 
-1. Press `n` in main view
-2. Select context: `production` (arrow keys + Enter)
-3. Select namespace: `default` (arrow keys + Enter)
-4. Select type: `Service` (arrow keys + Enter)
-5. Select service: `postgres` (arrow keys + Enter)
-6. Enter remote port: `5432` (type + Enter)
-7. Enter local port: `5432` (type + Enter)
-8. Add alias: `prod-db` (optional, type + Enter)
-9. Confirm: Select "Add to .kportal.yaml" (Enter)
+## 🔄 Auto Hot-Reload
 
-Done! The forward starts automatically within seconds.
+Changes are applied automatically:
+1. Wizard writes to `.kportal.yaml` atomically
+2. File watcher detects change (~100ms)
+3. Manager reloads and starts forward
+4. UI updates
 
-## Architecture
+## Error Handling
 
-The wizards use:
-- **Config Mutator**: Safe, atomic YAML writes (temp file + rename)
-- **K8s Discovery**: Lists contexts, namespaces, pods, services
-- **Modal Overlays**: Wizards appear centered over the main view
-- **Async Validation**: Port checks and selector validation run in background
-- **Hot-Reload Integration**: File watcher picks up changes automatically
+The wizards handle:
+- Cluster unreachable - allows manual entry
+- Port conflicts - shows which process is using the port
+- Invalid selectors - real-time validation
+- Duplicate ports - prevents conflicts
 
-## Troubleshooting
+## 🐛 Troubleshooting
 
-### Wizards not appearing?
-Check that kportal can connect to your Kubernetes cluster:
+### Wizard not appearing
+
+Verify cluster connectivity:
 ```bash
 kubectl cluster-info
 ```
 
-### Port check showing wrong status?
-The port check happens asynchronously. Wait a moment after typing for validation.
+### Port validation delayed
 
-### Changes not appearing?
-The file watcher triggers within 100ms. If changes aren't visible, check:
+Port checks run asynchronously. Wait briefly after typing.
+
+### Changes not visible
+
+Check:
 1. `.kportal.yaml` was written correctly
-2. No validation errors in the file
-3. kportal process is still running
-
----
-
-**Navigation Summary**
-
-Main View:
-- `n` - New forward wizard
-- `d` - Delete forward wizard
-- `Space` - Toggle forward on/off
-- `↑↓/jk` - Navigate forwards
-- `q` - Quit
-
-Wizards:
-- `Enter` - Next step / Confirm
-- `Esc` - Previous step / Cancel
-- `Ctrl+C` - Hard cancel
-- `↑↓/jk` - Navigate
-- `Space` - Toggle (in delete wizard)
+2. No validation errors in file
+3. kportal process is running

cmd/kportal/main.go

@@ -1,6 +1,7 @@
 package main
 
 import (
+	"context"
 	"flag"
 	"fmt"
 	"io"
@@ -18,7 +19,9 @@ import (
 	"github.com/nvm/kportal/internal/forward"
 	"github.com/nvm/kportal/internal/k8s"
 	"github.com/nvm/kportal/internal/logger"
+	"github.com/nvm/kportal/internal/mdns"
 	"github.com/nvm/kportal/internal/ui"
+	"github.com/nvm/kportal/internal/version"
 	"k8s.io/klog/v2"
 )
 
@@ -26,6 +29,10 @@ const (
 	defaultConfigFile        = ".kportal.yaml"
 	initialForwardSettleTime = 100 * time.Millisecond
 	tableUpdateInterval      = 2 * time.Second
+
+	// GitHub repository info for update checks
+	githubOwner = "lukaszraczylo"
+	githubRepo  = "kportal"
 )
 
 var (
@@ -34,16 +41,22 @@ var (
 	logFormat     = flag.String("log-format", "text", "Log format: text or json")
 	check         = flag.Bool("check", false, "Validate configuration and exit")
 	showVersion   = flag.Bool("version", false, "Show version and exit")
+	checkUpdate   = flag.Bool("update", false, "Check for updates and exit")
 	convertInput  = flag.String("convert", "", "Convert kftray JSON config to kportal YAML (provide input file path)")
 	convertOutput = flag.String("convert-output", ".kportal.yaml", "Output file for converted configuration")
-	version       = "0.1.0" // Set via ldflags during build
+	appVersion    = "0.1.0" // Set via ldflags during build
 )
 
 func main() {
 	flag.Parse()
 
 	if *showVersion {
-		fmt.Printf("kportal version %s\n", version)
+		fmt.Printf("kportal version %s\n", appVersion)
+		os.Exit(0)
+	}
+
+	if *checkUpdate {
+		checkForUpdates()
 		os.Exit(0)
 	}
 
@@ -177,7 +190,7 @@ func main() {
 
 	// Only log startup messages in verbose mode
 	if *verbose {
-		log.Printf("kportal v%s", version)
+		log.Printf("kportal v%s", appVersion)
 		log.Printf("Loading configuration from: %s", *configFile)
 	}
 
@@ -197,6 +210,14 @@ func main() {
 		os.Exit(1)
 	}
 
+	// Create mDNS publisher if enabled in config
+	mdnsPublisher := mdns.NewPublisher(cfg.IsMDNSEnabled())
+	manager.SetMDNSPublisher(mdnsPublisher)
+
+	if cfg.IsMDNSEnabled() && *verbose {
+		log.Printf("mDNS hostname publishing enabled - aliases will be accessible via <alias>.local")
+	}
+
 	// Create UI (bubbletea for interactive, simple table for verbose)
 	var bubbleTeaUI *ui.BubbleTeaUI
 	var tableUI *ui.TableUI
@@ -209,17 +230,40 @@ func main() {
 			} else {
 				manager.DisableForward(id)
 			}
-		}, version)
+		}, appVersion)
 
 		// Set wizard dependencies
 		// Note: mutator is always available (for delete/edit), discovery requires valid kubeconfig (for add)
 		bubbleTeaUI.SetWizardDependencies(discovery, mutator, *configFile)
 
+		// Check for updates in background (non-blocking)
+		go func() {
+			checker := version.NewChecker(githubOwner, githubRepo, appVersion)
+			ctx, cancel := context.WithTimeout(context.Background(), 10*time.Second)
+			defer cancel()
+
+			if update := checker.CheckForUpdate(ctx); update != nil {
+				bubbleTeaUI.SetUpdateAvailable(update.LatestVersion, update.ReleaseURL)
+			}
+		}()
+
 		manager.SetStatusUI(bubbleTeaUI)
 	} else {
 		// Verbose mode with simple table
 		tableUI = ui.NewTableUI(*verbose)
 		manager.SetStatusUI(tableUI)
+
+		// Check for updates and print to log
+		go func() {
+			checker := version.NewChecker(githubOwner, githubRepo, appVersion)
+			ctx, cancel := context.WithTimeout(context.Background(), 10*time.Second)
+			defer cancel()
+
+			if update := checker.CheckForUpdate(ctx); update != nil {
+				log.Printf("Update available: v%s (current: v%s) - %s",
+					update.LatestVersion, update.CurrentVersion, update.ReleaseURL)
+			}
+		}()
 	}
 
 	// Start forwards
@@ -283,7 +327,23 @@ func main() {
 
 			case os.Interrupt, syscall.SIGTERM:
 				log.Printf("Received shutdown signal, stopping...")
-				manager.Stop()
+
+				// Graceful shutdown with timeout - force exit if it takes too long
+				shutdownDone := make(chan struct{})
+				go func() {
+					manager.Stop()
+					close(shutdownDone)
+				}()
+
+				select {
+				case <-shutdownDone:
+					log.Printf("Graceful shutdown complete")
+				case <-time.After(5 * time.Second):
+					log.Printf("Shutdown timed out, forcing exit...")
+				case sig := <-sigChan:
+					// Second signal received - force exit immediately
+					log.Printf("Received second signal (%v), forcing exit...", sig)
+				}
 				os.Exit(0)
 			}
 		}
@@ -322,3 +382,24 @@ func main() {
 		manager.Stop()
 	}
 }
+
+// checkForUpdates checks for available updates and prints the result
+func checkForUpdates() {
+	fmt.Printf("kportal version %s\n", appVersion)
+	fmt.Println("Checking for updates...")
+
+	checker := version.NewChecker(githubOwner, githubRepo, appVersion)
+	ctx, cancel := context.WithTimeout(context.Background(), 10*time.Second)
+	defer cancel()
+
+	update := checker.CheckForUpdate(ctx)
+	if update == nil {
+		fmt.Println("You are running the latest version.")
+		return
+	}
+
+	fmt.Printf("\nUpdate available: v%s\n", update.LatestVersion)
+	fmt.Printf("Download: %s\n", update.ReleaseURL)
+	fmt.Println("\nTo update, download the latest release from the URL above")
+	fmt.Println("or use your package manager (e.g., 'brew upgrade kportal').")
+}

go.mod

@@ -17,6 +17,7 @@ require (
 
 require (
 	github.com/aymanbagabas/go-osc52/v2 v2.0.1 // indirect
+	github.com/cenkalti/backoff v2.2.1+incompatible // indirect
 	github.com/charmbracelet/colorprofile v0.3.3 // indirect
 	github.com/charmbracelet/x/ansi v0.11.1 // indirect
 	github.com/charmbracelet/x/cellbuf v0.0.14 // indirect
@@ -25,6 +26,7 @@ require (
 	github.com/clipperhouse/stringish v0.1.1 // indirect
 	github.com/clipperhouse/uax29/v2 v2.3.0 // indirect
 	github.com/davecgh/go-spew v1.1.1 // indirect
+	github.com/dustin/go-humanize v1.0.1 // indirect
 	github.com/emicklei/go-restful/v3 v3.13.0 // indirect
 	github.com/erikgeiser/coninput v0.0.0-20211004153227-1c3628e74d0f // indirect
 	github.com/fxamacker/cbor/v2 v2.9.0 // indirect
@@ -47,11 +49,13 @@ require (
 	github.com/google/gnostic-models v0.7.1 // indirect
 	github.com/google/uuid v1.6.0 // indirect
 	github.com/gorilla/websocket v1.5.4-0.20250319132907-e064f32e3674 // indirect
+	github.com/grandcat/zeroconf v1.0.0 // indirect
 	github.com/json-iterator/go v1.1.12 // indirect
 	github.com/lucasb-eyer/go-colorful v1.3.0 // indirect
 	github.com/mattn/go-isatty v0.0.20 // indirect
 	github.com/mattn/go-localereader v0.0.1 // indirect
 	github.com/mattn/go-runewidth v0.0.19 // indirect
+	github.com/miekg/dns v1.1.27 // indirect
 	github.com/moby/spdystream v0.5.0 // indirect
 	github.com/modern-go/concurrent v0.0.0-20180306012644-bacd9c7ef1dd // indirect
 	github.com/modern-go/reflect2 v1.0.3-0.20250322232337-35a7c28c31ee // indirect
@@ -67,6 +71,7 @@ require (
 	github.com/xo/terminfo v0.0.0-20220910002029-abceb7e1c41e // indirect
 	go.yaml.in/yaml/v2 v2.4.3 // indirect
 	go.yaml.in/yaml/v3 v3.0.4 // indirect
+	golang.org/x/crypto v0.44.0 // indirect
 	golang.org/x/net v0.47.0 // indirect
 	golang.org/x/oauth2 v0.33.0 // indirect
 	golang.org/x/sys v0.38.0 // indirect

go.sum

@@ -2,6 +2,8 @@ github.com/armon/go-socks5 v0.0.0-20160902184237-e75332964ef5 h1:0CwZNZbxp69SHPd
 github.com/armon/go-socks5 v0.0.0-20160902184237-e75332964ef5/go.mod h1:wHh0iHkYZB8zMSxRWpUBQtwG5a7fFgvEO+odwuTv2gs=
 github.com/aymanbagabas/go-osc52/v2 v2.0.1 h1:HwpRHbFMcZLEVr42D4p7XBqjyuxQH5SMiErDT4WkJ2k=
 github.com/aymanbagabas/go-osc52/v2 v2.0.1/go.mod h1:uYgXzlJ7ZpABp8OJ+exZzJJhRNQ2ASbcXHWsFqH8hp8=
+github.com/cenkalti/backoff v2.2.1+incompatible h1:tNowT99t7UNflLxfYYSlKYsBpXdEet03Pg2g16Swow4=
+github.com/cenkalti/backoff v2.2.1+incompatible/go.mod h1:90ReRw6GdpyfrHakVjL/QHaoyV4aDUVVkXQJJJ3NXXM=
 github.com/charmbracelet/bubbletea v1.3.10 h1:otUDHWMMzQSB0Pkc87rm691KZ3SWa4KUlvF9nRvCICw=
 github.com/charmbracelet/bubbletea v1.3.10/go.mod h1:ORQfo0fk8U+po9VaNvnV95UPWA1BitP1E0N6xJPlHr4=
 github.com/charmbracelet/colorprofile v0.3.3 h1:DjJzJtLP6/NZ8p7Cgjno0CKGr7wwRJGxWUwh2IyhfAI=
@@ -23,6 +25,8 @@ github.com/clipperhouse/uax29/v2 v2.3.0/go.mod h1:Wn1g7MK6OoeDT0vL+Q0SQLDz/KpfsV
 github.com/davecgh/go-spew v1.1.0/go.mod h1:J7Y8YcW2NihsgmVo/mv3lAwl/skON4iLHjSsI+c5H38=
 github.com/davecgh/go-spew v1.1.1 h1:vj9j/u1bqnvCEfJOwUhtlOARqs3+rkHYY13jYWTU97c=
 github.com/davecgh/go-spew v1.1.1/go.mod h1:J7Y8YcW2NihsgmVo/mv3lAwl/skON4iLHjSsI+c5H38=
+github.com/dustin/go-humanize v1.0.1 h1:GzkhY7T5VNhEkwH0PVJgjz+fX1rhBrR7pRT3mDkpeCY=
+github.com/dustin/go-humanize v1.0.1/go.mod h1:Mu1zIs6XwVuF/gI1OepvI0qD18qycQx+mFykh5fBlto=
 github.com/emicklei/go-restful/v3 v3.13.0 h1:C4Bl2xDndpU6nJ4bc1jXd+uTmYPVUwkD6bFY/oTyCes=
 github.com/emicklei/go-restful/v3 v3.13.0/go.mod h1:6n3XBCmQQb25CM2LCACGz8ukIrRry+4bhvbpWn3mrbc=
 github.com/erikgeiser/coninput v0.0.0-20211004153227-1c3628e74d0f h1:Y/CXytFA4m6baUTXGLOoWe4PQhGxaX0KpnayAqC48p4=
@@ -82,6 +86,8 @@ github.com/google/uuid v1.6.0 h1:NIvaJDMOsjHA8n1jAhLSgzrAzy1Hgr+hNrb57e+94F0=
 github.com/google/uuid v1.6.0/go.mod h1:TIyPZe4MgqvfeYDBFedMoGGpEw/LqOeaOT+nhxU+yHo=
 github.com/gorilla/websocket v1.5.4-0.20250319132907-e064f32e3674 h1:JeSE6pjso5THxAzdVpqr6/geYxZytqFMBCOtn/ujyeo=
 github.com/gorilla/websocket v1.5.4-0.20250319132907-e064f32e3674/go.mod h1:r4w70xmWCQKmi1ONH4KIaBptdivuRPyosB9RmPlGEwA=
+github.com/grandcat/zeroconf v1.0.0 h1:uHhahLBKqwWBV6WZUDAT71044vwOTL+McW0mBJvo6kE=
+github.com/grandcat/zeroconf v1.0.0/go.mod h1:lTKmG1zh86XyCoUeIHSA4FJMBwCJiQmGfcP2PdzytEs=
 github.com/json-iterator/go v1.1.12 h1:PV8peI4a0ysnczrg+LtxykD8LfKY9ML6u2jnxaEnrnM=
 github.com/json-iterator/go v1.1.12/go.mod h1:e30LSqwooZae/UwlEbR2852Gd8hjQvJoHmT4TnhNGBo=
 github.com/kisielk/errcheck v1.5.0/go.mod h1:pFxgyoBC7bSaBwPgfKdkLd5X25qrDl4LWUI2bnpBCr8=
@@ -98,6 +104,8 @@ github.com/mattn/go-localereader v0.0.1 h1:ygSAOl7ZXTx4RdPYinUpg6W99U8jWvWi9Ye2J
 github.com/mattn/go-localereader v0.0.1/go.mod h1:8fBrzywKY7BI3czFoHkuzRoWE9C+EiG4R1k4Cjx5p88=
 github.com/mattn/go-runewidth v0.0.19 h1:v++JhqYnZuu5jSKrk9RbgF5v4CGUjqRfBm05byFGLdw=
 github.com/mattn/go-runewidth v0.0.19/go.mod h1:XBkDxAl56ILZc9knddidhrOlY5R/pDhgLpndooCuJAs=
+github.com/miekg/dns v1.1.27 h1:aEH/kqUzUxGJ/UHcEKdJY+ugH6WEzsEBBSPa8zuy1aM=
+github.com/miekg/dns v1.1.27/go.mod h1:KNUDUusw/aVsxyTYZM1oqvCicbwhgbNgztCETuNZ7xM=
 github.com/moby/spdystream v0.5.0 h1:7r0J1Si3QO/kjRitvSLVVFUjxMEb/YLj6S9FF62JBCU=
 github.com/moby/spdystream v0.5.0/go.mod h1:xBAYlnt/ay+11ShkdFKNAG7LsyK/tmNBVvVOwrfMgdI=
 github.com/modern-go/concurrent v0.0.0-20180228061459-e0a39a4cb421/go.mod h1:6dJC0mAP4ikYIbvyc7fijjWJddQyLn8Ig3JB5CqoB9Q=
@@ -120,6 +128,7 @@ github.com/onsi/ginkgo/v2 v2.21.0 h1:7rg/4f3rB88pb5obDgNZrNHrQ4e6WpjonchcpuBRnZM
 github.com/onsi/ginkgo/v2 v2.21.0/go.mod h1:7Du3c42kxCUegi0IImZ1wUQzMBVecgIHjR1C+NkhLQo=
 github.com/onsi/gomega v1.35.1 h1:Cwbd75ZBPxFSuZ6T+rN/WCb/gOc6YgFBXLlZLhC7Ds4=
 github.com/onsi/gomega v1.35.1/go.mod h1:PvZbdDc8J6XJEpDK4HCuRBm8a6Fzp9/DmhC9C7yFlog=
+github.com/pkg/errors v0.9.1/go.mod h1:bwawxfHBFNV+L2hUp1rHADufV3IMtnDRdf1r5NINEl0=
 github.com/pmezard/go-difflib v1.0.0 h1:4DBwDE0NGyQoBHbLQYPwSUPoCMWR5BEzIk/f1lZbAQM=
 github.com/pmezard/go-difflib v1.0.0/go.mod h1:iKH77koFhYxTK1pcRnkKkqfTogsbg7gZNVY4sRDYZ/4=
 github.com/rivo/uniseg v0.4.7 h1:WUdvkW8uEhrYfLC4ZzdpI2ztxP1I582+49Oc5Mq64VQ=
@@ -147,12 +156,17 @@ go.yaml.in/yaml/v3 v3.0.4/go.mod h1:DhzuOOF2ATzADvBadXxruRBLzYTpT36CKvDb3+aBEFg=
 golang.org/x/crypto v0.0.0-20190308221718-c2843e01d9a2/go.mod h1:djNgcEr1/C05ACkg1iLfiJU5Ep61QUkGW8qpdssI0+w=
 golang.org/x/crypto v0.0.0-20191011191535-87dc89f01550/go.mod h1:yigFU9vqHzYiE8UmvKecakEJjdnWj3jj499lnFckfCI=
 golang.org/x/crypto v0.0.0-20200622213623-75b288015ac9/go.mod h1:LzIPMQfyMNhhGPhUkYOs5KpL4U8rLKemX1yGLhDgUto=
+golang.org/x/crypto v0.44.0 h1:A97SsFvM3AIwEEmTBiaxPPTYpDC47w720rdiiUvgoAU=
+golang.org/x/crypto v0.44.0/go.mod h1:013i+Nw79BMiQiMsOPcVCB5ZIJbYkerPrGnOa00tvmc=
 golang.org/x/exp v0.0.0-20231006140011-7918f672742d h1:jtJma62tbqLibJ5sFQz8bKtEM8rJBtfilJ2qTU199MI=
 golang.org/x/exp v0.0.0-20231006140011-7918f672742d/go.mod h1:ldy0pHrwJyGW56pPQzzkH36rKxoZW1tw7ZJpeKx+hdo=
+golang.org/x/mod v0.1.1-0.20191105210325-c90efee705ee/go.mod h1:QqPTAvyqsEbceGzBzNggFXnrqF1CaUcvgkdR5Ot7KZg=
 golang.org/x/mod v0.2.0/go.mod h1:s0Qsj1ACt9ePp/hMypM3fl4fZqREWJwdYDEqhRiZZUA=
 golang.org/x/mod v0.3.0/go.mod h1:s0Qsj1ACt9ePp/hMypM3fl4fZqREWJwdYDEqhRiZZUA=
 golang.org/x/net v0.0.0-20190404232315-eb5bcb51f2a3/go.mod h1:t9HGtf8HONx5eT2rtn7q6eTqICYqUVnKs3thJo3Qplg=
 golang.org/x/net v0.0.0-20190620200207-3b0461eec859/go.mod h1:z5CRVTTTmAJ677TzLLGU+0bjPO0LkuOLi4/5GtJWs/s=
+golang.org/x/net v0.0.0-20190923162816-aa69164e4478/go.mod h1:z5CRVTTTmAJ677TzLLGU+0bjPO0LkuOLi4/5GtJWs/s=
+golang.org/x/net v0.0.0-20200114155413-6afb5195e5aa/go.mod h1:z5CRVTTTmAJ677TzLLGU+0bjPO0LkuOLi4/5GtJWs/s=
 golang.org/x/net v0.0.0-20200226121028-0de0cce0169b/go.mod h1:z5CRVTTTmAJ677TzLLGU+0bjPO0LkuOLi4/5GtJWs/s=
 golang.org/x/net v0.0.0-20201021035429-f5854403a974/go.mod h1:sp8m0HH+o8qH0wwXwYZr8TS3Oi6o0r6Gce1SSxlDquU=
 golang.org/x/net v0.47.0 h1:Mx+4dIFzqraBXUugkia1OOvlD6LemFo1ALMHjrXDOhY=
@@ -164,6 +178,7 @@ golang.org/x/sync v0.0.0-20190911185100-cd5d95a43a6e/go.mod h1:RxMgew5VJxzue5/jJ
 golang.org/x/sync v0.0.0-20201020160332-67f06af15bc9/go.mod h1:RxMgew5VJxzue5/jJTE5uejpjVlOe/izrB70Jof72aM=
 golang.org/x/sys v0.0.0-20190215142949-d0b11bdaac8a/go.mod h1:STP8DvDyc/dI5b8T5hshtkjS+E42TnysNCUPdjciGhY=
 golang.org/x/sys v0.0.0-20190412213103-97732733099d/go.mod h1:h1NjWce9XRLGQEsW7wpKNCjG9DtNlClVuFLEZdDNbEs=
+golang.org/x/sys v0.0.0-20190924154521-2837fb4f24fe/go.mod h1:h1NjWce9XRLGQEsW7wpKNCjG9DtNlClVuFLEZdDNbEs=
 golang.org/x/sys v0.0.0-20200930185726-fdedc70b468f/go.mod h1:h1NjWce9XRLGQEsW7wpKNCjG9DtNlClVuFLEZdDNbEs=
 golang.org/x/sys v0.0.0-20210809222454-d867a43fc93e/go.mod h1:oPkhp1MJrh7nUepCBck5+mAzfO9JrbApNNgaTdGDITg=
 golang.org/x/sys v0.6.0/go.mod h1:oPkhp1MJrh7nUepCBck5+mAzfO9JrbApNNgaTdGDITg=
@@ -179,6 +194,7 @@ golang.org/x/time v0.14.0 h1:MRx4UaLrDotUKUdCIqzPC48t1Y9hANFKIRpNx+Te8PI=
 golang.org/x/time v0.14.0/go.mod h1:eL/Oa2bBBK0TkX57Fyni+NgnyQQN4LitPmob2Hjnqw4=
 golang.org/x/tools v0.0.0-20180917221912-90fa682c2a6e/go.mod h1:n7NCudcB/nEzxVGmLbDWY5pfWTLqBcC2KZ6jyYvM4mQ=
 golang.org/x/tools v0.0.0-20191119224855-298f0cb1881e/go.mod h1:b+2E5dAYhXwXZwtnZ6UAqBI28+e2cm9otk0dWdXHAEo=
+golang.org/x/tools v0.0.0-20191216052735-49a3e744a425/go.mod h1:TB2adYChydJhpapKDTa4BR/hXlZSLoq2Wpct/0txZ28=
 golang.org/x/tools v0.0.0-20200619180055-7c47624df98f/go.mod h1:EkVYQZoAsY45+roYkvgYkIh4xh/qjgUK9TdY2XT94GE=
 golang.org/x/tools v0.0.0-20210106214847-113979e3529a/go.mod h1:emZCQorbCU4vsT4fOWvOPXz4eW1wZW4PmDk9uLelYpA=
 golang.org/x/tools v0.38.0 h1:Hx2Xv8hISq8Lm16jvBZ2VQf+RLmbd7wVUsALibYI/IQ=

internal/config/config.go

@@ -1,19 +1,149 @@
 package config
 
 import (
+	"bytes"
 	"fmt"
 	"os"
+	"strings"
+	"time"
 
 	"gopkg.in/yaml.v3"
 )
 
 const (
-	maxConfigSize = 10 * 1024 * 1024 // 10MB
+	// maxConfigSize is the maximum allowed configuration file size (10MB)
+	maxConfigSize = 10 * 1024 * 1024
+
+	// Default health check settings
+	DefaultHealthCheckInterval = 3 * time.Second  // How often to check connection health
+	DefaultHealthCheckTimeout  = 2 * time.Second  // Timeout for health check probes
+	DefaultHealthCheckMethod   = "data-transfer"  // More reliable than tcp-dial
+	DefaultMaxConnectionAge    = 25 * time.Minute // Reconnect before k8s 30min timeout
+	DefaultMaxIdleTime         = 10 * time.Minute // Reconnect if no activity
+
+	// Default reliability settings
+	DefaultTCPKeepalive   = 30 * time.Second // OS-level TCP keepalive interval
+	DefaultDialTimeout    = 30 * time.Second // Connection establishment timeout
+	DefaultWatchdogPeriod = 30 * time.Second // Goroutine health check interval
 )
 
 // Config represents the root configuration structure from .kportal.yaml
 type Config struct {
-	Contexts []Context `yaml:"contexts"`
+	Contexts    []Context        `yaml:"contexts"`
+	HealthCheck *HealthCheckSpec `yaml:"healthCheck,omitempty"`
+	Reliability *ReliabilitySpec `yaml:"reliability,omitempty"`
+	MDNS        *MDNSSpec        `yaml:"mdns,omitempty"`
+}
+
+// MDNSSpec configures mDNS (multicast DNS) hostname publishing
+// When enabled, forwards with aliases can be accessed via <alias>.local hostnames
+type MDNSSpec struct {
+	Enabled bool `yaml:"enabled"` // Enable mDNS hostname publishing
+}
+
+// HealthCheckSpec configures health check behavior
+type HealthCheckSpec struct {
+	Interval         string `yaml:"interval,omitempty"`         // e.g., "3s", "5s"
+	Timeout          string `yaml:"timeout,omitempty"`          // e.g., "2s"
+	Method           string `yaml:"method,omitempty"`           // "tcp-dial" | "data-transfer"
+	MaxConnectionAge string `yaml:"maxConnectionAge,omitempty"` // e.g., "25m" - reconnect before k8s timeout
+	MaxIdleTime      string `yaml:"maxIdleTime,omitempty"`      // e.g., "10m" - reconnect if no activity
+}
+
+// ReliabilitySpec configures connection reliability features
+type ReliabilitySpec struct {
+	TCPKeepalive   string `yaml:"tcpKeepalive,omitempty"`   // e.g., "30s" - OS-level keepalive
+	DialTimeout    string `yaml:"dialTimeout,omitempty"`    // e.g., "30s" - connection dial timeout
+	RetryOnStale   bool   `yaml:"retryOnStale,omitempty"`   // Auto-reconnect on stale detection
+	WatchdogPeriod string `yaml:"watchdogPeriod,omitempty"` // e.g., "30s" - goroutine watchdog interval
+}
+
+// parseDurationOrDefault parses a duration string and returns the default if empty or invalid.
+func parseDurationOrDefault(value string, defaultDur time.Duration) time.Duration {
+	if value == "" {
+		return defaultDur
+	}
+	if d, err := time.ParseDuration(value); err == nil {
+		return d
+	}
+	return defaultDur
+}
+
+// GetHealthCheckIntervalOrDefault returns the health check interval or default value
+func (c *Config) GetHealthCheckIntervalOrDefault() time.Duration {
+	if c.HealthCheck == nil {
+		return DefaultHealthCheckInterval
+	}
+	return parseDurationOrDefault(c.HealthCheck.Interval, DefaultHealthCheckInterval)
+}
+
+// GetHealthCheckTimeoutOrDefault returns the health check timeout or default value
+func (c *Config) GetHealthCheckTimeoutOrDefault() time.Duration {
+	if c.HealthCheck == nil {
+		return DefaultHealthCheckTimeout
+	}
+	return parseDurationOrDefault(c.HealthCheck.Timeout, DefaultHealthCheckTimeout)
+}
+
+// GetHealthCheckMethod returns the health check method or default
+func (c *Config) GetHealthCheckMethod() string {
+	if c.HealthCheck != nil && c.HealthCheck.Method != "" {
+		return c.HealthCheck.Method
+	}
+	return DefaultHealthCheckMethod
+}
+
+// GetMaxConnectionAge returns the max connection age or default
+func (c *Config) GetMaxConnectionAge() time.Duration {
+	if c.HealthCheck == nil {
+		return DefaultMaxConnectionAge
+	}
+	return parseDurationOrDefault(c.HealthCheck.MaxConnectionAge, DefaultMaxConnectionAge)
+}
+
+// GetMaxIdleTime returns the max idle time or default
+func (c *Config) GetMaxIdleTime() time.Duration {
+	if c.HealthCheck == nil {
+		return DefaultMaxIdleTime
+	}
+	return parseDurationOrDefault(c.HealthCheck.MaxIdleTime, DefaultMaxIdleTime)
+}
+
+// GetTCPKeepalive returns the TCP keepalive duration or default
+func (c *Config) GetTCPKeepalive() time.Duration {
+	if c.Reliability == nil {
+		return DefaultTCPKeepalive
+	}
+	return parseDurationOrDefault(c.Reliability.TCPKeepalive, DefaultTCPKeepalive)
+}
+
+// GetRetryOnStale returns whether to retry on stale connections
+func (c *Config) GetRetryOnStale() bool {
+	if c.Reliability != nil {
+		return c.Reliability.RetryOnStale
+	}
+	return true // Default: enabled
+}
+
+// GetWatchdogPeriod returns the goroutine watchdog check period or default
+func (c *Config) GetWatchdogPeriod() time.Duration {
+	if c.Reliability == nil {
+		return DefaultWatchdogPeriod
+	}
+	return parseDurationOrDefault(c.Reliability.WatchdogPeriod, DefaultWatchdogPeriod)
+}
+
+// GetDialTimeout returns the connection dial timeout or default
+func (c *Config) GetDialTimeout() time.Duration {
+	if c.Reliability == nil {
+		return DefaultDialTimeout
+	}
+	return parseDurationOrDefault(c.Reliability.DialTimeout, DefaultDialTimeout)
+}
+
+// IsMDNSEnabled returns whether mDNS hostname publishing is enabled
+func (c *Config) IsMDNSEnabled() bool {
+	return c.MDNS != nil && c.MDNS.Enabled
 }
 
 // Context represents a Kubernetes context with its namespaces
@@ -82,6 +212,25 @@ func (f *Forward) GetNamespace() string {
 	return f.namespaceName
 }
 
+// GetMDNSAlias returns the alias to use for mDNS hostname registration.
+// If an explicit alias is set, it returns that.
+// Otherwise, it generates one from the resource name (e.g., "service/logto" -> "logto").
+func (f *Forward) GetMDNSAlias() string {
+	if f.Alias != "" {
+		return f.Alias
+	}
+
+	// Generate alias from resource name
+	// Format is "type/name" (e.g., "service/logto", "pod/my-app")
+	parts := strings.SplitN(f.Resource, "/", 2)
+	if len(parts) == 2 && parts[1] != "" {
+		return parts[1]
+	}
+
+	// Fallback: can't generate a valid alias (e.g., "pod" with selector)
+	return ""
+}
+
 // LoadConfig loads and parses the configuration file from the given path.
 func LoadConfig(path string) (*Config, error) {
 	// Validate file size before reading
@@ -103,9 +252,15 @@ func LoadConfig(path string) (*Config, error) {
 }
 
 // ParseConfig parses YAML configuration data into a Config struct.
+// It uses strict parsing that rejects unknown keys to catch typos.
 func ParseConfig(data []byte) (*Config, error) {
 	var cfg Config
-	if err := yaml.Unmarshal(data, &cfg); err != nil {
+
+	// Use decoder with KnownFields to reject unknown keys (catches typos)
+	decoder := yaml.NewDecoder(bytes.NewReader(data))
+	decoder.KnownFields(true)
+
+	if err := decoder.Decode(&cfg); err != nil {
 		return nil, fmt.Errorf("failed to parse YAML: %w", err)
 	}
 

internal/config/validator.go

@@ -6,10 +6,15 @@ import (
 )
 
 const (
-	minPort = 1
-	maxPort = 65535
+	MinPort = 1
+	MaxPort = 65535
 )
 
+// IsValidPort returns true if the port number is within the valid range (1-65535).
+func IsValidPort(port int) bool {
+	return port >= MinPort && port <= MaxPort
+}
+
 // ValidationError represents a configuration validation error with context.
 type ValidationError struct {
 	Field   string            // The field that failed validation
@@ -56,6 +61,11 @@ func (v *Validator) ValidateConfig(cfg *Config) []ValidationError {
 	// Check for duplicate local ports
 	errs = append(errs, v.validateDuplicatePorts(cfg)...)
 
+	// Validate mDNS configuration
+	if cfg.IsMDNSEnabled() {
+		errs = append(errs, v.validateMDNS(cfg)...)
+	}
+
 	return errs
 }
 
@@ -84,7 +94,7 @@ func (v *Validator) validateStructure(cfg *Config) []ValidationError {
 				Field:   fmt.Sprintf("contexts[%d].namespaces", i),
 				Message: fmt.Sprintf("Context '%s' must have at least one namespace", ctx.Name),
 			})
-			continue
+			// Don't continue - still validate other aspects of the context if any
 		}
 
 		for j, ns := range ctx.Namespaces {
@@ -130,17 +140,17 @@ func (v *Validator) validateForward(fwd *Forward) []ValidationError {
 	}
 
 	// Validate ports
-	if fwd.Port < minPort || fwd.Port > maxPort {
+	if fwd.Port < MinPort || fwd.Port > MaxPort {
 		errs = append(errs, ValidationError{
 			Field:   "port",
-			Message: fmt.Sprintf("Invalid port %d for forward %s (must be between %d and %d)", fwd.Port, fwd.ID(), minPort, maxPort),
+			Message: fmt.Sprintf("Invalid port %d for forward %s (must be between %d and %d)", fwd.Port, fwd.ID(), MinPort, MaxPort),
 		})
 	}
 
-	if fwd.LocalPort < minPort || fwd.LocalPort > maxPort {
+	if fwd.LocalPort < MinPort || fwd.LocalPort > MaxPort {
 		errs = append(errs, ValidationError{
 			Field:   "localPort",
-			Message: fmt.Sprintf("Invalid localPort %d for forward %s (must be between %d and %d)", fwd.LocalPort, fwd.ID(), minPort, maxPort),
+			Message: fmt.Sprintf("Invalid localPort %d for forward %s (must be between %d and %d)", fwd.LocalPort, fwd.ID(), MinPort, MaxPort),
 		})
 	}
 
@@ -265,3 +275,85 @@ func FormatValidationErrors(errs []ValidationError) string {
 
 	return sb.String()
 }
+
+// validateMDNS validates mDNS configuration when enabled.
+// It checks that aliases used for mDNS hostnames are valid and unique.
+// This includes both explicit aliases and auto-generated ones from resource names.
+func (v *Validator) validateMDNS(cfg *Config) []ValidationError {
+	var errs []ValidationError
+
+	aliasMap := make(map[string][]string) // alias -> list of forward IDs using it
+
+	for _, ctx := range cfg.Contexts {
+		for _, ns := range ctx.Namespaces {
+			for _, fwd := range ns.Forwards {
+				// Get the mDNS alias (explicit or generated from resource name)
+				mdnsAlias := fwd.GetMDNSAlias()
+				if mdnsAlias == "" {
+					// No alias available (e.g., "pod" with selector only)
+					continue
+				}
+
+				// Validate alias is a valid hostname (RFC 1123)
+				if !isValidHostname(mdnsAlias) {
+					errs = append(errs, ValidationError{
+						Field:   "alias",
+						Message: fmt.Sprintf("Forward %s has invalid mDNS hostname '%s' (must be a valid RFC 1123 hostname)", fwd.ID(), mdnsAlias),
+					})
+				}
+
+				aliasMap[mdnsAlias] = append(aliasMap[mdnsAlias], fwd.ID())
+			}
+		}
+	}
+
+	// Check for duplicate aliases (would cause mDNS conflicts)
+	for alias, forwards := range aliasMap {
+		if len(forwards) > 1 {
+			errs = append(errs, ValidationError{
+				Field:   "alias",
+				Message: fmt.Sprintf("Duplicate mDNS hostname '%s' used by multiple forwards (would cause conflict)", alias),
+				Context: map[string]string{
+					"alias":    alias,
+					"forwards": strings.Join(forwards, ", "),
+				},
+			})
+		}
+	}
+
+	return errs
+}
+
+// isValidHostname checks if a string is a valid RFC 1123 hostname.
+// Hostnames must start with alphanumeric, contain only alphanumeric and hyphens,
+// and be 1-63 characters long.
+func isValidHostname(name string) bool {
+	if len(name) == 0 || len(name) > 63 {
+		return false
+	}
+
+	// Must start with alphanumeric
+	if !isAlphanumeric(name[0]) {
+		return false
+	}
+
+	// Must end with alphanumeric
+	if !isAlphanumeric(name[len(name)-1]) {
+		return false
+	}
+
+	// Check all characters
+	for i := 0; i < len(name); i++ {
+		c := name[i]
+		if !isAlphanumeric(c) && c != '-' {
+			return false
+		}
+	}
+
+	return true
+}
+
+// isAlphanumeric returns true if the character is a letter or digit.
+func isAlphanumeric(c byte) bool {
+	return (c >= 'a' && c <= 'z') || (c >= 'A' && c <= 'Z') || (c >= '0' && c <= '9')
+}

internal/config/validator_test.go

@@ -701,3 +701,274 @@ func TestValidator_ValidateStructure(t *testing.T) {
 		})
 	}
 }
+
+func TestValidator_ValidateMDNS(t *testing.T) {
+	validator := NewValidator()
+
+	tests := []struct {
+		name          string
+		config        *Config
+		expectErrors  bool
+		errorContains []string
+	}{
+		{
+			name: "mDNS disabled - no validation",
+			config: &Config{
+				Contexts: []Context{
+					{
+						Name: "dev",
+						Namespaces: []Namespace{
+							{
+								Name: "default",
+								Forwards: []Forward{
+									{Resource: "pod/app", Port: 8080, LocalPort: 8080, Alias: "invalid_alias", contextName: "dev", namespaceName: "default"},
+								},
+							},
+						},
+					},
+				},
+			},
+			expectErrors: false,
+		},
+		{
+			name: "mDNS enabled - valid aliases",
+			config: &Config{
+				MDNS: &MDNSSpec{Enabled: true},
+				Contexts: []Context{
+					{
+						Name: "dev",
+						Namespaces: []Namespace{
+							{
+								Name: "default",
+								Forwards: []Forward{
+									{Resource: "pod/app1", Port: 8080, LocalPort: 8080, Alias: "my-app", contextName: "dev", namespaceName: "default"},
+									{Resource: "pod/app2", Port: 8081, LocalPort: 8081, Alias: "my-service", contextName: "dev", namespaceName: "default"},
+								},
+							},
+						},
+					},
+				},
+			},
+			expectErrors: false,
+		},
+		{
+			name: "mDNS enabled - no alias (allowed)",
+			config: &Config{
+				MDNS: &MDNSSpec{Enabled: true},
+				Contexts: []Context{
+					{
+						Name: "dev",
+						Namespaces: []Namespace{
+							{
+								Name: "default",
+								Forwards: []Forward{
+									{Resource: "pod/app", Port: 8080, LocalPort: 8080, contextName: "dev", namespaceName: "default"},
+								},
+							},
+						},
+					},
+				},
+			},
+			expectErrors: false,
+		},
+		{
+			name: "mDNS enabled - invalid alias with underscore",
+			config: &Config{
+				MDNS: &MDNSSpec{Enabled: true},
+				Contexts: []Context{
+					{
+						Name: "dev",
+						Namespaces: []Namespace{
+							{
+								Name: "default",
+								Forwards: []Forward{
+									{Resource: "pod/app", Port: 8080, LocalPort: 8080, Alias: "my_app", contextName: "dev", namespaceName: "default"},
+								},
+							},
+						},
+					},
+				},
+			},
+			expectErrors:  true,
+			errorContains: []string{"invalid mDNS hostname", "RFC 1123"},
+		},
+		{
+			name: "mDNS enabled - alias starts with hyphen",
+			config: &Config{
+				MDNS: &MDNSSpec{Enabled: true},
+				Contexts: []Context{
+					{
+						Name: "dev",
+						Namespaces: []Namespace{
+							{
+								Name: "default",
+								Forwards: []Forward{
+									{Resource: "pod/app", Port: 8080, LocalPort: 8080, Alias: "-myapp", contextName: "dev", namespaceName: "default"},
+								},
+							},
+						},
+					},
+				},
+			},
+			expectErrors:  true,
+			errorContains: []string{"invalid mDNS hostname"},
+		},
+		{
+			name: "mDNS enabled - alias ends with hyphen",
+			config: &Config{
+				MDNS: &MDNSSpec{Enabled: true},
+				Contexts: []Context{
+					{
+						Name: "dev",
+						Namespaces: []Namespace{
+							{
+								Name: "default",
+								Forwards: []Forward{
+									{Resource: "pod/app", Port: 8080, LocalPort: 8080, Alias: "myapp-", contextName: "dev", namespaceName: "default"},
+								},
+							},
+						},
+					},
+				},
+			},
+			expectErrors:  true,
+			errorContains: []string{"invalid mDNS hostname"},
+		},
+		{
+			name: "mDNS enabled - duplicate aliases",
+			config: &Config{
+				MDNS: &MDNSSpec{Enabled: true},
+				Contexts: []Context{
+					{
+						Name: "dev",
+						Namespaces: []Namespace{
+							{
+								Name: "default",
+								Forwards: []Forward{
+									{Resource: "pod/app1", Port: 8080, LocalPort: 8080, Alias: "myapp", contextName: "dev", namespaceName: "default"},
+									{Resource: "pod/app2", Port: 8081, LocalPort: 8081, Alias: "myapp", contextName: "dev", namespaceName: "default"},
+								},
+							},
+						},
+					},
+				},
+			},
+			expectErrors:  true,
+			errorContains: []string{"Duplicate mDNS hostname", "conflict"},
+		},
+		{
+			name: "mDNS enabled - duplicate aliases across contexts",
+			config: &Config{
+				MDNS: &MDNSSpec{Enabled: true},
+				Contexts: []Context{
+					{
+						Name: "cluster1",
+						Namespaces: []Namespace{
+							{
+								Name: "default",
+								Forwards: []Forward{
+									{Resource: "pod/app1", Port: 8080, LocalPort: 8080, Alias: "shared-name", contextName: "cluster1", namespaceName: "default"},
+								},
+							},
+						},
+					},
+					{
+						Name: "cluster2",
+						Namespaces: []Namespace{
+							{
+								Name: "default",
+								Forwards: []Forward{
+									{Resource: "pod/app2", Port: 8081, LocalPort: 8081, Alias: "shared-name", contextName: "cluster2", namespaceName: "default"},
+								},
+							},
+						},
+					},
+				},
+			},
+			expectErrors:  true,
+			errorContains: []string{"Duplicate mDNS hostname", "shared-name"},
+		},
+	}
+
+	for _, tt := range tests {
+		t.Run(tt.name, func(t *testing.T) {
+			errs := validator.ValidateConfig(tt.config)
+
+			if tt.expectErrors {
+				assert.NotEmpty(t, errs, "expected validation errors")
+
+				// Check that expected error messages are present
+				for _, expectedMsg := range tt.errorContains {
+					found := false
+					for _, err := range errs {
+						if strings.Contains(err.Message, expectedMsg) {
+							found = true
+							break
+						}
+					}
+					assert.True(t, found, "expected error message '%s' not found in errors: %v", expectedMsg, errs)
+				}
+			} else {
+				assert.Empty(t, errs, "expected no validation errors, got: %v", errs)
+			}
+		})
+	}
+}
+
+func TestIsValidHostname(t *testing.T) {
+	tests := []struct {
+		name     string
+		hostname string
+		valid    bool
+	}{
+		{"valid simple", "myservice", true},
+		{"valid with hyphen", "my-service", true},
+		{"valid with numbers", "service123", true},
+		{"valid mixed", "my-service-123", true},
+		{"valid uppercase", "MyService", true},
+		{"valid single char", "a", true},
+		{"valid single digit", "1", true},
+		{"valid max length (63)", "aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa", true},
+		{"invalid empty", "", false},
+		{"invalid too long (64)", "aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa", false},
+		{"invalid starts with hyphen", "-myservice", false},
+		{"invalid ends with hyphen", "myservice-", false},
+		{"invalid underscore", "my_service", false},
+		{"invalid dot", "my.service", false},
+		{"invalid space", "my service", false},
+		{"invalid special char", "my@service", false},
+	}
+
+	for _, tt := range tests {
+		t.Run(tt.name, func(t *testing.T) {
+			result := isValidHostname(tt.hostname)
+			assert.Equal(t, tt.valid, result, "isValidHostname(%q) = %v, want %v", tt.hostname, result, tt.valid)
+		})
+	}
+}
+
+func TestIsAlphanumeric(t *testing.T) {
+	tests := []struct {
+		char  byte
+		valid bool
+	}{
+		{'a', true},
+		{'z', true},
+		{'A', true},
+		{'Z', true},
+		{'0', true},
+		{'9', true},
+		{'-', false},
+		{'_', false},
+		{'.', false},
+		{' ', false},
+		{'@', false},
+	}
+
+	for _, tt := range tests {
+		t.Run(string(tt.char), func(t *testing.T) {
+			result := isAlphanumeric(tt.char)
+			assert.Equal(t, tt.valid, result, "isAlphanumeric(%q) = %v, want %v", tt.char, result, tt.valid)
+		})
+	}
+}

internal/forward/manager.go

@@ -10,11 +10,7 @@ import (
 	"github.com/nvm/kportal/internal/healthcheck"
 	"github.com/nvm/kportal/internal/k8s"
 	"github.com/nvm/kportal/internal/logger"
-)
-
-const (
-	healthCheckInterval = 5 * time.Second
-	healthCheckTimeout  = 2 * time.Second
+	"github.com/nvm/kportal/internal/mdns"
 )
 
 // StatusUpdater is an interface for updating forward status
@@ -34,12 +30,16 @@ type Manager struct {
 	portForwarder *k8s.PortForwarder
 	portChecker   *PortChecker
 	healthChecker *healthcheck.Checker
+	watchdog      *Watchdog
+	mdnsPublisher *mdns.Publisher
 	verbose       bool
 	currentConfig *config.Config
 	statusUI      StatusUpdater
 }
 
 // NewManager creates a new forward Manager.
+// The health checker will be created with default settings and can be
+// reconfigured via SetConfig().
 func NewManager(verbose bool) (*Manager, error) {
 	clientPool, err := k8s.NewClientPool()
 	if err != nil {
@@ -49,8 +49,13 @@ func NewManager(verbose bool) (*Manager, error) {
 	resolver := k8s.NewResourceResolver(clientPool)
 	portForwarder := k8s.NewPortForwarder(clientPool, resolver)
 
-	// Create health checker: check every 5 seconds with 2 second timeout
-	healthChecker := healthcheck.NewChecker(healthCheckInterval, healthCheckTimeout)
+	// Create health checker with defaults: check every 3 seconds with 2 second timeout
+	// Will be reconfigured when config is loaded
+	healthChecker := healthcheck.NewChecker(3*time.Second, 2*time.Second)
+
+	// Create watchdog with default settings: check every 30 seconds, 60 second hang threshold
+	// Will be reconfigured when config is loaded
+	watchdog := NewWatchdog(30*time.Second, 60*time.Second)
 
 	return &Manager{
 		workers:       make(map[string]*ForwardWorker),
@@ -59,15 +64,66 @@ func NewManager(verbose bool) (*Manager, error) {
 		portForwarder: portForwarder,
 		portChecker:   NewPortChecker(),
 		healthChecker: healthChecker,
+		watchdog:      watchdog,
 		verbose:       verbose,
 	}, nil
 }
 
+// configureHealthChecker creates a new health checker with settings from config
+func (m *Manager) configureHealthChecker(cfg *config.Config) {
+	// Stop existing health checker
+	if m.healthChecker != nil {
+		m.healthChecker.Stop()
+	}
+
+	// Parse check method
+	methodStr := cfg.GetHealthCheckMethod()
+	var method healthcheck.CheckMethod
+	switch methodStr {
+	case "tcp-dial":
+		method = healthcheck.CheckMethodTCPDial
+	case "data-transfer":
+		method = healthcheck.CheckMethodDataTransfer
+	default:
+		method = healthcheck.CheckMethodDataTransfer
+	}
+
+	// Create new health checker with config settings
+	m.healthChecker = healthcheck.NewCheckerWithOptions(healthcheck.CheckerOptions{
+		Interval:         cfg.GetHealthCheckIntervalOrDefault(),
+		Timeout:          cfg.GetHealthCheckTimeoutOrDefault(),
+		Method:           method,
+		MaxConnectionAge: cfg.GetMaxConnectionAge(),
+		MaxIdleTime:      cfg.GetMaxIdleTime(),
+	})
+
+	// Configure TCP settings on port forwarder
+	tcpKeepalive := cfg.GetTCPKeepalive()
+	dialTimeout := cfg.GetDialTimeout()
+	m.portForwarder.SetTCPKeepalive(tcpKeepalive)
+	m.portForwarder.SetDialTimeout(dialTimeout)
+
+	logger.Info("Health checker and reliability configured", map[string]interface{}{
+		"interval":           cfg.GetHealthCheckIntervalOrDefault().String(),
+		"timeout":            cfg.GetHealthCheckTimeoutOrDefault().String(),
+		"method":             methodStr,
+		"max_connection_age": cfg.GetMaxConnectionAge().String(),
+		"max_idle_time":      cfg.GetMaxIdleTime().String(),
+		"tcp_keepalive":      tcpKeepalive.String(),
+		"dial_timeout":       dialTimeout.String(),
+	})
+}
+
 // SetStatusUI sets the status updater for the manager
 func (m *Manager) SetStatusUI(ui StatusUpdater) {
 	m.statusUI = ui
 }
 
+// SetMDNSPublisher sets the mDNS publisher for the manager
+func (m *Manager) SetMDNSPublisher(publisher *mdns.Publisher) {
+	m.mdnsPublisher = publisher
+}
+
 // Start initializes and starts all port-forwards from the configuration.
 func (m *Manager) Start(cfg *config.Config) error {
 	if cfg == nil {
@@ -76,6 +132,20 @@ func (m *Manager) Start(cfg *config.Config) error {
 
 	m.currentConfig = cfg
 
+	// Configure health checker with settings from config
+	m.configureHealthChecker(cfg)
+
+	// Start watchdog
+	watchdogPeriod := cfg.GetWatchdogPeriod()
+	m.watchdog.checkInterval = watchdogPeriod
+	m.watchdog.hangThreshold = watchdogPeriod * 2 // Hang threshold is 2x check interval
+	m.watchdog.Start()
+
+	logger.Info("Watchdog started", map[string]interface{}{
+		"check_interval": watchdogPeriod.String(),
+		"hang_threshold": (watchdogPeriod * 2).String(),
+	})
+
 	// Get all forwards from config
 	forwards := cfg.GetAllForwards()
 
@@ -119,8 +189,14 @@ func (m *Manager) Start(cfg *config.Config) error {
 func (m *Manager) Stop() {
 	log.Printf("Stopping all port-forwards...")
 
-	// Stop health checker first
+	// Stop health checker and watchdog first
 	m.healthChecker.Stop()
+	m.watchdog.Stop()
+
+	// Stop mDNS publisher
+	if m.mdnsPublisher != nil {
+		m.mdnsPublisher.Stop()
+	}
 
 	m.workersMu.Lock()
 	workers := make([]*ForwardWorker, 0, len(m.workers))
@@ -273,26 +349,76 @@ func (m *Manager) startWorker(fwd config.Forward) error {
 		m.statusUI.AddForward(fwd.ID(), &fwd)
 	}
 
+	// Register with watchdog
+	m.watchdog.RegisterWorker(fwd.ID(), func(forwardID string) {
+		logger.Warn("Watchdog triggered reconnection for hung worker", map[string]interface{}{
+			"forward_id": forwardID,
+		})
+
+		// Find and trigger reconnect on hung worker
+		m.workersMu.RLock()
+		worker, exists := m.workers[forwardID]
+		m.workersMu.RUnlock()
+
+		if exists {
+			worker.TriggerReconnect("watchdog detected hung worker")
+		}
+	})
+
 	// Register with health checker
 	m.healthChecker.Register(fwd.ID(), fwd.LocalPort, func(forwardID string, status healthcheck.Status, errorMsg string) {
 		if m.statusUI != nil {
 			m.statusUI.UpdateStatus(forwardID, string(status))
+
 			// Send error separately if there is one
-			if status == healthcheck.StatusUnhealthy && errorMsg != "" {
+			if (status == healthcheck.StatusUnhealthy || status == healthcheck.StatusStale) && errorMsg != "" {
 				if ui, ok := m.statusUI.(interface{ SetError(id, msg string) }); ok {
 					ui.SetError(forwardID, errorMsg)
 				}
 			}
 		}
+
+		// Handle stale connections: trigger reconnection if retryOnStale is enabled
+		if status == healthcheck.StatusStale && m.currentConfig.GetRetryOnStale() {
+			logger.Info("Stale connection detected, triggering reconnection", map[string]interface{}{
+				"forward_id": forwardID,
+				"reason":     errorMsg,
+			})
+
+			// Find and notify the worker to reconnect
+			m.workersMu.RLock()
+			worker, exists := m.workers[forwardID]
+			m.workersMu.RUnlock()
+
+			if exists {
+				worker.TriggerReconnect("stale connection")
+			}
+		}
 	})
 
 	// Create and start worker
-	worker := NewForwardWorker(fwd, m.portForwarder, m.verbose, m.statusUI, m.healthChecker)
+	worker := NewForwardWorker(fwd, m.portForwarder, m.verbose, m.statusUI, m.healthChecker, m.watchdog)
 	worker.Start()
 
 	// Store worker
 	m.workers[fwd.ID()] = worker
 
+	// Register mDNS hostname if enabled
+	// Uses explicit alias if set, otherwise generates from resource name
+	if m.mdnsPublisher != nil {
+		mdnsAlias := fwd.GetMDNSAlias()
+		if mdnsAlias != "" {
+			if err := m.mdnsPublisher.Register(fwd.ID(), mdnsAlias, fwd.LocalPort); err != nil {
+				logger.Warn("Failed to register mDNS hostname", map[string]interface{}{
+					"forward_id": fwd.ID(),
+					"alias":      mdnsAlias,
+					"error":      err.Error(),
+				})
+				// Don't fail the forward start - mDNS is optional
+			}
+		}
+	}
+
 	return nil
 }
 
@@ -312,8 +438,14 @@ func (m *Manager) stopWorkerInternal(id string, removeFromUI bool) error {
 	delete(m.workers, id)
 	m.workersMu.Unlock()
 
-	// Unregister from health checker
+	// Unregister from health checker and watchdog
 	m.healthChecker.Unregister(id)
+	m.watchdog.UnregisterWorker(id)
+
+	// Unregister mDNS hostname
+	if m.mdnsPublisher != nil {
+		m.mdnsPublisher.Unregister(id)
+	}
 
 	// Notify UI - either remove or update to disabled status
 	if m.statusUI != nil {

internal/forward/portcheck.go

@@ -6,11 +6,20 @@ import (
 	"os/exec"
 	"runtime"
 	"strings"
+
+	"github.com/nvm/kportal/internal/logger"
+)
+
+const (
+	// maxPIDLength is the maximum length of a valid PID string (9 digits covers PIDs up to 999,999,999)
+	maxPIDLength = 9
+	// minNetstatFields is the minimum number of fields expected in netstat output
+	minNetstatFields = 5
 )
 
 // isValidPID validates that a PID string contains only digits
 func isValidPID(pid string) bool {
-	if len(pid) == 0 || len(pid) > 9 {
+	if len(pid) == 0 || len(pid) > maxPIDLength {
 		return false
 	}
 	for _, c := range pid {
@@ -21,6 +30,72 @@ func isValidPID(pid string) bool {
 	return true
 }
 
+// processInfo holds information about a process using a port
+type processInfo struct {
+	pid     string
+	name    string
+	isValid bool
+}
+
+// formatProcessInfo formats process information for display
+func formatProcessInfo(info processInfo) string {
+	if !info.isValid {
+		return "unknown"
+	}
+	if info.name != "" {
+		return fmt.Sprintf("%s (PID %s)", info.name, info.pid)
+	}
+	return fmt.Sprintf("PID %s", info.pid)
+}
+
+// formatProcessList formats a list of processes into a human-readable string.
+// Returns "unknown" if the list is empty.
+func formatProcessList(processes []processInfo) string {
+	if len(processes) == 0 {
+		return "unknown"
+	}
+	if len(processes) == 1 {
+		return formatProcessInfo(processes[0])
+	}
+	// Multiple processes - format as comma-separated list
+	parts := make([]string, len(processes))
+	for i, p := range processes {
+		parts[i] = formatProcessInfo(p)
+	}
+	return strings.Join(parts, ", ")
+}
+
+// getProcessNameByPID retrieves the process name for a given PID on Unix systems
+func getProcessNameByPID(pid string) string {
+	cmd := exec.Command("ps", "-p", pid, "-o", "comm=")
+	output, err := cmd.Output()
+	if err != nil {
+		return ""
+	}
+	return strings.TrimSpace(string(output))
+}
+
+// getProcessNameByPIDWindows retrieves the process name for a given PID on Windows
+func getProcessNameByPIDWindows(pid string) string {
+	cmd := exec.Command("tasklist", "/FI", fmt.Sprintf("PID eq %s", pid), "/FO", "CSV", "/NH")
+	output, err := cmd.Output()
+	if err != nil {
+		return ""
+	}
+
+	// Parse CSV output: "process.exe","1234","Console","1","12,345 K"
+	csvLine := strings.TrimSpace(string(output))
+	if csvLine == "" {
+		return ""
+	}
+
+	parts := strings.Split(csvLine, ",")
+	if len(parts) > 0 {
+		return strings.Trim(parts[0], "\"")
+	}
+	return ""
+}
+
 // PortConflict represents a local port that is already in use.
 type PortConflict struct {
 	Port     int    // The conflicting port number
@@ -102,27 +177,55 @@ func (pc *PortChecker) getProcessUsingPortUnix(port int) string {
 		return "unknown"
 	}
 
-	// Get the first PID if multiple are returned
+	// Handle multiple PIDs (multiple processes on same port)
 	pids := strings.Split(pidStr, "\n")
-	pid := pids[0]
+	var validProcesses []processInfo
 
-	if !isValidPID(pid) {
-		return "unknown"
+	for _, pid := range pids {
+		pid = strings.TrimSpace(pid)
+		if pid == "" {
+			continue
+		}
+
+		if !isValidPID(pid) {
+			logger.Debug("Invalid PID format from lsof output", map[string]interface{}{
+				"port":    port,
+				"raw_pid": pid,
+			})
+			continue
+		}
+
+		procName := getProcessNameByPID(pid)
+		validProcesses = append(validProcesses, processInfo{
+			pid:     pid,
+			name:    procName,
+			isValid: true,
+		})
 	}
 
-	// Get process name using ps
-	cmd = exec.Command("ps", "-p", pid, "-o", "comm=")
-	output, err = cmd.Output()
-	if err != nil {
-		return fmt.Sprintf("PID %s", pid)
+	return formatProcessList(validProcesses)
+}
+
+// isListeningState checks if a netstat line indicates a listening state.
+// This handles both English and potentially other locales by checking for common patterns.
+func isListeningState(line string, fields []string) bool {
+	upperLine := strings.ToUpper(line)
+
+	// Check for common listening state indicators across locales
+	// English: LISTENING, German: ABHÖREN, French: ÉCOUTE, etc.
+	// The most reliable check is the state field position (4th field, 0-indexed = 3)
+	// and that it's a TCP connection with 0.0.0.0:0 or *:* as foreign address
+	if len(fields) >= minNetstatFields {
+		state := strings.ToUpper(fields[3])
+		// Common listening state values across Windows locales
+		if state == "LISTENING" || state == "ABHÖREN" || state == "ÉCOUTE" ||
+			state == "ESCUCHANDO" || state == "ASCOLTO" || state == "NASŁUCHIWANIE" {
+			return true
+		}
 	}
 
-	procName := strings.TrimSpace(string(output))
-	if procName == "" {
-		return fmt.Sprintf("PID %s", pid)
-	}
-
-	return fmt.Sprintf("%s (PID %s)", procName, pid)
+	// Fallback: check if line contains LISTENING (most common case)
+	return strings.Contains(upperLine, "LISTENING")
 }
 
 // getProcessUsingPortWindows uses netstat to find the process using a port on Windows.
@@ -138,6 +241,8 @@ func (pc *PortChecker) getProcessUsingPortWindows(port int) string {
 	lines := strings.Split(string(output), "\n")
 	portStr := fmt.Sprintf(":%d", port)
 
+	var validProcesses []processInfo
+
 	for _, line := range lines {
 		if !strings.Contains(line, portStr) {
 			continue
@@ -146,44 +251,42 @@ func (pc *PortChecker) getProcessUsingPortWindows(port int) string {
 		// Parse the line to extract PID
 		// Format: TCP    0.0.0.0:8080    0.0.0.0:0    LISTENING    1234
 		fields := strings.Fields(line)
-		if len(fields) < 5 {
+		if len(fields) < minNetstatFields {
 			continue
 		}
 
-		// Check if this is a LISTENING state
-		if !strings.Contains(strings.ToUpper(line), "LISTENING") {
+		// Check if this is a LISTENING state (locale-aware)
+		if !isListeningState(line, fields) {
+			continue
+		}
+
+		// Verify the local address field actually contains our port
+		// (avoid matching port in foreign address)
+		localAddr := fields[1]
+		if !strings.HasSuffix(localAddr, portStr) {
 			continue
 		}
 
 		pid := fields[len(fields)-1]
 
 		if !isValidPID(pid) {
-			return "unknown"
+			logger.Debug("Invalid PID format from netstat output", map[string]interface{}{
+				"port":    port,
+				"raw_pid": pid,
+				"line":    line,
+			})
+			continue
 		}
 
-		// Get process name using tasklist
-		cmd = exec.Command("tasklist", "/FI", fmt.Sprintf("PID eq %s", pid), "/FO", "CSV", "/NH")
-		output, err = cmd.Output()
-		if err != nil {
-			return fmt.Sprintf("PID %s", pid)
-		}
-
-		// Parse CSV output: "process.exe","1234","Console","1","12,345 K"
-		csvLine := strings.TrimSpace(string(output))
-		if csvLine == "" {
-			return fmt.Sprintf("PID %s", pid)
-		}
-
-		parts := strings.Split(csvLine, ",")
-		if len(parts) > 0 {
-			procName := strings.Trim(parts[0], "\"")
-			return fmt.Sprintf("%s (PID %s)", procName, pid)
-		}
-
-		return fmt.Sprintf("PID %s", pid)
+		procName := getProcessNameByPIDWindows(pid)
+		validProcesses = append(validProcesses, processInfo{
+			pid:     pid,
+			name:    procName,
+			isValid: true,
+		})
 	}
 
-	return "unknown"
+	return formatProcessList(validProcesses)
 }
 
 // FormatConflicts formats port conflicts into a human-readable error message.

internal/forward/watchdog.go

@@ -0,0 +1,174 @@
+package forward
+
+import (
+	"context"
+	"sync"
+	"time"
+
+	"github.com/nvm/kportal/internal/logger"
+)
+
+// Watchdog monitors worker goroutines to detect hung workers
+type Watchdog struct {
+	mu            sync.RWMutex
+	workers       map[string]*workerState // key: forward ID
+	checkInterval time.Duration
+	hangThreshold time.Duration // How long without heartbeat before considered hung
+	ctx           context.Context
+	cancel        context.CancelFunc
+	wg            sync.WaitGroup
+}
+
+// workerState tracks the health of a single worker
+type workerState struct {
+	forwardID      string
+	lastHeartbeat  time.Time
+	heartbeatCount uint64
+	isHung         bool
+	onHungCallback func(forwardID string)
+}
+
+// NewWatchdog creates a new goroutine watchdog
+func NewWatchdog(checkInterval, hangThreshold time.Duration) *Watchdog {
+	ctx, cancel := context.WithCancel(context.Background())
+	return &Watchdog{
+		workers:       make(map[string]*workerState),
+		checkInterval: checkInterval,
+		hangThreshold: hangThreshold,
+		ctx:           ctx,
+		cancel:        cancel,
+	}
+}
+
+// Start begins the watchdog monitoring loop
+func (w *Watchdog) Start() {
+	w.wg.Add(1)
+	go w.monitorLoop()
+}
+
+// Stop stops the watchdog
+func (w *Watchdog) Stop() {
+	w.cancel()
+	w.wg.Wait()
+}
+
+// RegisterWorker adds a worker to monitor
+func (w *Watchdog) RegisterWorker(forwardID string, onHungCallback func(string)) {
+	w.mu.Lock()
+	defer w.mu.Unlock()
+
+	w.workers[forwardID] = &workerState{
+		forwardID:      forwardID,
+		lastHeartbeat:  time.Now(),
+		heartbeatCount: 0,
+		isHung:         false,
+		onHungCallback: onHungCallback,
+	}
+
+	logger.Debug("Watchdog registered worker", map[string]interface{}{
+		"forward_id": forwardID,
+	})
+}
+
+// UnregisterWorker removes a worker from monitoring
+func (w *Watchdog) UnregisterWorker(forwardID string) {
+	w.mu.Lock()
+	defer w.mu.Unlock()
+
+	delete(w.workers, forwardID)
+
+	logger.Debug("Watchdog unregistered worker", map[string]interface{}{
+		"forward_id": forwardID,
+	})
+}
+
+// Heartbeat records that a worker is alive and processing
+// Workers should call this periodically (e.g., in their main loop)
+func (w *Watchdog) Heartbeat(forwardID string) {
+	w.mu.Lock()
+	defer w.mu.Unlock()
+
+	if state, exists := w.workers[forwardID]; exists {
+		state.lastHeartbeat = time.Now()
+		state.heartbeatCount++
+		state.isHung = false
+	}
+}
+
+// GetWorkerState returns the current state of a worker (for testing)
+func (w *Watchdog) GetWorkerState(forwardID string) (lastHeartbeat time.Time, count uint64, exists bool) {
+	w.mu.RLock()
+	defer w.mu.RUnlock()
+
+	if state, ok := w.workers[forwardID]; ok {
+		return state.lastHeartbeat, state.heartbeatCount, true
+	}
+	return time.Time{}, 0, false
+}
+
+// monitorLoop periodically checks all workers
+func (w *Watchdog) monitorLoop() {
+	defer w.wg.Done()
+
+	ticker := time.NewTicker(w.checkInterval)
+	defer ticker.Stop()
+
+	for {
+		select {
+		case <-w.ctx.Done():
+			return
+		case <-ticker.C:
+			w.checkWorkers()
+		}
+	}
+}
+
+// hungWorkerInfo stores information about a hung worker for deferred callback execution
+type hungWorkerInfo struct {
+	forwardID string
+	callback  func(string)
+}
+
+// checkWorkers checks all registered workers for hung state
+func (w *Watchdog) checkWorkers() {
+	// Collect hung workers while holding the lock
+	var hungWorkers []hungWorkerInfo
+
+	w.mu.Lock()
+	now := time.Now()
+	for forwardID, state := range w.workers {
+		timeSinceHeartbeat := now.Sub(state.lastHeartbeat)
+
+		// Check if worker is hung
+		if timeSinceHeartbeat > w.hangThreshold {
+			if !state.isHung {
+				// First time detecting hung state
+				state.isHung = true
+
+				logger.Warn("Watchdog detected hung worker", map[string]interface{}{
+					"forward_id":           forwardID,
+					"time_since_heartbeat": timeSinceHeartbeat.String(),
+					"hang_threshold":       w.hangThreshold.String(),
+					"heartbeat_count":      state.heartbeatCount,
+				})
+
+				// Collect callback for deferred execution outside the lock
+				if state.onHungCallback != nil {
+					hungWorkers = append(hungWorkers, hungWorkerInfo{
+						forwardID: forwardID,
+						callback:  state.onHungCallback,
+					})
+				}
+			}
+		}
+	}
+	w.mu.Unlock()
+
+	// Execute callbacks outside the lock to prevent deadlocks and ensure
+	// consistent state during callback execution. Callbacks are idempotent
+	// (they trigger reconnection via channels), so concurrent state changes
+	// between detection and callback execution are safe.
+	for _, hw := range hungWorkers {
+		hw.callback(hw.forwardID)
+	}
+}

internal/forward/watchdog_test.go

@@ -0,0 +1,310 @@
+package forward
+
+import (
+	"sync"
+	"testing"
+	"time"
+
+	"github.com/stretchr/testify/assert"
+	"github.com/stretchr/testify/require"
+	"github.com/stretchr/testify/suite"
+)
+
+// WatchdogTestSuite contains tests for the watchdog
+type WatchdogTestSuite struct {
+	suite.Suite
+	watchdog *Watchdog
+}
+
+func TestWatchdogSuite(t *testing.T) {
+	suite.Run(t, new(WatchdogTestSuite))
+}
+
+func (s *WatchdogTestSuite) SetupTest() {
+	// Create watchdog with fast intervals for testing
+	s.watchdog = NewWatchdog(100*time.Millisecond, 300*time.Millisecond)
+	s.watchdog.Start()
+}
+
+func (s *WatchdogTestSuite) TearDownTest() {
+	if s.watchdog != nil {
+		s.watchdog.Stop()
+	}
+}
+
+// TestRegisterUnregister tests basic registration and unregistration
+func (s *WatchdogTestSuite) TestRegisterUnregister() {
+	callbackCalled := false
+	callback := func(forwardID string) {
+		callbackCalled = true
+	}
+
+	// Register worker
+	s.watchdog.RegisterWorker("test-forward", callback)
+
+	// Verify worker is registered
+	_, _, exists := s.watchdog.GetWorkerState("test-forward")
+	assert.True(s.T(), exists, "Worker should be registered")
+
+	// Unregister worker
+	s.watchdog.UnregisterWorker("test-forward")
+
+	// Verify worker is unregistered
+	_, _, exists = s.watchdog.GetWorkerState("test-forward")
+	assert.False(s.T(), exists, "Worker should be unregistered")
+	assert.False(s.T(), callbackCalled, "Callback should not have been called")
+}
+
+// TestHeartbeat tests that heartbeats update worker state
+func (s *WatchdogTestSuite) TestHeartbeat() {
+	s.watchdog.RegisterWorker("test-forward", nil)
+
+	// Send initial heartbeat
+	s.watchdog.Heartbeat("test-forward")
+
+	lastHeartbeat1, count1, exists := s.watchdog.GetWorkerState("test-forward")
+	require.True(s.T(), exists)
+	assert.Equal(s.T(), uint64(1), count1)
+
+	// Wait a bit
+	time.Sleep(50 * time.Millisecond)
+
+	// Send another heartbeat
+	s.watchdog.Heartbeat("test-forward")
+
+	lastHeartbeat2, count2, exists := s.watchdog.GetWorkerState("test-forward")
+	require.True(s.T(), exists)
+	assert.Equal(s.T(), uint64(2), count2)
+	assert.True(s.T(), lastHeartbeat2.After(lastHeartbeat1), "Second heartbeat should be after first")
+}
+
+// TestHungWorkerDetection tests that hung workers are detected
+func (s *WatchdogTestSuite) TestHungWorkerDetection() {
+	callbackCalled := make(chan string, 1)
+	callback := func(forwardID string) {
+		callbackCalled <- forwardID
+	}
+
+	s.watchdog.RegisterWorker("test-forward", callback)
+
+	// Send initial heartbeat
+	s.watchdog.Heartbeat("test-forward")
+
+	// Wait for worker to be considered hung (300ms threshold + 100ms check interval)
+	timeout := time.After(1 * time.Second)
+
+	select {
+	case forwardID := <-callbackCalled:
+		assert.Equal(s.T(), "test-forward", forwardID)
+	case <-timeout:
+		s.T().Fatal("Timeout waiting for hung worker callback")
+	}
+}
+
+// TestHealthyWorkerNotDetectedAsHung tests that workers sending heartbeats are not considered hung
+func (s *WatchdogTestSuite) TestHealthyWorkerNotDetectedAsHung() {
+	callbackCalled := false
+	var mu sync.Mutex
+	callback := func(forwardID string) {
+		mu.Lock()
+		defer mu.Unlock()
+		callbackCalled = true
+	}
+
+	s.watchdog.RegisterWorker("test-forward", callback)
+
+	// Send periodic heartbeats (faster than hang threshold)
+	ticker := time.NewTicker(50 * time.Millisecond)
+	defer ticker.Stop()
+
+	done := make(chan bool)
+	go func() {
+		for i := 0; i < 10; i++ {
+			<-ticker.C
+			s.watchdog.Heartbeat("test-forward")
+		}
+		done <- true
+	}()
+
+	// Wait for all heartbeats to complete
+	<-done
+
+	// Check that callback was not called
+	mu.Lock()
+	assert.False(s.T(), callbackCalled, "Callback should not be called for healthy worker")
+	mu.Unlock()
+}
+
+// TestMultipleWorkers tests monitoring multiple workers simultaneously
+func (s *WatchdogTestSuite) TestMultipleWorkers() {
+	callbacks := make(map[string]int)
+	var mu sync.Mutex
+
+	makeCallback := func(id string) func(string) {
+		return func(forwardID string) {
+			mu.Lock()
+			defer mu.Unlock()
+			callbacks[id]++
+		}
+	}
+
+	// Register multiple workers
+	s.watchdog.RegisterWorker("worker-1", makeCallback("worker-1"))
+	s.watchdog.RegisterWorker("worker-2", makeCallback("worker-2"))
+	s.watchdog.RegisterWorker("worker-3", makeCallback("worker-3"))
+
+	// worker-1: Keep sending heartbeats (healthy)
+	ticker1 := time.NewTicker(50 * time.Millisecond)
+	defer ticker1.Stop()
+	go func() {
+		for i := 0; i < 10; i++ {
+			<-ticker1.C
+			s.watchdog.Heartbeat("worker-1")
+		}
+	}()
+
+	// worker-2: Send initial heartbeat then stop (will become hung)
+	s.watchdog.Heartbeat("worker-2")
+
+	// worker-3: Send initial heartbeat then stop (will become hung)
+	s.watchdog.Heartbeat("worker-3")
+
+	// Wait for hung workers to be detected
+	time.Sleep(600 * time.Millisecond)
+
+	// Check results
+	mu.Lock()
+	defer mu.Unlock()
+
+	assert.Equal(s.T(), 0, callbacks["worker-1"], "worker-1 should not trigger callback (healthy)")
+	assert.Greater(s.T(), callbacks["worker-2"], 0, "worker-2 should trigger callback (hung)")
+	assert.Greater(s.T(), callbacks["worker-3"], 0, "worker-3 should trigger callback (hung)")
+}
+
+// TestCallbackOnlyOnFirstDetection tests that callback is only called once when hung is first detected
+func (s *WatchdogTestSuite) TestCallbackOnlyOnFirstDetection() {
+	callbackCount := 0
+	var mu sync.Mutex
+	callback := func(forwardID string) {
+		mu.Lock()
+		defer mu.Unlock()
+		callbackCount++
+	}
+
+	s.watchdog.RegisterWorker("test-forward", callback)
+
+	// Send initial heartbeat
+	s.watchdog.Heartbeat("test-forward")
+
+	// Wait for multiple check cycles
+	time.Sleep(1 * time.Second)
+
+	// Check that callback was only called once
+	mu.Lock()
+	assert.Equal(s.T(), 1, callbackCount, "Callback should only be called once")
+	mu.Unlock()
+}
+
+// TestHeartbeatResetsHungState tests that sending heartbeat after hung detection resets state
+func (s *WatchdogTestSuite) TestHeartbeatResetsHungState() {
+	callbackCount := 0
+	var mu sync.Mutex
+	callback := func(forwardID string) {
+		mu.Lock()
+		defer mu.Unlock()
+		callbackCount++
+	}
+
+	s.watchdog.RegisterWorker("test-forward", callback)
+
+	// Send initial heartbeat
+	s.watchdog.Heartbeat("test-forward")
+
+	// Wait for hung detection
+	time.Sleep(500 * time.Millisecond)
+
+	mu.Lock()
+	firstCount := callbackCount
+	mu.Unlock()
+
+	assert.Equal(s.T(), 1, firstCount, "First hung detection should trigger callback")
+
+	// Send heartbeat to reset hung state
+	s.watchdog.Heartbeat("test-forward")
+
+	// Wait for worker to become hung again
+	time.Sleep(500 * time.Millisecond)
+
+	mu.Lock()
+	secondCount := callbackCount
+	mu.Unlock()
+
+	assert.Equal(s.T(), 2, secondCount, "Second hung detection should trigger callback again")
+}
+
+// TestConcurrentOperations tests thread safety
+func (s *WatchdogTestSuite) TestConcurrentOperations() {
+	var wg sync.WaitGroup
+	numWorkers := 10
+
+	for i := 0; i < numWorkers; i++ {
+		wg.Add(1)
+		go func(id int) {
+			defer wg.Done()
+			forwardID := string(rune('a' + id))
+			s.watchdog.RegisterWorker(forwardID, nil)
+			for j := 0; j < 10; j++ {
+				s.watchdog.Heartbeat(forwardID)
+				time.Sleep(10 * time.Millisecond)
+			}
+			s.watchdog.UnregisterWorker(forwardID)
+		}(i)
+	}
+
+	wg.Wait()
+	// If we get here without deadlocks or panics, test passes
+}
+
+// TestStopWatchdog tests that stopping watchdog cleans up properly
+func TestStopWatchdog(t *testing.T) {
+	watchdog := NewWatchdog(100*time.Millisecond, 300*time.Millisecond)
+	watchdog.Start()
+
+	callbackCalled := false
+	callback := func(forwardID string) {
+		callbackCalled = true
+	}
+
+	watchdog.RegisterWorker("test-forward", callback)
+	watchdog.Heartbeat("test-forward")
+
+	// Stop watchdog before hang detection
+	time.Sleep(100 * time.Millisecond)
+	watchdog.Stop()
+
+	// Wait to ensure no more callbacks after stop
+	time.Sleep(500 * time.Millisecond)
+
+	assert.False(t, callbackCalled, "Callback should not be called after watchdog is stopped")
+}
+
+// TestWatchdogWithZeroHeartbeats tests detecting hung worker that never sends heartbeats
+func (s *WatchdogTestSuite) TestWatchdogWithZeroHeartbeats() {
+	callbackCalled := make(chan string, 1)
+	callback := func(forwardID string) {
+		callbackCalled <- forwardID
+	}
+
+	// Register worker but never send heartbeat
+	s.watchdog.RegisterWorker("test-forward", callback)
+
+	// Wait for hung detection
+	timeout := time.After(1 * time.Second)
+
+	select {
+	case forwardID := <-callbackCalled:
+		assert.Equal(s.T(), "test-forward", forwardID)
+	case <-timeout:
+		s.T().Fatal("Timeout waiting for hung worker callback")
+	}
+}

internal/forward/worker.go

@@ -5,6 +5,7 @@ import (
 	"fmt"
 	"io"
 	"log"
+	"sync"
 	"time"
 
 	"github.com/nvm/kportal/internal/config"
@@ -20,21 +21,25 @@ const (
 
 // ForwardWorker manages a single port-forward connection with automatic retry.
 type ForwardWorker struct {
-	forward       config.Forward
-	portForwarder *k8s.PortForwarder
-	ctx           context.Context
-	cancel        context.CancelFunc
-	stopChan      chan struct{}
-	doneChan      chan struct{}
-	verbose       bool
-	lastPod       string // Track the last pod we connected to
-	statusUI      StatusUpdater
-	healthChecker *healthcheck.Checker
-	startTime     time.Time // Track when the worker started
+	forward         config.Forward
+	portForwarder   *k8s.PortForwarder
+	ctx             context.Context
+	cancel          context.CancelFunc
+	stopChan        chan struct{}
+	doneChan        chan struct{}
+	reconnectChan   chan string // Channel to trigger reconnection
+	verbose         bool
+	lastPod         string // Track the last pod we connected to
+	statusUI        StatusUpdater
+	healthChecker   *healthcheck.Checker
+	watchdog        *Watchdog
+	startTime       time.Time          // Track when the worker started
+	forwardCancel   context.CancelFunc // Cancel function for current forward attempt
+	forwardCancelMu sync.Mutex         // Protects forwardCancel
 }
 
 // NewForwardWorker creates a new ForwardWorker for a single forward configuration.
-func NewForwardWorker(fwd config.Forward, portForwarder *k8s.PortForwarder, verbose bool, statusUI StatusUpdater, healthChecker *healthcheck.Checker) *ForwardWorker {
+func NewForwardWorker(fwd config.Forward, portForwarder *k8s.PortForwarder, verbose bool, statusUI StatusUpdater, healthChecker *healthcheck.Checker, watchdog *Watchdog) *ForwardWorker {
 	ctx, cancel := context.WithCancel(context.Background())
 
 	return &ForwardWorker{
@@ -44,13 +49,32 @@ func NewForwardWorker(fwd config.Forward, portForwarder *k8s.PortForwarder, verb
 		cancel:        cancel,
 		stopChan:      make(chan struct{}),
 		doneChan:      make(chan struct{}),
+		reconnectChan: make(chan string, 1), // Buffered to avoid blocking
 		verbose:       verbose,
 		statusUI:      statusUI,
 		healthChecker: healthChecker,
+		watchdog:      watchdog,
 		startTime:     time.Now(),
 	}
 }
 
+// TriggerReconnect triggers a reconnection (e.g., due to stale connection)
+func (w *ForwardWorker) TriggerReconnect(reason string) {
+	// Cancel current forward if running
+	w.forwardCancelMu.Lock()
+	if w.forwardCancel != nil {
+		w.forwardCancel()
+	}
+	w.forwardCancelMu.Unlock()
+
+	// Send reconnect signal (non-blocking)
+	select {
+	case w.reconnectChan <- reason:
+	default:
+		// Channel already has pending reconnect
+	}
+}
+
 // Start begins the port-forward worker in a goroutine.
 // The worker will continuously retry on failures with exponential backoff.
 func (w *ForwardWorker) Start() {
@@ -61,13 +85,28 @@ func (w *ForwardWorker) Start() {
 func (w *ForwardWorker) Stop() {
 	w.cancel()
 	close(w.stopChan)
-	<-w.doneChan // Wait for worker to finish
+
+	// Wait for worker to finish with timeout to prevent blocking forever
+	select {
+	case <-w.doneChan:
+		// Worker finished gracefully
+	case <-time.After(3 * time.Second):
+		// Worker didn't finish in time, but we've cancelled its context
+		// so it will clean up eventually
+		log.Printf("[%s] Worker stop timed out, continuing...", w.forward.ID())
+	}
 }
 
 // run is the main worker loop that handles retries.
 func (w *ForwardWorker) run() {
 	defer close(w.doneChan)
 
+	// Start heartbeat goroutine to continuously send heartbeats to watchdog
+	// This prevents false "hung worker" detection when connections are long-lived
+	if w.watchdog != nil {
+		go w.heartbeatLoop()
+	}
+
 	backoff := retry.NewBackoff()
 
 	for {
@@ -173,6 +212,26 @@ func (w *ForwardWorker) run() {
 	}
 }
 
+// heartbeatLoop sends periodic heartbeats to the watchdog to prove the worker is alive
+// This runs in a separate goroutine and continues throughout the worker's lifetime
+func (w *ForwardWorker) heartbeatLoop() {
+	// Send heartbeats every 15 seconds (well within typical 60s watchdog timeout)
+	ticker := time.NewTicker(15 * time.Second)
+	defer ticker.Stop()
+
+	// Send immediate heartbeat
+	w.watchdog.Heartbeat(w.forward.ID())
+
+	for {
+		select {
+		case <-ticker.C:
+			w.watchdog.Heartbeat(w.forward.ID())
+		case <-w.ctx.Done():
+			return
+		}
+	}
+}
+
 // establishForward establishes a port-forward connection.
 // This blocks until the connection is closed or an error occurs.
 func (w *ForwardWorker) establishForward(podName string) error {
@@ -184,11 +243,24 @@ func (w *ForwardWorker) establishForward(podName string) error {
 	forwardCtx, forwardCancel := context.WithCancel(w.ctx)
 	defer forwardCancel()
 
-	// Start a goroutine to monitor for stop signal
+	// Store cancel function so TriggerReconnect can use it
+	w.forwardCancelMu.Lock()
+	w.forwardCancel = forwardCancel
+	w.forwardCancelMu.Unlock()
+
+	defer func() {
+		w.forwardCancelMu.Lock()
+		w.forwardCancel = nil
+		w.forwardCancelMu.Unlock()
+	}()
+
+	// Start a goroutine to monitor for stop signal and reconnect triggers
 	go func() {
 		select {
 		case <-w.stopChan:
 			close(stopChan)
+		case <-w.reconnectChan:
+			close(stopChan)
 		case <-forwardCtx.Done():
 			close(stopChan)
 		}
@@ -230,6 +302,10 @@ func (w *ForwardWorker) establishForward(podName string) error {
 		if w.verbose {
 			log.Printf("[%s] Port-forward connection established", w.forward.ID())
 		}
+		// Mark connection as established in health checker
+		if w.healthChecker != nil {
+			w.healthChecker.MarkConnected(w.forward.ID())
+		}
 	case err := <-errChan:
 		return fmt.Errorf("failed to establish forward: %w", err)
 	case <-w.ctx.Done():

internal/healthcheck/checker.go

@@ -3,13 +3,17 @@ package healthcheck
 import (
 	"context"
 	"fmt"
+	"io"
 	"net"
 	"sync"
 	"time"
+
+	"github.com/nvm/kportal/internal/config"
 )
 
 const (
 	startupGracePeriod = 10 * time.Second
+	dataTransferSize   = 1024 // bytes to read in data transfer test
 )
 
 // Status represents the health status of a port forward
@@ -20,15 +24,26 @@ const (
 	StatusUnhealthy Status = "Error"
 	StatusStarting  Status = "Starting"
 	StatusReconnect Status = "Reconnecting"
+	StatusStale     Status = "Stale" // Connection is old or idle
+)
+
+// CheckMethod represents the health check method
+type CheckMethod string
+
+const (
+	CheckMethodTCPDial      CheckMethod = "tcp-dial"      // Simple TCP connection test
+	CheckMethodDataTransfer CheckMethod = "data-transfer" // Try to read data from connection
 )
 
 // PortHealth represents the health status of a single port
 type PortHealth struct {
-	Port         int
-	LastCheck    time.Time
-	Status       Status
-	ErrorMessage string
-	RegisteredAt time.Time // When this port was registered
+	Port           int
+	LastCheck      time.Time
+	Status         Status
+	ErrorMessage   string
+	RegisteredAt   time.Time // When this port was registered
+	ConnectionTime time.Time // When current connection was established
+	LastActivity   time.Time // Last time data was transferred
 }
 
 // StatusCallback is called when a port's health status changes
@@ -36,26 +51,52 @@ type StatusCallback func(forwardID string, status Status, errorMsg string)
 
 // Checker performs periodic health checks on local ports
 type Checker struct {
-	mu        sync.RWMutex
-	ports     map[string]*PortHealth // key: forward ID
-	callbacks map[string]StatusCallback
-	interval  time.Duration
-	timeout   time.Duration
-	ctx       context.Context
-	cancel    context.CancelFunc
-	wg        sync.WaitGroup
+	mu               sync.RWMutex
+	ports            map[string]*PortHealth // key: forward ID
+	callbacks        map[string]StatusCallback
+	interval         time.Duration
+	timeout          time.Duration
+	method           CheckMethod
+	maxConnectionAge time.Duration
+	maxIdleTime      time.Duration
+	ctx              context.Context
+	cancel           context.CancelFunc
+	wg               sync.WaitGroup
 }
 
-// NewChecker creates a new health checker
+// CheckerOptions configures the health checker
+type CheckerOptions struct {
+	Interval         time.Duration
+	Timeout          time.Duration
+	Method           CheckMethod
+	MaxConnectionAge time.Duration
+	MaxIdleTime      time.Duration
+}
+
+// NewChecker creates a new health checker with default options
 func NewChecker(interval, timeout time.Duration) *Checker {
+	return NewCheckerWithOptions(CheckerOptions{
+		Interval:         interval,
+		Timeout:          timeout,
+		Method:           CheckMethodDataTransfer,
+		MaxConnectionAge: config.DefaultMaxConnectionAge,
+		MaxIdleTime:      config.DefaultMaxIdleTime,
+	})
+}
+
+// NewCheckerWithOptions creates a new health checker with custom options
+func NewCheckerWithOptions(opts CheckerOptions) *Checker {
 	ctx, cancel := context.WithCancel(context.Background())
 	return &Checker{
-		ports:     make(map[string]*PortHealth),
-		callbacks: make(map[string]StatusCallback),
-		interval:  interval,
-		timeout:   timeout,
-		ctx:       ctx,
-		cancel:    cancel,
+		ports:            make(map[string]*PortHealth),
+		callbacks:        make(map[string]StatusCallback),
+		interval:         opts.Interval,
+		timeout:          opts.Timeout,
+		method:           opts.Method,
+		maxConnectionAge: opts.MaxConnectionAge,
+		maxIdleTime:      opts.MaxIdleTime,
+		ctx:              ctx,
+		cancel:           cancel,
 	}
 }
 
@@ -64,11 +105,14 @@ func (c *Checker) Register(forwardID string, port int, callback StatusCallback)
 	c.mu.Lock()
 	defer c.mu.Unlock()
 
+	now := time.Now()
 	c.ports[forwardID] = &PortHealth{
-		Port:         port,
-		LastCheck:    time.Time{},
-		Status:       StatusStarting,
-		RegisteredAt: time.Now(),
+		Port:           port,
+		LastCheck:      time.Time{},
+		Status:         StatusStarting,
+		RegisteredAt:   now,
+		ConnectionTime: now,
+		LastActivity:   now,
 	}
 	c.callbacks[forwardID] = callback
 
@@ -77,6 +121,28 @@ func (c *Checker) Register(forwardID string, port int, callback StatusCallback)
 	go c.checkLoop(forwardID)
 }
 
+// MarkConnected marks a forward as having established a new connection
+func (c *Checker) MarkConnected(forwardID string) {
+	c.mu.Lock()
+	defer c.mu.Unlock()
+
+	if health, exists := c.ports[forwardID]; exists {
+		now := time.Now()
+		health.ConnectionTime = now
+		health.LastActivity = now
+	}
+}
+
+// RecordActivity records data transfer activity for a forward
+func (c *Checker) RecordActivity(forwardID string) {
+	c.mu.Lock()
+	defer c.mu.Unlock()
+
+	if health, exists := c.ports[forwardID]; exists {
+		health.LastActivity = time.Now()
+	}
+}
+
 // Unregister removes a port from monitoring
 func (c *Checker) Unregister(forwardID string) {
 	c.mu.Lock()
@@ -86,44 +152,34 @@ func (c *Checker) Unregister(forwardID string) {
 	delete(c.callbacks, forwardID)
 }
 
-// MarkReconnecting marks a forward as reconnecting (called by worker)
-func (c *Checker) MarkReconnecting(forwardID string) {
+// markStatus is a helper to set a forward's status and notify on change.
+func (c *Checker) markStatus(forwardID string, newStatus Status) {
 	c.mu.Lock()
 
-	if health, exists := c.ports[forwardID]; exists {
-		oldStatus := health.Status
-		health.Status = StatusReconnect
-		health.LastCheck = time.Now()
-
+	health, exists := c.ports[forwardID]
+	if !exists {
 		c.mu.Unlock()
-
-		if oldStatus != StatusReconnect {
-			c.notifyStatusChange(forwardID, StatusReconnect, "")
-		}
 		return
 	}
 
+	oldStatus := health.Status
+	health.Status = newStatus
+	health.LastCheck = time.Now()
 	c.mu.Unlock()
+
+	if oldStatus != newStatus {
+		c.notifyStatusChange(forwardID, newStatus, "")
+	}
+}
+
+// MarkReconnecting marks a forward as reconnecting (called by worker)
+func (c *Checker) MarkReconnecting(forwardID string) {
+	c.markStatus(forwardID, StatusReconnect)
 }
 
 // MarkStarting marks a forward as starting (called by worker)
 func (c *Checker) MarkStarting(forwardID string) {
-	c.mu.Lock()
-
-	if health, exists := c.ports[forwardID]; exists {
-		oldStatus := health.Status
-		health.Status = StatusStarting
-		health.LastCheck = time.Now()
-
-		c.mu.Unlock()
-
-		if oldStatus != StatusStarting {
-			c.notifyStatusChange(forwardID, StatusStarting, "")
-		}
-		return
-	}
-
-	c.mu.Unlock()
+	c.markStatus(forwardID, StatusStarting)
 }
 
 // GetStatus returns the current health status of a forward
@@ -137,6 +193,17 @@ func (c *Checker) GetStatus(forwardID string) (Status, bool) {
 	return StatusUnhealthy, false
 }
 
+// GetLastCheckTime returns the last health check time for a forward
+func (c *Checker) GetLastCheckTime(forwardID string) (time.Time, bool) {
+	c.mu.RLock()
+	defer c.mu.RUnlock()
+
+	if health, exists := c.ports[forwardID]; exists {
+		return health.LastCheck, true
+	}
+	return time.Time{}, false
+}
+
 // GetAllErrors returns all forwards with errors and their error messages
 func (c *Checker) GetAllErrors() map[string]string {
 	c.mu.RLock()
@@ -197,38 +264,64 @@ func (c *Checker) checkPort(forwardID string) {
 	port := health.Port
 	oldStatus := health.Status
 	registeredAt := health.RegisteredAt
+	connectionTime := health.ConnectionTime
+	lastActivity := health.LastActivity
 	c.mu.RUnlock()
 
-	// Attempt to connect to the local port
-	ctx, cancel := context.WithTimeout(c.ctx, c.timeout)
-	defer cancel()
-
-	var d net.Dialer
-	conn, err := d.DialContext(ctx, "tcp", fmt.Sprintf("127.0.0.1:%d", port))
-
+	now := time.Now()
 	newStatus := StatusHealthy
 	errorMsg := ""
 
-	if err != nil {
-		// Grace period: if forward is less than 10 seconds old, keep it as "Starting"
-		// This avoids scary "Error" messages during initial connection attempts
-		timeSinceStart := time.Since(registeredAt)
-		if timeSinceStart < startupGracePeriod {
-			newStatus = StatusStarting
-		} else {
-			newStatus = StatusUnhealthy
-		}
-		errorMsg = err.Error()
+	// Check for stale connections based on age or idle time
+	connectionAge := now.Sub(connectionTime)
+	idleTime := now.Sub(lastActivity)
+
+	// Only enforce max connection age if the connection is ALSO idle
+	// This prevents interrupting active transfers (e.g., database dumps)
+	if c.maxConnectionAge > 0 && connectionAge > c.maxConnectionAge && idleTime > c.maxIdleTime {
+		newStatus = StatusStale
+		errorMsg = fmt.Sprintf("connection age %v exceeds max %v (and idle for %v)",
+			connectionAge.Round(time.Second), c.maxConnectionAge, idleTime.Round(time.Second))
+	} else if c.maxIdleTime > 0 && idleTime > c.maxIdleTime {
+		newStatus = StatusStale
+		errorMsg = fmt.Sprintf("idle time %v exceeds max %v", idleTime.Round(time.Second), c.maxIdleTime)
 	} else {
-		conn.Close()
+		// Perform connectivity check
+		var checkErr error
+		switch c.method {
+		case CheckMethodDataTransfer:
+			checkErr = c.checkDataTransfer(port)
+		case CheckMethodTCPDial:
+			checkErr = c.checkTCPDial(port)
+		default:
+			checkErr = c.checkTCPDial(port)
+		}
+
+		if checkErr != nil {
+			// Grace period: if forward is less than 10 seconds old, keep it as "Starting"
+			// This avoids scary "Error" messages during initial connection attempts
+			timeSinceStart := now.Sub(registeredAt)
+			if timeSinceStart < startupGracePeriod {
+				newStatus = StatusStarting
+			} else {
+				newStatus = StatusUnhealthy
+			}
+			errorMsg = checkErr.Error()
+		}
 	}
 
 	// Update health status
 	c.mu.Lock()
 	if health, exists := c.ports[forwardID]; exists {
 		health.Status = newStatus
-		health.LastCheck = time.Now()
+		health.LastCheck = now
 		health.ErrorMessage = errorMsg
+
+		// Successful health check indicates connection is active
+		// This prevents false positives where healthy connections are marked as idle
+		if newStatus == StatusHealthy {
+			health.LastActivity = now
+		}
 	}
 	c.mu.Unlock()
 
@@ -238,6 +331,62 @@ func (c *Checker) checkPort(forwardID string) {
 	}
 }
 
+// checkTCPDial performs a simple TCP dial test
+func (c *Checker) checkTCPDial(port int) error {
+	ctx, cancel := context.WithTimeout(c.ctx, c.timeout)
+	defer cancel()
+
+	var d net.Dialer
+	conn, err := d.DialContext(ctx, "tcp", fmt.Sprintf("127.0.0.1:%d", port))
+	if err != nil {
+		return err
+	}
+	conn.Close()
+	return nil
+}
+
+// checkDataTransfer attempts to read data from the connection to verify tunnel health
+func (c *Checker) checkDataTransfer(port int) error {
+	ctx, cancel := context.WithTimeout(c.ctx, c.timeout)
+	defer cancel()
+
+	var d net.Dialer
+	conn, err := d.DialContext(ctx, "tcp", fmt.Sprintf("127.0.0.1:%d", port))
+	if err != nil {
+		return err
+	}
+	defer conn.Close()
+
+	// Set a short read deadline to detect hung connections
+	// We don't expect to receive data, but we want to verify the connection isn't hung
+	conn.SetReadDeadline(time.Now().Add(c.timeout))
+
+	// Try to read a small amount of data
+	// Most servers will either:
+	// 1. Send a banner (SSH, FTP, etc) - we'll read it successfully
+	// 2. Wait for client to send first (HTTP, postgres) - we'll timeout (which is OK)
+	// 3. Hung/stale connection - will timeout with different error
+	buf := make([]byte, dataTransferSize)
+	_, err = conn.Read(buf)
+
+	// We expect either:
+	// - No error (banner received)
+	// - EOF (connection closed by server after connect)
+	// - Timeout (server waiting for client)
+	// All of these indicate the tunnel is working
+	if err == nil || err == io.EOF {
+		return nil
+	}
+
+	// Timeout is acceptable - server is waiting for us to send data first
+	if netErr, ok := err.(net.Error); ok && netErr.Timeout() {
+		return nil
+	}
+
+	// Other errors indicate a problem
+	return fmt.Errorf("data transfer check failed: %w", err)
+}
+
 // notifyStatusChange calls the callback for a forward
 func (c *Checker) notifyStatusChange(forwardID string, status Status, errorMsg string) {
 	c.mu.RLock()

internal/healthcheck/checker_test.go

@@ -0,0 +1,551 @@
+package healthcheck
+
+import (
+	"fmt"
+	"net"
+	"sync"
+	"testing"
+	"time"
+
+	"github.com/stretchr/testify/assert"
+	"github.com/stretchr/testify/require"
+	"github.com/stretchr/testify/suite"
+)
+
+// HealthCheckTestSuite contains tests for the health checker
+type HealthCheckTestSuite struct {
+	suite.Suite
+	checker  *Checker
+	listener net.Listener
+	port     int
+}
+
+func TestHealthCheckSuite(t *testing.T) {
+	suite.Run(t, new(HealthCheckTestSuite))
+}
+
+func (s *HealthCheckTestSuite) SetupTest() {
+	// Create a test listener on a random port
+	ln, err := net.Listen("tcp", "127.0.0.1:0")
+	require.NoError(s.T(), err)
+	s.listener = ln
+	s.port = ln.Addr().(*net.TCPAddr).Port
+
+	// Create checker with fast intervals for testing
+	s.checker = NewCheckerWithOptions(CheckerOptions{
+		Interval:         100 * time.Millisecond,
+		Timeout:          50 * time.Millisecond,
+		Method:           CheckMethodTCPDial,
+		MaxConnectionAge: 500 * time.Millisecond,
+		MaxIdleTime:      300 * time.Millisecond,
+	})
+}
+
+func (s *HealthCheckTestSuite) TearDownTest() {
+	if s.checker != nil {
+		s.checker.Stop()
+	}
+	if s.listener != nil {
+		s.listener.Close()
+	}
+}
+
+// TestRegisterAndUnregister tests basic registration and unregistration
+func (s *HealthCheckTestSuite) TestRegisterAndUnregister() {
+	callbackCalled := false
+	var callbackStatus Status
+	var mu sync.Mutex
+
+	callback := func(forwardID string, status Status, errorMsg string) {
+		mu.Lock()
+		defer mu.Unlock()
+		callbackCalled = true
+		callbackStatus = status
+	}
+
+	// Register port
+	s.checker.Register("test-forward", s.port, callback)
+
+	// Wait for health check to run
+	time.Sleep(200 * time.Millisecond)
+
+	// Verify callback was called with healthy status
+	mu.Lock()
+	assert.True(s.T(), callbackCalled, "Callback should have been called")
+	assert.Equal(s.T(), StatusHealthy, callbackStatus)
+	mu.Unlock()
+
+	// Unregister
+	s.checker.Unregister("test-forward")
+
+	// Verify port is no longer monitored
+	status, exists := s.checker.GetStatus("test-forward")
+	assert.False(s.T(), exists, "Port should no longer exist after unregister")
+	assert.Equal(s.T(), StatusUnhealthy, status)
+}
+
+// TestTCPDialMethod tests the TCP dial health check method
+func (s *HealthCheckTestSuite) TestTCPDialMethod() {
+	tests := []struct {
+		name           string
+		setupPort      bool
+		expectedStatus Status
+		description    string
+	}{
+		{
+			name:           "port available - healthy",
+			setupPort:      true,
+			expectedStatus: StatusHealthy,
+			description:    "When port is listening, status should be healthy",
+		},
+		{
+			name:           "port unavailable - unhealthy",
+			setupPort:      false,
+			expectedStatus: StatusUnhealthy,
+			description:    "When port is not listening, status should be unhealthy",
+		},
+	}
+
+	for _, tt := range tests {
+		s.Run(tt.name, func() {
+			var testPort int
+			var testListener net.Listener
+
+			if tt.setupPort {
+				// Use the existing listener
+				testPort = s.port
+			} else {
+				// Use a port that's not listening
+				testPort = 54321 // Likely unused port
+			}
+
+			// Create a new checker for this test
+			checker := NewCheckerWithOptions(CheckerOptions{
+				Interval:         100 * time.Millisecond,
+				Timeout:          50 * time.Millisecond,
+				Method:           CheckMethodTCPDial,
+				MaxConnectionAge: 0, // Disable for this test
+				MaxIdleTime:      0, // Disable for this test
+			})
+			defer checker.Stop()
+
+			checker.Register("test-forward", testPort, nil)
+
+			// Wait for health checks to complete
+			if !tt.setupPort {
+				// For unhealthy case, wait for grace period
+				time.Sleep(startupGracePeriod + 200*time.Millisecond)
+			} else {
+				time.Sleep(200 * time.Millisecond)
+			}
+
+			// Check status directly
+			status, exists := checker.GetStatus("test-forward")
+			assert.True(s.T(), exists)
+			assert.Equal(s.T(), tt.expectedStatus, status, tt.description)
+
+			if testListener != nil {
+				testListener.Close()
+			}
+		})
+	}
+}
+
+// TestDataTransferMethod tests the data transfer health check method
+func (s *HealthCheckTestSuite) TestDataTransferMethod() {
+	tests := []struct {
+		name           string
+		serverBehavior string // "banner", "silent", "close", "none"
+		expectedStatus Status
+	}{
+		{
+			name:           "server sends banner - healthy",
+			serverBehavior: "banner",
+			expectedStatus: StatusHealthy,
+		},
+		{
+			name:           "server waits silently - healthy (timeout OK)",
+			serverBehavior: "silent",
+			expectedStatus: StatusHealthy,
+		},
+		{
+			name:           "server closes connection - healthy (EOF OK)",
+			serverBehavior: "close",
+			expectedStatus: StatusHealthy,
+		},
+		{
+			name:           "no server listening - unhealthy",
+			serverBehavior: "none",
+			expectedStatus: StatusUnhealthy,
+		},
+	}
+
+	for _, tt := range tests {
+		s.Run(tt.name, func() {
+			var testPort int
+			var testListener net.Listener
+			var err error
+
+			if tt.serverBehavior != "none" {
+				// Start test server
+				testListener, err = net.Listen("tcp", "127.0.0.1:0")
+				require.NoError(s.T(), err)
+				testPort = testListener.Addr().(*net.TCPAddr).Port
+
+				// Handle connections based on behavior
+				go func() {
+					for {
+						conn, err := testListener.Accept()
+						if err != nil {
+							return
+						}
+						switch tt.serverBehavior {
+						case "banner":
+							conn.Write([]byte("220 Welcome\r\n"))
+							time.Sleep(50 * time.Millisecond)
+							conn.Close()
+						case "close":
+							conn.Close()
+						case "silent":
+							// Just keep connection open
+							time.Sleep(200 * time.Millisecond)
+							conn.Close()
+						}
+					}
+				}()
+				defer testListener.Close()
+			} else {
+				testPort = 54322 // Unused port
+			}
+
+			// Create checker with data transfer method
+			checker := NewCheckerWithOptions(CheckerOptions{
+				Interval:         100 * time.Millisecond,
+				Timeout:          50 * time.Millisecond,
+				Method:           CheckMethodDataTransfer,
+				MaxConnectionAge: 0, // Disable for this test
+				MaxIdleTime:      0, // Disable for this test
+			})
+			defer checker.Stop()
+
+			checker.Register("test-forward", testPort, nil)
+
+			// Wait for health checks to complete
+			if tt.serverBehavior == "none" {
+				// For unhealthy case, wait for grace period
+				time.Sleep(startupGracePeriod + 200*time.Millisecond)
+			} else {
+				time.Sleep(300 * time.Millisecond)
+			}
+
+			// Check status directly
+			status, exists := checker.GetStatus("test-forward")
+			assert.True(s.T(), exists)
+			assert.Equal(s.T(), tt.expectedStatus, status)
+		})
+	}
+}
+
+// TestConnectionAgeDetection tests max connection age detection
+func (s *HealthCheckTestSuite) TestConnectionAgeDetection() {
+	statusChanges := make(chan Status, 10)
+	callback := func(forwardID string, status Status, errorMsg string) {
+		statusChanges <- status
+	}
+
+	// Create checker with very short max connection age
+	checker := NewCheckerWithOptions(CheckerOptions{
+		Interval:         50 * time.Millisecond,
+		Timeout:          25 * time.Millisecond,
+		Method:           CheckMethodTCPDial,
+		MaxConnectionAge: 150 * time.Millisecond, // Very short for testing
+		MaxIdleTime:      0,                      // Disable idle detection
+	})
+	defer checker.Stop()
+
+	checker.Register("test-forward", s.port, callback)
+
+	// Wait for initial healthy status
+	var gotHealthy, gotStale bool
+	timeout := time.After(1 * time.Second)
+
+	for {
+		select {
+		case status := <-statusChanges:
+			if status == StatusHealthy || status == StatusStarting {
+				gotHealthy = true
+			}
+			if status == StatusStale {
+				gotStale = true
+			}
+			if gotHealthy && gotStale {
+				return // Test passed
+			}
+		case <-timeout:
+			s.T().Fatalf("Expected StatusStale after max connection age exceeded. gotHealthy=%v, gotStale=%v",
+				gotHealthy, gotStale)
+		}
+	}
+}
+
+// TestIdleTimeDetection tests that connections with passing health checks are NOT marked as stale
+// This verifies that successful health checks update LastActivity, preventing false idle detection
+func (s *HealthCheckTestSuite) TestIdleTimeDetection() {
+	statusChanges := make(chan Status, 10)
+	callback := func(forwardID string, status Status, errorMsg string) {
+		statusChanges <- status
+	}
+
+	// Create checker with very short max idle time
+	checker := NewCheckerWithOptions(CheckerOptions{
+		Interval:         50 * time.Millisecond,
+		Timeout:          25 * time.Millisecond,
+		Method:           CheckMethodTCPDial,
+		MaxConnectionAge: 0,                      // Disable age detection
+		MaxIdleTime:      150 * time.Millisecond, // Very short for testing
+	})
+	defer checker.Stop()
+
+	checker.Register("test-forward", s.port, callback)
+
+	// Wait long enough that idle time WOULD be exceeded if health checks didn't update LastActivity
+	time.Sleep(500 * time.Millisecond)
+
+	// Verify connection is still healthy, not stale
+	// This proves that successful health checks are updating LastActivity
+	status, exists := checker.GetStatus("test-forward")
+	require.True(s.T(), exists)
+	assert.Equal(s.T(), StatusHealthy, status, "Connection with passing health checks should NOT be marked as stale")
+
+	// Verify we never received a StatusStale callback
+	select {
+	case status := <-statusChanges:
+		if status == StatusStale {
+			s.T().Fatal("Connection should NOT be marked as stale when health checks are passing")
+		}
+	default:
+		// No stale status - this is correct
+	}
+}
+
+// TestMarkConnected tests that MarkConnected resets connection time
+func (s *HealthCheckTestSuite) TestMarkConnected() {
+	checker := NewCheckerWithOptions(CheckerOptions{
+		Interval:         50 * time.Millisecond,
+		Timeout:          25 * time.Millisecond,
+		Method:           CheckMethodTCPDial,
+		MaxConnectionAge: 200 * time.Millisecond,
+		MaxIdleTime:      0,
+	})
+	defer checker.Stop()
+
+	statusChanges := make(chan Status, 10)
+	callback := func(forwardID string, status Status, errorMsg string) {
+		statusChanges <- status
+	}
+
+	checker.Register("test-forward", s.port, callback)
+
+	// Wait a bit
+	time.Sleep(100 * time.Millisecond)
+
+	// Mark as reconnected (resets connection time)
+	checker.MarkConnected("test-forward")
+
+	// Wait for connection age to exceed (relative to first connection time)
+	time.Sleep(200 * time.Millisecond)
+
+	// Check status - should still be healthy because we reset connection time
+	status, exists := checker.GetStatus("test-forward")
+	assert.True(s.T(), exists)
+	// Note: Might be StatusStale by now, but the key is that MarkConnected delayed it
+	// This is a timing-sensitive test, so we just verify the functionality exists
+	_ = status
+}
+
+// TestRecordActivity tests that RecordActivity resets idle time
+func (s *HealthCheckTestSuite) TestRecordActivity() {
+	checker := NewCheckerWithOptions(CheckerOptions{
+		Interval:         50 * time.Millisecond,
+		Timeout:          25 * time.Millisecond,
+		Method:           CheckMethodTCPDial,
+		MaxConnectionAge: 0,
+		MaxIdleTime:      200 * time.Millisecond,
+	})
+	defer checker.Stop()
+
+	statusChanges := make(chan Status, 10)
+	callback := func(forwardID string, status Status, errorMsg string) {
+		statusChanges <- status
+	}
+
+	checker.Register("test-forward", s.port, callback)
+
+	// Periodically record activity to prevent idle detection
+	ticker := time.NewTicker(80 * time.Millisecond)
+	defer ticker.Stop()
+
+	go func() {
+		for i := 0; i < 5; i++ {
+			<-ticker.C
+			checker.RecordActivity("test-forward")
+		}
+	}()
+
+	// Wait longer than idle timeout
+	time.Sleep(500 * time.Millisecond)
+
+	// Should still be healthy due to activity
+	status, exists := checker.GetStatus("test-forward")
+	assert.True(s.T(), exists)
+	// May transition to stale eventually, but activity recording should have delayed it
+	_ = status
+}
+
+// TestMarkReconnecting tests the MarkReconnecting functionality
+func (s *HealthCheckTestSuite) TestMarkReconnecting() {
+	statusChanges := make(chan Status, 10)
+	callback := func(forwardID string, status Status, errorMsg string) {
+		statusChanges <- status
+	}
+
+	s.checker.Register("test-forward", s.port, callback)
+
+	// Wait for initial status
+	time.Sleep(150 * time.Millisecond)
+
+	// Mark as reconnecting
+	s.checker.MarkReconnecting("test-forward")
+
+	// Should receive reconnecting status
+	timeout := time.After(500 * time.Millisecond)
+	gotReconnect := false
+	for !gotReconnect {
+		select {
+		case status := <-statusChanges:
+			if status == StatusReconnect {
+				gotReconnect = true
+			}
+		case <-timeout:
+			s.T().Fatal("Expected StatusReconnect")
+		}
+	}
+}
+
+// TestStartingGracePeriod tests that errors during grace period show as "Starting"
+func (s *HealthCheckTestSuite) TestStartingGracePeriod() {
+	// Use a port that's not listening
+	unavailablePort := 54323
+
+	checker := NewCheckerWithOptions(CheckerOptions{
+		Interval:         50 * time.Millisecond,
+		Timeout:          25 * time.Millisecond,
+		Method:           CheckMethodTCPDial,
+		MaxConnectionAge: 0,
+		MaxIdleTime:      0,
+	})
+	defer checker.Stop()
+
+	// Register without callback - we'll check status directly
+	checker.Register("test-forward", unavailablePort, nil)
+
+	// Immediately check status - should be Starting or not yet checked
+	status, exists := checker.GetStatus("test-forward")
+	assert.True(s.T(), exists)
+	// Initially should be Starting
+	assert.Equal(s.T(), StatusStarting, status)
+
+	// Wait for grace period to expire
+	time.Sleep(startupGracePeriod + 200*time.Millisecond)
+
+	// Now should be Unhealthy
+	status, exists = checker.GetStatus("test-forward")
+	assert.True(s.T(), exists)
+	assert.Equal(s.T(), StatusUnhealthy, status)
+}
+
+// TestGetAllErrors tests retrieving all error messages
+func (s *HealthCheckTestSuite) TestGetAllErrors() {
+	// Create a new checker with faster intervals for this test
+	checker := NewCheckerWithOptions(CheckerOptions{
+		Interval:         100 * time.Millisecond,
+		Timeout:          50 * time.Millisecond,
+		Method:           CheckMethodTCPDial,
+		MaxConnectionAge: 0,
+		MaxIdleTime:      0,
+	})
+	defer checker.Stop()
+
+	// Register multiple forwards
+	checker.Register("forward1", s.port, nil)
+	checker.Register("forward2", 54324, nil) // Unavailable port
+
+	// Wait for grace period to expire
+	time.Sleep(startupGracePeriod + 300*time.Millisecond)
+
+	errors := checker.GetAllErrors()
+
+	// forward2 should have an error
+	_, hasError := errors["forward2"]
+	assert.True(s.T(), hasError, "forward2 should have an error")
+
+	// forward1 should not have an error
+	_, hasError = errors["forward1"]
+	assert.False(s.T(), hasError, "forward1 should not have an error")
+}
+
+// TestConcurrentOperations tests thread safety
+func (s *HealthCheckTestSuite) TestConcurrentOperations() {
+	var wg sync.WaitGroup
+	numGoroutines := 10
+
+	for i := 0; i < numGoroutines; i++ {
+		wg.Add(1)
+		go func(id int) {
+			defer wg.Done()
+			forwardID := fmt.Sprintf("forward-%d", id)
+			s.checker.Register(forwardID, s.port, nil)
+			time.Sleep(50 * time.Millisecond)
+			s.checker.MarkConnected(forwardID)
+			s.checker.RecordActivity(forwardID)
+			status, _ := s.checker.GetStatus(forwardID)
+			_ = status
+			s.checker.Unregister(forwardID)
+		}(i)
+	}
+
+	wg.Wait()
+	// If we get here without deadlocks or panics, test passes
+}
+
+// TestDefaultOptions tests that NewChecker uses sensible defaults
+func TestDefaultOptions(t *testing.T) {
+	checker := NewChecker(5*time.Second, 2*time.Second)
+	defer checker.Stop()
+
+	assert.Equal(t, 5*time.Second, checker.interval)
+	assert.Equal(t, 2*time.Second, checker.timeout)
+	assert.Equal(t, CheckMethodDataTransfer, checker.method)
+	assert.Equal(t, 25*time.Minute, checker.maxConnectionAge)
+	assert.Equal(t, 10*time.Minute, checker.maxIdleTime)
+}
+
+// TestCustomOptions tests NewCheckerWithOptions
+func TestCustomOptions(t *testing.T) {
+	opts := CheckerOptions{
+		Interval:         1 * time.Second,
+		Timeout:          500 * time.Millisecond,
+		Method:           CheckMethodTCPDial,
+		MaxConnectionAge: 5 * time.Minute,
+		MaxIdleTime:      2 * time.Minute,
+	}
+
+	checker := NewCheckerWithOptions(opts)
+	defer checker.Stop()
+
+	assert.Equal(t, 1*time.Second, checker.interval)
+	assert.Equal(t, 500*time.Millisecond, checker.timeout)
+	assert.Equal(t, CheckMethodTCPDial, checker.method)
+	assert.Equal(t, 5*time.Minute, checker.maxConnectionAge)
+	assert.Equal(t, 2*time.Minute, checker.maxIdleTime)
+}

internal/k8s/discovery.go

@@ -9,6 +9,8 @@ import (
 
 	corev1 "k8s.io/api/core/v1"
 	metav1 "k8s.io/apimachinery/pkg/apis/meta/v1"
+	"k8s.io/apimachinery/pkg/util/intstr"
+	"k8s.io/client-go/kubernetes"
 )
 
 // Discovery provides cluster introspection capabilities for the UI wizards.
@@ -41,9 +43,10 @@ type ContainerInfo struct {
 
 // PortInfo describes a port exposed by a container or service.
 type PortInfo struct {
-	Name     string
-	Port     int32
-	Protocol string
+	Name       string
+	Port       int32
+	TargetPort int32 // For services: the actual pod port to forward to
+	Protocol   string
 }
 
 // ServiceInfo contains information about a service.
@@ -205,7 +208,60 @@ func (d *Discovery) ListPodsWithSelector(ctx context.Context, contextName, names
 	return pods, nil
 }
 
+// resolveTargetPort resolves a service's targetPort to an actual port number.
+// If targetPort is numeric, it returns that number directly.
+// If targetPort is a named port, it looks up the port number from the backing pods.
+// Falls back to the service port if resolution fails.
+func (d *Discovery) resolveTargetPort(ctx context.Context, client kubernetes.Interface, namespace string, svc *corev1.Service, port *corev1.ServicePort) int32 {
+	// If targetPort is not set, Kubernetes defaults to the service port
+	if port.TargetPort.Type == intstr.Int && port.TargetPort.IntVal == 0 {
+		return port.Port
+	}
+
+	// If targetPort is numeric, use it directly
+	if port.TargetPort.Type == intstr.Int {
+		return port.TargetPort.IntVal
+	}
+
+	// targetPort is a named port - need to look up from pods
+	namedPort := port.TargetPort.StrVal
+	if namedPort == "" {
+		return port.Port
+	}
+
+	// Get a backing pod to resolve the named port
+	if len(svc.Spec.Selector) == 0 {
+		// No selector, can't resolve - fall back to service port
+		return port.Port
+	}
+
+	selector := metav1.FormatLabelSelector(&metav1.LabelSelector{MatchLabels: svc.Spec.Selector})
+	pods, err := client.CoreV1().Pods(namespace).List(ctx, metav1.ListOptions{
+		LabelSelector: selector,
+		Limit:         1, // We only need one pod to resolve the port name
+	})
+	if err != nil || len(pods.Items) == 0 {
+		// Can't get pods - fall back to service port
+		return port.Port
+	}
+
+	// Look up the named port in the pod's containers
+	pod := &pods.Items[0]
+	for _, container := range pod.Spec.Containers {
+		for _, containerPort := range container.Ports {
+			if containerPort.Name == namedPort {
+				return containerPort.ContainerPort
+			}
+		}
+	}
+
+	// Named port not found - fall back to service port
+	return port.Port
+}
+
 // ListServices returns all services in the given namespace.
+// For each service port, it resolves the targetPort to an actual port number
+// by looking up the backing pods when the targetPort is a named port.
 func (d *Discovery) ListServices(ctx context.Context, contextName, namespace string) ([]ServiceInfo, error) {
 	client, err := d.pool.GetClient(contextName)
 	if err != nil {
@@ -221,10 +277,13 @@ func (d *Discovery) ListServices(ctx context.Context, contextName, namespace str
 	for _, svc := range svcList.Items {
 		ports := make([]PortInfo, 0, len(svc.Spec.Ports))
 		for _, port := range svc.Spec.Ports {
+			targetPort := d.resolveTargetPort(ctx, client, namespace, &svc, &port)
+
 			ports = append(ports, PortInfo{
-				Name:     port.Name,
-				Port:     port.Port,
-				Protocol: string(port.Protocol),
+				Name:       port.Name,
+				Port:       port.Port,
+				TargetPort: targetPort,
+				Protocol:   string(port.Protocol),
 			})
 		}
 
@@ -292,12 +351,8 @@ func CheckPortAvailability(port int) (bool, string, error) {
 	addr := fmt.Sprintf(":%d", port)
 	listener, err := net.Listen("tcp", addr)
 	if err != nil {
-		// Port is in use
-		// Try to get process info (best-effort)
-		processInfo := "unknown process"
-		// Note: Getting process info requires platform-specific code
-		// For now, just return a generic message
-		return false, processInfo, nil
+		// Port is in use - return error details
+		return false, err.Error(), nil
 	}
 
 	// Port is available, close the listener

internal/k8s/discovery_test.go

@@ -0,0 +1,308 @@
+package k8s
+
+import (
+	"context"
+	"testing"
+
+	"github.com/stretchr/testify/assert"
+	corev1 "k8s.io/api/core/v1"
+	metav1 "k8s.io/apimachinery/pkg/apis/meta/v1"
+	"k8s.io/apimachinery/pkg/runtime"
+	"k8s.io/apimachinery/pkg/util/intstr"
+	"k8s.io/client-go/kubernetes/fake"
+)
+
+func TestResolveTargetPort(t *testing.T) {
+	tests := []struct {
+		name         string
+		servicePort  corev1.ServicePort
+		service      *corev1.Service
+		pods         []corev1.Pod
+		expectedPort int32
+		description  string
+	}{
+		{
+			name: "numeric targetPort",
+			servicePort: corev1.ServicePort{
+				Port:       80,
+				TargetPort: intstr.FromInt(8000),
+			},
+			service: &corev1.Service{
+				ObjectMeta: metav1.ObjectMeta{
+					Name:      "test-svc",
+					Namespace: "default",
+				},
+				Spec: corev1.ServiceSpec{
+					Selector: map[string]string{"app": "test"},
+				},
+			},
+			pods:         nil, // No pods needed for numeric targetPort
+			expectedPort: 8000,
+			description:  "should use numeric targetPort directly",
+		},
+		{
+			name: "named targetPort resolved from pod",
+			servicePort: corev1.ServicePort{
+				Name:       "http",
+				Port:       80,
+				TargetPort: intstr.FromString("http"),
+			},
+			service: &corev1.Service{
+				ObjectMeta: metav1.ObjectMeta{
+					Name:      "test-svc",
+					Namespace: "default",
+				},
+				Spec: corev1.ServiceSpec{
+					Selector: map[string]string{"app": "test"},
+				},
+			},
+			pods: []corev1.Pod{
+				{
+					ObjectMeta: metav1.ObjectMeta{
+						Name:      "test-pod",
+						Namespace: "default",
+						Labels:    map[string]string{"app": "test"},
+					},
+					Spec: corev1.PodSpec{
+						Containers: []corev1.Container{
+							{
+								Name: "main",
+								Ports: []corev1.ContainerPort{
+									{Name: "http", ContainerPort: 8000},
+								},
+							},
+						},
+					},
+				},
+			},
+			expectedPort: 8000,
+			description:  "should resolve named port from pod container",
+		},
+		{
+			name: "targetPort not set - defaults to service port",
+			servicePort: corev1.ServicePort{
+				Port:       80,
+				TargetPort: intstr.FromInt(0), // Not set
+			},
+			service: &corev1.Service{
+				ObjectMeta: metav1.ObjectMeta{
+					Name:      "test-svc",
+					Namespace: "default",
+				},
+				Spec: corev1.ServiceSpec{
+					Selector: map[string]string{"app": "test"},
+				},
+			},
+			pods:         nil,
+			expectedPort: 80,
+			description:  "should fall back to service port when targetPort is not set",
+		},
+		{
+			name: "named targetPort with no matching pod",
+			servicePort: corev1.ServicePort{
+				Name:       "http",
+				Port:       80,
+				TargetPort: intstr.FromString("http"),
+			},
+			service: &corev1.Service{
+				ObjectMeta: metav1.ObjectMeta{
+					Name:      "test-svc",
+					Namespace: "default",
+				},
+				Spec: corev1.ServiceSpec{
+					Selector: map[string]string{"app": "test"},
+				},
+			},
+			pods:         nil, // No pods available
+			expectedPort: 80,
+			description:  "should fall back to service port when no pods found",
+		},
+		{
+			name: "service without selector",
+			servicePort: corev1.ServicePort{
+				Name:       "http",
+				Port:       80,
+				TargetPort: intstr.FromString("http"),
+			},
+			service: &corev1.Service{
+				ObjectMeta: metav1.ObjectMeta{
+					Name:      "test-svc",
+					Namespace: "default",
+				},
+				Spec: corev1.ServiceSpec{
+					Selector: nil, // No selector
+				},
+			},
+			pods:         nil,
+			expectedPort: 80,
+			description:  "should fall back to service port when service has no selector",
+		},
+		{
+			name: "named targetPort not found in pod containers",
+			servicePort: corev1.ServicePort{
+				Name:       "http",
+				Port:       80,
+				TargetPort: intstr.FromString("nonexistent"),
+			},
+			service: &corev1.Service{
+				ObjectMeta: metav1.ObjectMeta{
+					Name:      "test-svc",
+					Namespace: "default",
+				},
+				Spec: corev1.ServiceSpec{
+					Selector: map[string]string{"app": "test"},
+				},
+			},
+			pods: []corev1.Pod{
+				{
+					ObjectMeta: metav1.ObjectMeta{
+						Name:      "test-pod",
+						Namespace: "default",
+						Labels:    map[string]string{"app": "test"},
+					},
+					Spec: corev1.PodSpec{
+						Containers: []corev1.Container{
+							{
+								Name: "main",
+								Ports: []corev1.ContainerPort{
+									{Name: "http", ContainerPort: 8000},
+								},
+							},
+						},
+					},
+				},
+			},
+			expectedPort: 80,
+			description:  "should fall back to service port when named port not found in pod",
+		},
+		{
+			name: "multiple containers with named port in second container",
+			servicePort: corev1.ServicePort{
+				Name:       "metrics",
+				Port:       9090,
+				TargetPort: intstr.FromString("metrics"),
+			},
+			service: &corev1.Service{
+				ObjectMeta: metav1.ObjectMeta{
+					Name:      "test-svc",
+					Namespace: "default",
+				},
+				Spec: corev1.ServiceSpec{
+					Selector: map[string]string{"app": "test"},
+				},
+			},
+			pods: []corev1.Pod{
+				{
+					ObjectMeta: metav1.ObjectMeta{
+						Name:      "test-pod",
+						Namespace: "default",
+						Labels:    map[string]string{"app": "test"},
+					},
+					Spec: corev1.PodSpec{
+						Containers: []corev1.Container{
+							{
+								Name: "main",
+								Ports: []corev1.ContainerPort{
+									{Name: "http", ContainerPort: 8000},
+								},
+							},
+							{
+								Name: "sidecar",
+								Ports: []corev1.ContainerPort{
+									{Name: "metrics", ContainerPort: 9100},
+								},
+							},
+						},
+					},
+				},
+			},
+			expectedPort: 9100,
+			description:  "should find named port in any container",
+		},
+	}
+
+	for _, tt := range tests {
+		t.Run(tt.name, func(t *testing.T) {
+			// Create fake client with pods
+			var objects []runtime.Object
+			for i := range tt.pods {
+				objects = append(objects, &tt.pods[i])
+			}
+			fakeClient := fake.NewSimpleClientset(objects...)
+
+			// Create discovery instance (we only need it to call resolveTargetPort)
+			d := &Discovery{}
+
+			// Call resolveTargetPort
+			result := d.resolveTargetPort(
+				context.Background(),
+				fakeClient,
+				"default",
+				tt.service,
+				&tt.servicePort,
+			)
+
+			assert.Equal(t, tt.expectedPort, result, tt.description)
+		})
+	}
+}
+
+func TestPortInfoTargetPort(t *testing.T) {
+	// Test that PortInfo correctly stores TargetPort
+	portInfo := PortInfo{
+		Name:       "http",
+		Port:       80,
+		TargetPort: 8000,
+		Protocol:   "TCP",
+	}
+
+	assert.Equal(t, int32(80), portInfo.Port)
+	assert.Equal(t, int32(8000), portInfo.TargetPort)
+	assert.Equal(t, "http", portInfo.Name)
+	assert.Equal(t, "TCP", portInfo.Protocol)
+}
+
+func TestGetUniquePorts(t *testing.T) {
+	// Test GetUniquePorts still works with the new PortInfo struct
+	pods := []PodInfo{
+		{
+			Name: "pod1",
+			Containers: []ContainerInfo{
+				{
+					Name: "main",
+					Ports: []PortInfo{
+						{Name: "http", Port: 8080},
+						{Name: "metrics", Port: 9090},
+					},
+				},
+			},
+		},
+		{
+			Name: "pod2",
+			Containers: []ContainerInfo{
+				{
+					Name: "main",
+					Ports: []PortInfo{
+						{Name: "http", Port: 8080}, // Duplicate
+						{Name: "grpc", Port: 50051},
+					},
+				},
+			},
+		},
+	}
+
+	ports := GetUniquePorts(pods)
+
+	// Should have 3 unique ports
+	assert.Len(t, ports, 3)
+
+	// Should be sorted by port number
+	assert.Equal(t, int32(8080), ports[0].Port)
+	assert.Equal(t, int32(9090), ports[1].Port)
+	assert.Equal(t, int32(50051), ports[2].Port)
+
+	// Names should be preserved
+	assert.Equal(t, "http", ports[0].Name)
+	assert.Equal(t, "metrics", ports[1].Name)
+	assert.Equal(t, "grpc", ports[2].Name)
+}

internal/k8s/portforward.go

@@ -4,9 +4,13 @@ import (
 	"context"
 	"fmt"
 	"io"
+	"net"
 	"net/http"
 	"net/url"
 	"strings"
+	"time"
+
+	"github.com/nvm/kportal/internal/config"
 
 	corev1 "k8s.io/api/core/v1"
 	metav1 "k8s.io/apimachinery/pkg/apis/meta/v1"
@@ -17,18 +21,32 @@ import (
 
 // PortForwarder handles Kubernetes port-forwarding operations.
 type PortForwarder struct {
-	clientPool *ClientPool
-	resolver   *ResourceResolver
+	clientPool   *ClientPool
+	resolver     *ResourceResolver
+	tcpKeepalive time.Duration // TCP keepalive interval
+	dialTimeout  time.Duration // Connection dial timeout
 }
 
-// NewPortForwarder creates a new PortForwarder instance.
+// NewPortForwarder creates a new PortForwarder instance with default settings.
 func NewPortForwarder(clientPool *ClientPool, resolver *ResourceResolver) *PortForwarder {
 	return &PortForwarder{
-		clientPool: clientPool,
-		resolver:   resolver,
+		clientPool:   clientPool,
+		resolver:     resolver,
+		tcpKeepalive: config.DefaultTCPKeepalive,
+		dialTimeout:  config.DefaultDialTimeout,
 	}
 }
 
+// SetTCPKeepalive configures the TCP keepalive interval for new connections.
+func (pf *PortForwarder) SetTCPKeepalive(keepalive time.Duration) {
+	pf.tcpKeepalive = keepalive
+}
+
+// SetDialTimeout configures the connection dial timeout.
+func (pf *PortForwarder) SetDialTimeout(timeout time.Duration) {
+	pf.dialTimeout = timeout
+}
+
 // ForwardRequest contains the parameters for a port-forward request.
 type ForwardRequest struct {
 	ContextName string // Kubernetes context name
@@ -124,6 +142,9 @@ func (pf *PortForwarder) forwardToService(ctx context.Context, req *ForwardReque
 	}
 
 	// Get pods backing the service using label selector
+	if len(service.Spec.Selector) == 0 {
+		return fmt.Errorf("service %s has no selector (headless service without selector cannot be port-forwarded)", serviceName)
+	}
 	selector := metav1.FormatLabelSelector(&metav1.LabelSelector{MatchLabels: service.Spec.Selector})
 	pods, err := client.CoreV1().Pods(req.Namespace).List(ctx, metav1.ListOptions{
 		LabelSelector: selector,
@@ -164,6 +185,19 @@ func (pf *PortForwarder) forwardToService(ctx context.Context, req *ForwardReque
 
 // executePortForward performs the actual port-forward operation.
 func (pf *PortForwarder) executePortForward(config *rest.Config, url *url.URL, req *ForwardRequest) error {
+	// Configure TCP settings on the underlying connection
+	// This is set in the rest.Config which will be used by the SPDY transport
+	if config.Dial == nil {
+		// Create a custom dialer with configurable timeout and keepalive
+		// - Timeout: How long to wait for connection to establish
+		// - KeepAlive: TCP keepalive helps OS detect dead connections at network layer
+		dialer := &net.Dialer{
+			Timeout:   pf.dialTimeout,  // Configurable dial timeout
+			KeepAlive: pf.tcpKeepalive, // Configurable keepalive interval
+		}
+		config.Dial = dialer.DialContext
+	}
+
 	// Create SPDY roundtripper
 	transport, upgrader, err := spdy.RoundTripperFor(config)
 	if err != nil {
@@ -228,6 +262,9 @@ func (pf *PortForwarder) GetPodForResource(ctx context.Context, contextName, nam
 			return "", fmt.Errorf("failed to get service: %w", err)
 		}
 
+		if len(service.Spec.Selector) == 0 {
+			return "", fmt.Errorf("service %s has no selector (headless service without selector cannot be port-forwarded)", resourceName)
+		}
 		selector := metav1.FormatLabelSelector(&metav1.LabelSelector{MatchLabels: service.Spec.Selector})
 		pods, err := client.CoreV1().Pods(namespace).List(ctx, metav1.ListOptions{
 			LabelSelector: selector,

internal/mdns/publisher.go

@@ -0,0 +1,220 @@
+package mdns
+
+import (
+	"fmt"
+	"net"
+	"sync"
+	"time"
+
+	"github.com/grandcat/zeroconf"
+	"github.com/nvm/kportal/internal/logger"
+)
+
+const (
+	// shutdownTimeout is the maximum time to wait for mDNS server shutdown
+	shutdownTimeout = 2 * time.Second
+
+	// mdnsDomain is the standard mDNS domain (RFC 6762)
+	// This is always ".local" for multicast DNS - it's not configurable
+	// and is different from your network's DNS search domain
+	mdnsDomain = "local"
+)
+
+// Publisher manages mDNS hostname registrations for port forwards.
+// It allows forwards with aliases to be accessible via <alias>.local hostnames.
+type Publisher struct {
+	mu       sync.RWMutex
+	servers  map[string]*zeroconf.Server // forwardID -> server
+	aliases  map[string]string           // forwardID -> alias (for logging)
+	enabled  bool
+	localIPs []string
+}
+
+// NewPublisher creates a new mDNS Publisher.
+// If enabled is false, all registration calls will be no-ops.
+func NewPublisher(enabled bool) *Publisher {
+	p := &Publisher{
+		servers:  make(map[string]*zeroconf.Server),
+		aliases:  make(map[string]string),
+		enabled:  enabled,
+		localIPs: getLocalIPs(),
+	}
+
+	if enabled {
+		logger.Info("mDNS publisher initialized", map[string]interface{}{
+			"domain":    mdnsDomain,
+			"local_ips": p.localIPs,
+		})
+	}
+
+	return p
+}
+
+// Register publishes an mDNS hostname for a forward.
+// The hostname will be <alias>.local and will resolve to 127.0.0.1.
+// If the forward has no alias or mDNS is disabled, this is a no-op.
+func (p *Publisher) Register(forwardID, alias string, localPort int) error {
+	if !p.enabled || alias == "" {
+		return nil
+	}
+
+	p.mu.Lock()
+	defer p.mu.Unlock()
+
+	// Check if already registered
+	if _, exists := p.servers[forwardID]; exists {
+		logger.Debug("mDNS hostname already registered", map[string]interface{}{
+			"forward_id": forwardID,
+			"alias":      alias,
+		})
+		return nil
+	}
+
+	// Register the mDNS service
+	// We use a generic service type and rely on the hostname registration
+	server, err := zeroconf.RegisterProxy(
+		alias,                 // Instance name (shown in service discovery)
+		"_kportal._tcp",       // Service type (custom for kportal)
+		"local.",              // Domain
+		localPort,             // Port
+		alias,                 // Hostname (will be <alias>.local)
+		[]string{"127.0.0.1"}, // IPs to resolve to
+		[]string{fmt.Sprintf("forward=%s", forwardID)}, // TXT records
+		nil, // interfaces (nil = all)
+	)
+	if err != nil {
+		return fmt.Errorf("failed to register mDNS for %s: %w", alias, err)
+	}
+
+	p.servers[forwardID] = server
+	p.aliases[forwardID] = alias
+
+	logger.Info("mDNS hostname registered", map[string]interface{}{
+		"forward_id": forwardID,
+		"hostname":   GetHostname(alias),
+		"port":       localPort,
+	})
+
+	return nil
+}
+
+// Unregister removes the mDNS hostname for a forward.
+func (p *Publisher) Unregister(forwardID string) {
+	if !p.enabled {
+		return
+	}
+
+	p.mu.Lock()
+	defer p.mu.Unlock()
+
+	server, exists := p.servers[forwardID]
+	if !exists {
+		return
+	}
+
+	alias := p.aliases[forwardID]
+	shutdownWithTimeout(server, forwardID)
+	delete(p.servers, forwardID)
+	delete(p.aliases, forwardID)
+
+	logger.Info("mDNS hostname unregistered", map[string]interface{}{
+		"forward_id": forwardID,
+		"hostname":   GetHostname(alias),
+	})
+}
+
+// Stop shuts down all mDNS registrations.
+func (p *Publisher) Stop() {
+	if !p.enabled {
+		return
+	}
+
+	p.mu.Lock()
+	defer p.mu.Unlock()
+
+	// Shutdown all servers concurrently with timeout
+	var wg sync.WaitGroup
+	for forwardID, server := range p.servers {
+		wg.Add(1)
+		go func(id string, srv *zeroconf.Server) {
+			defer wg.Done()
+			shutdownWithTimeout(srv, id)
+		}(forwardID, server)
+	}
+
+	// Wait for all shutdowns to complete (or timeout)
+	wg.Wait()
+
+	p.servers = make(map[string]*zeroconf.Server)
+	p.aliases = make(map[string]string)
+
+	logger.Info("mDNS publisher stopped", nil)
+}
+
+// shutdownWithTimeout attempts to shutdown a zeroconf server with a timeout.
+// If shutdown hangs, it logs a warning and returns anyway.
+func shutdownWithTimeout(server *zeroconf.Server, forwardID string) {
+	done := make(chan struct{})
+
+	go func() {
+		server.Shutdown()
+		close(done)
+	}()
+
+	select {
+	case <-done:
+		// Shutdown completed successfully
+	case <-time.After(shutdownTimeout):
+		logger.Warn("mDNS shutdown timed out, continuing anyway", map[string]interface{}{
+			"forward_id": forwardID,
+			"timeout":    shutdownTimeout.String(),
+		})
+	}
+}
+
+// IsEnabled returns whether mDNS publishing is enabled.
+func (p *Publisher) IsEnabled() bool {
+	return p.enabled
+}
+
+// GetDomain returns the mDNS domain being used (always "local" per RFC 6762).
+func (p *Publisher) GetDomain() string {
+	return mdnsDomain
+}
+
+// GetHostname returns the full mDNS hostname for an alias.
+// Example: GetHostname("myapp") returns "myapp.local"
+func GetHostname(alias string) string {
+	return alias + "." + mdnsDomain
+}
+
+// GetRegisteredCount returns the number of currently registered hostnames.
+func (p *Publisher) GetRegisteredCount() int {
+	p.mu.RLock()
+	defer p.mu.RUnlock()
+	return len(p.servers)
+}
+
+// getLocalIPs returns the local IP addresses for logging purposes.
+func getLocalIPs() []string {
+	var ips []string
+
+	addrs, err := net.InterfaceAddrs()
+	if err != nil {
+		return []string{"127.0.0.1"}
+	}
+
+	for _, addr := range addrs {
+		if ipnet, ok := addr.(*net.IPNet); ok && !ipnet.IP.IsLoopback() {
+			if ipnet.IP.To4() != nil {
+				ips = append(ips, ipnet.IP.String())
+			}
+		}
+	}
+
+	if len(ips) == 0 {
+		return []string{"127.0.0.1"}
+	}
+
+	return ips
+}

internal/mdns/publisher_test.go

@@ -0,0 +1,154 @@
+package mdns
+
+import (
+	"testing"
+
+	"github.com/stretchr/testify/assert"
+)
+
+// Note: Tests that actually register mDNS services require network I/O
+// and can be slow or hang in CI environments. We test the logic paths
+// without actually calling zeroconf for most tests.
+
+func TestNewPublisher_Disabled(t *testing.T) {
+	p := NewPublisher(false)
+
+	assert.False(t, p.IsEnabled())
+	assert.Equal(t, 0, p.GetRegisteredCount())
+}
+
+func TestNewPublisher_Enabled(t *testing.T) {
+	p := NewPublisher(true)
+
+	assert.True(t, p.IsEnabled())
+	assert.Equal(t, 0, p.GetRegisteredCount())
+}
+
+func TestRegister_WhenDisabled_NoOp(t *testing.T) {
+	p := NewPublisher(false)
+
+	err := p.Register("forward-1", "test-alias", 8080)
+
+	assert.NoError(t, err)
+	assert.Equal(t, 0, p.GetRegisteredCount())
+}
+
+func TestRegister_EmptyAlias_NoOp(t *testing.T) {
+	p := NewPublisher(true)
+
+	err := p.Register("forward-1", "", 8080)
+
+	assert.NoError(t, err)
+	assert.Equal(t, 0, p.GetRegisteredCount())
+}
+
+func TestUnregister_WhenDisabled_NoOp(t *testing.T) {
+	p := NewPublisher(false)
+
+	// Should not panic
+	p.Unregister("forward-1")
+}
+
+func TestUnregister_NotRegistered_NoOp(t *testing.T) {
+	p := NewPublisher(true)
+
+	// Should not panic
+	p.Unregister("non-existent")
+	assert.Equal(t, 0, p.GetRegisteredCount())
+}
+
+func TestStop_WhenDisabled_NoOp(t *testing.T) {
+	p := NewPublisher(false)
+
+	// Should not panic
+	p.Stop()
+}
+
+func TestStop_WhenNoRegistrations(t *testing.T) {
+	p := NewPublisher(true)
+
+	// Should not panic
+	p.Stop()
+	assert.Equal(t, 0, p.GetRegisteredCount())
+}
+
+func TestGetLocalIPs(t *testing.T) {
+	ips := getLocalIPs()
+
+	// Should return at least one IP
+	assert.NotEmpty(t, ips, "getLocalIPs should return at least one IP")
+
+	// All IPs should be non-empty strings
+	for _, ip := range ips {
+		assert.NotEmpty(t, ip, "IP address should not be empty")
+	}
+}
+
+// Integration tests - only run when explicitly requested
+// These tests actually register mDNS services and require network access
+
+func TestRegister_Integration(t *testing.T) {
+	if testing.Short() {
+		t.Skip("Skipping mDNS integration test in short mode")
+	}
+
+	p := NewPublisher(true)
+	defer p.Stop()
+
+	err := p.Register("forward-1", "test-service", 8080)
+
+	assert.NoError(t, err)
+	assert.Equal(t, 1, p.GetRegisteredCount())
+}
+
+func TestRegister_Duplicate_Idempotent_Integration(t *testing.T) {
+	if testing.Short() {
+		t.Skip("Skipping mDNS integration test in short mode")
+	}
+
+	p := NewPublisher(true)
+	defer p.Stop()
+
+	// First registration
+	err := p.Register("forward-1", "test-service", 8080)
+	assert.NoError(t, err)
+	assert.Equal(t, 1, p.GetRegisteredCount())
+
+	// Second registration with same ID should be idempotent
+	err = p.Register("forward-1", "test-service", 8080)
+	assert.NoError(t, err)
+	assert.Equal(t, 1, p.GetRegisteredCount())
+}
+
+func TestRegister_MultipleForwards_Integration(t *testing.T) {
+	if testing.Short() {
+		t.Skip("Skipping mDNS integration test in short mode")
+	}
+
+	p := NewPublisher(true)
+	defer p.Stop()
+
+	err1 := p.Register("forward-1", "service-a", 8080)
+	err2 := p.Register("forward-2", "service-b", 8081)
+	err3 := p.Register("forward-3", "service-c", 8082)
+
+	assert.NoError(t, err1)
+	assert.NoError(t, err2)
+	assert.NoError(t, err3)
+	assert.Equal(t, 3, p.GetRegisteredCount())
+}
+
+func TestUnregister_Success_Integration(t *testing.T) {
+	if testing.Short() {
+		t.Skip("Skipping mDNS integration test in short mode")
+	}
+
+	p := NewPublisher(true)
+	defer p.Stop()
+
+	p.Register("forward-1", "test-service", 8080)
+	assert.Equal(t, 1, p.GetRegisteredCount())
+
+	p.Unregister("forward-1")
+	assert.Equal(t, 0, p.GetRegisteredCount())
+}

internal/ui/bubbletea_ui.go

@@ -46,6 +46,11 @@ type BubbleTeaUI struct {
 	version        string
 	errors         map[string]string // Track error messages by forward ID
 
+	// Update notification
+	updateAvailable bool
+	updateVersion   string
+	updateURL       string
+
 	// Modal wizard state
 	viewMode     ViewMode
 	addWizard    *AddWizardState
@@ -96,6 +101,16 @@ func (ui *BubbleTeaUI) SetWizardDependencies(discovery *k8s.Discovery, mutator *
 	ui.configPath = configPath
 }
 
+// SetUpdateAvailable sets the update notification to be displayed
+func (ui *BubbleTeaUI) SetUpdateAvailable(version, url string) {
+	ui.mu.Lock()
+	defer ui.mu.Unlock()
+
+	ui.updateAvailable = true
+	ui.updateVersion = version
+	ui.updateURL = url
+}
+
 // Start starts the bubbletea application
 func (ui *BubbleTeaUI) Start() error {
 	m := model{ui: ui}
@@ -169,8 +184,9 @@ func (ui *BubbleTeaUI) UpdateStatus(id string, status string) {
 	if fwd, ok := ui.forwards[id]; ok {
 		fwd.Status = status
 	}
-	// Clear error if status is not Error
-	if status != "Error" {
+	// Only clear error when forward becomes Active again
+	// This keeps error visible during Reconnecting/Starting states
+	if status == "Active" {
 		delete(ui.errors, id)
 	}
 	ui.mu.Unlock()
@@ -266,7 +282,7 @@ func (m model) Update(msg tea.Msg) (tea.Model, tea.Cmd) {
 		m.ui.addWizard = nil
 		m.ui.removeWizard = nil
 		m.ui.mu.Unlock()
-		return m, nil
+		return m, tea.ClearScreen
 	}
 
 	return m, nil
@@ -356,6 +372,15 @@ func (m model) renderMainView() string {
 	// Title with version
 	title := fmt.Sprintf("kportal v%s - Port Forwarding Status", m.ui.version)
 	b.WriteString(titleStyle.Render(title))
+
+	// Show update notification if available
+	if m.ui.updateAvailable {
+		updateStyle := lipgloss.NewStyle().
+			Foreground(lipgloss.Color("42")). // Green
+			Bold(true)
+		updateMsg := fmt.Sprintf("  Update available: v%s", m.ui.updateVersion)
+		b.WriteString(updateStyle.Render(updateMsg))
+	}
 	b.WriteString("\n\n")
 
 	// Header
@@ -574,6 +599,15 @@ func (ui *BubbleTeaUI) moveSelection(delta int) {
 	}
 }
 
+// resetDeleteConfirmation resets the delete confirmation dialog state.
+// Caller must hold ui.mu lock.
+func (ui *BubbleTeaUI) resetDeleteConfirmation() {
+	ui.deleteConfirming = false
+	ui.deleteConfirmID = ""
+	ui.deleteConfirmAlias = ""
+	ui.deleteConfirmCursor = 0
+}
+
 // renderDeleteConfirmation renders the delete confirmation dialog
 func (m model) renderDeleteConfirmation() string {
 	m.ui.mu.RLock()

internal/ui/wizard_commands.go

@@ -144,8 +144,25 @@ func validateSelectorCmd(discovery *k8s.Discovery, contextName, namespace, selec
 }
 
 // checkPortCmd checks if a local port is available
-func checkPortCmd(port int) tea.Cmd {
+func checkPortCmd(port int, configPath string) tea.Cmd {
 	return func() tea.Msg {
+		// First check if port is already in the configuration
+		cfg, err := config.LoadConfig(configPath)
+		if err == nil {
+			// Check all forwards in config for this port
+			allForwards := cfg.GetAllForwards()
+			for _, fwd := range allForwards {
+				if fwd.LocalPort == port {
+					return PortCheckedMsg{
+						port:      port,
+						available: false,
+						message:   fmt.Sprintf("✗ Port %d already assigned to %s", port, fwd.ID()),
+					}
+				}
+			}
+		}
+
+		// Then check if port is available at OS level
 		available, processInfo, err := k8s.CheckPortAvailability(port)
 
 		msg := ""

internal/ui/wizard_handlers.go

@@ -173,12 +173,8 @@ func (m model) handleDeleteConfirmation(msg tea.KeyMsg) (tea.Model, tea.Cmd) {
 	switch msg.String() {
 	case "ctrl+c", "esc":
 		// Cancel deletion
-		m.ui.deleteConfirming = false
-		m.ui.deleteConfirmID = ""
-		m.ui.deleteConfirmAlias = ""
-		m.ui.deleteConfirmCursor = 0 // Reset cursor
+		m.ui.resetDeleteConfirmation()
 		m.ui.mu.Unlock()
-		// Force a repaint by returning the model
 		return m, tea.ClearScreen
 
 	case "left", "h", "right", "l":
@@ -191,26 +187,18 @@ func (m model) handleDeleteConfirmation(msg tea.KeyMsg) (tea.Model, tea.Cmd) {
 		// Confirm deletion (either Enter on Yes or pressing 'y')
 		if m.ui.deleteConfirmCursor == 0 || msg.String() == "y" {
 			id := m.ui.deleteConfirmID
-			m.ui.deleteConfirming = false
-			m.ui.deleteConfirmID = ""
-			m.ui.deleteConfirmAlias = ""
+			m.ui.resetDeleteConfirmation()
 			m.ui.mu.Unlock()
 			return m, removeForwardByIDCmd(m.ui.mutator, id)
 		}
 		// Enter on No = cancel
-		m.ui.deleteConfirming = false
-		m.ui.deleteConfirmID = ""
-		m.ui.deleteConfirmAlias = ""
-		m.ui.deleteConfirmCursor = 0 // Reset cursor
+		m.ui.resetDeleteConfirmation()
 		m.ui.mu.Unlock()
 		return m, tea.ClearScreen
 
 	case "n":
 		// Quick 'n' for no
-		m.ui.deleteConfirming = false
-		m.ui.deleteConfirmID = ""
-		m.ui.deleteConfirmAlias = ""
-		m.ui.deleteConfirmCursor = 0 // Reset cursor
+		m.ui.resetDeleteConfirmation()
 		m.ui.mu.Unlock()
 		return m, tea.ClearScreen
 	}
@@ -259,10 +247,7 @@ func (m model) handleAddWizardKeys(msg tea.KeyMsg) (tea.Model, tea.Cmd) {
 		} else {
 			// Go back one step
 			wizard.step--
-			wizard.cursor = 0
-			wizard.clearTextInput()
-			wizard.clearSearchFilter()
-			wizard.error = nil
+			wizard.resetInput()
 
 			// Reset input mode based on the step we're going back to
 			switch wizard.step {
@@ -374,6 +359,11 @@ func (m model) handleAddWizardKeys(msg tea.KeyMsg) (tea.Model, tea.Cmd) {
 func (m model) handleAddWizardEnter() (tea.Model, tea.Cmd) {
 	wizard := m.ui.addWizard
 
+	// Don't process Enter if we're currently loading
+	if wizard.loading {
+		return m, nil
+	}
+
 	switch wizard.step {
 	case StepSelectContext:
 		filteredContexts := wizard.getFilteredContexts()
@@ -452,12 +442,14 @@ func (m model) handleAddWizardEnter() (tea.Model, tea.Cmd) {
 			filteredServices := wizard.getFilteredServices()
 			if wizard.cursor >= 0 && wizard.cursor < len(filteredServices) {
 				wizard.resourceValue = filteredServices[wizard.cursor].Name
+
+				// Get ports from selected service (must do this BEFORE clearing search filter)
+				wizard.detectedPorts = filteredServices[wizard.cursor].Ports
+
 				wizard.step = StepEnterRemotePort
 				wizard.clearTextInput()
 				wizard.clearSearchFilter()
 
-				// Get ports from selected service
-				wizard.detectedPorts = filteredServices[wizard.cursor].Ports
 				if len(wizard.detectedPorts) > 0 {
 					wizard.inputMode = InputModeList
 					wizard.cursor = 0
@@ -476,7 +468,14 @@ func (m model) handleAddWizardEnter() (tea.Model, tea.Cmd) {
 				wizard.clearTextInput()
 			} else if wizard.cursor >= 0 && wizard.cursor < len(wizard.detectedPorts) {
 				// Selected a detected port
-				wizard.remotePort = int(wizard.detectedPorts[wizard.cursor].Port)
+				// For services, use TargetPort (actual pod port) if available
+				// For pods, TargetPort is 0, so use Port (container port)
+				selectedPort := wizard.detectedPorts[wizard.cursor]
+				if selectedPort.TargetPort > 0 {
+					wizard.remotePort = int(selectedPort.TargetPort)
+				} else {
+					wizard.remotePort = int(selectedPort.Port)
+				}
 				wizard.step = StepEnterLocalPort
 				wizard.clearTextInput()
 				wizard.inputMode = InputModeText
@@ -485,7 +484,7 @@ func (m model) handleAddWizardEnter() (tea.Model, tea.Cmd) {
 		} else {
 			// Text mode - manual entry
 			port, err := strconv.Atoi(wizard.textInput)
-			if err != nil || port < 1 || port > 65535 {
+			if err != nil || !config.IsValidPort(port) {
 				wizard.error = fmt.Errorf("invalid port number")
 			} else {
 				wizard.remotePort = port
@@ -497,17 +496,14 @@ func (m model) handleAddWizardEnter() (tea.Model, tea.Cmd) {
 
 	case StepEnterLocalPort:
 		port, err := strconv.Atoi(wizard.textInput)
-		if err != nil || port < 1 || port > 65535 {
+		if err != nil || !config.IsValidPort(port) {
 			wizard.error = fmt.Errorf("invalid port number")
 		} else {
+			// Check port availability before proceeding
 			wizard.localPort = port
-			wizard.step = StepConfirmation
-			wizard.clearTextInput()
-			wizard.cursor = 0
-			wizard.inputMode = InputModeList
-			wizard.error = nil
 			wizard.loading = true
-			return m, checkPortCmd(port)
+			wizard.error = nil
+			return m, checkPortCmd(port, m.ui.configPath)
 		}
 
 	case StepConfirmation:
@@ -520,6 +516,12 @@ func (m model) handleAddWizardEnter() (tea.Model, tea.Cmd) {
 
 		// Handle button selection
 		if wizard.cursor == 0 {
+			// Check if port is available before saving
+			if !wizard.portAvailable {
+				wizard.error = fmt.Errorf("port %d is not available. Please choose a different port", wizard.localPort)
+				return m, nil
+			}
+
 			// Confirmed - save the forward
 			wizard.alias = wizard.textInput
 
@@ -549,9 +551,10 @@ func (m model) handleAddWizardEnter() (tea.Model, tea.Cmd) {
 
 			return m, saveForwardCmd(m.ui.mutator, wizard.selectedContext, wizard.selectedNamespace, fwd)
 		} else {
-			// Cancelled
+			// Cancelled - return to main view with screen clear
 			m.ui.viewMode = ViewModeMain
 			m.ui.addWizard = nil
+			return m, tea.ClearScreen
 		}
 
 	case StepSuccess:
@@ -561,9 +564,10 @@ func (m model) handleAddWizardEnter() (tea.Model, tea.Cmd) {
 			m.ui.addWizard.loading = true
 			return m, loadContextsCmd(m.ui.discovery)
 		} else {
-			// Return to main view
+			// Return to main view with screen clear
 			m.ui.viewMode = ViewModeMain
 			m.ui.addWizard = nil
+			return m, tea.ClearScreen
 		}
 	}
 
@@ -771,6 +775,17 @@ func (m model) handlePortChecked(msg PortCheckedMsg) (tea.Model, tea.Cmd) {
 		m.ui.addWizard.loading = false
 		m.ui.addWizard.portAvailable = msg.available
 		m.ui.addWizard.portCheckMsg = msg.message
+
+		// Only proceed to confirmation if port is available
+		if msg.available {
+			m.ui.addWizard.step = StepConfirmation
+			m.ui.addWizard.clearTextInput()
+			m.ui.addWizard.cursor = 0
+			m.ui.addWizard.inputMode = InputModeList
+		} else {
+			// Port is not available - show error and stay on local port step
+			m.ui.addWizard.error = fmt.Errorf("port %d is in use, please choose another port", msg.port)
+		}
 	}
 
 	return m, nil
@@ -807,5 +822,5 @@ func (m model) handleForwardsRemoved(msg ForwardsRemovedMsg) (tea.Model, tea.Cmd
 	// If there was an error, it will be logged but we don't show it in UI for now
 	// The config watcher will either reload (success) or keep old config (failure)
 
-	return m, nil
+	return m, tea.ClearScreen
 }

internal/ui/wizard_state.go

@@ -363,3 +363,13 @@ func (w *AddWizardState) clearSearchFilter() {
 	w.cursor = 0
 	w.scrollOffset = 0
 }
+
+// resetInput clears text input, search filter, and error state.
+// Use this when navigating between wizard steps.
+func (w *AddWizardState) resetInput() {
+	w.textInput = ""
+	w.searchFilter = ""
+	w.cursor = 0
+	w.scrollOffset = 0
+	w.error = nil
+}

internal/ui/wizard_views.go

@@ -349,9 +349,20 @@ func (m model) renderEnterRemotePort() string {
 		// Render detected ports within viewport
 		for i := start; i < end && i < len(wizard.detectedPorts); i++ {
 			port := wizard.detectedPorts[i]
-			portDesc := fmt.Sprintf("%d", port.Port)
-			if port.Name != "" {
-				portDesc += fmt.Sprintf(" (%s)", port.Name)
+			// For services, show both service port and target port if they differ
+			var portDesc string
+			if port.TargetPort > 0 && port.TargetPort != port.Port {
+				// Service with different target port: "80 → 8000 (http)"
+				portDesc = fmt.Sprintf("%d → %d", port.Port, port.TargetPort)
+				if port.Name != "" {
+					portDesc += fmt.Sprintf(" (%s)", port.Name)
+				}
+			} else {
+				// Pod port or service with same port
+				portDesc = fmt.Sprintf("%d", port.Port)
+				if port.Name != "" {
+					portDesc += fmt.Sprintf(" (%s)", port.Name)
+				}
 			}
 
 			prefix := "  "
@@ -373,7 +384,7 @@ func (m model) renderEnterRemotePort() string {
 				prefix = "▸ "
 				b.WriteString(selectedStyle.Render(prefix + manualOption))
 			} else {
-				b.WriteString(mutedStyle.Render(prefix + manualOption))
+				b.WriteString(prefix + mutedStyle.Render(manualOption))
 			}
 			b.WriteString("\n")
 		}
@@ -390,9 +401,17 @@ func (m model) renderEnterRemotePort() string {
 		if len(wizard.detectedPorts) > 0 {
 			b.WriteString(mutedStyle.Render("Detected ports:\n"))
 			for _, port := range wizard.detectedPorts {
-				portDesc := fmt.Sprintf("%d", port.Port)
-				if port.Name != "" {
-					portDesc += fmt.Sprintf(" (%s)", port.Name)
+				var portDesc string
+				if port.TargetPort > 0 && port.TargetPort != port.Port {
+					portDesc = fmt.Sprintf("%d → %d", port.Port, port.TargetPort)
+					if port.Name != "" {
+						portDesc += fmt.Sprintf(" (%s)", port.Name)
+					}
+				} else {
+					portDesc = fmt.Sprintf("%d", port.Port)
+					if port.Name != "" {
+						portDesc += fmt.Sprintf(" (%s)", port.Name)
+					}
 				}
 				b.WriteString(mutedStyle.Render(fmt.Sprintf("  • %s\n", portDesc)))
 			}
@@ -443,7 +462,7 @@ func (m model) renderEnterLocalPort() string {
 		} else {
 			b.WriteString(errorStyle.Render(wizard.portCheckMsg))
 		}
-	} else if wizard.textInput != "" && wizard.localPort > 0 {
+	} else if wizard.textInput != "" {
 		b.WriteString(mutedStyle.Render("Press Enter to check availability"))
 	}
 

internal/version/checker.go

@@ -0,0 +1,158 @@
+package version
+
+import (
+	"context"
+	"encoding/json"
+	"fmt"
+	"net/http"
+	"strings"
+	"time"
+)
+
+const (
+	// GitHubAPIURL is the GitHub API endpoint for releases
+	githubReleasesURL = "https://api.github.com/repos/%s/%s/releases/latest"
+	// requestTimeout is the timeout for HTTP requests
+	requestTimeout = 5 * time.Second
+)
+
+// ReleaseInfo contains information about a GitHub release
+type ReleaseInfo struct {
+	TagName string `json:"tag_name"`
+	HTMLURL string `json:"html_url"`
+	Name    string `json:"name"`
+}
+
+// UpdateInfo contains information about an available update
+type UpdateInfo struct {
+	CurrentVersion string
+	LatestVersion  string
+	ReleaseURL     string
+	ReleaseName    string
+}
+
+// Checker checks for new versions on GitHub
+type Checker struct {
+	owner   string
+	repo    string
+	current string
+	client  *http.Client
+}
+
+// NewChecker creates a new version checker
+func NewChecker(owner, repo, currentVersion string) *Checker {
+	return &Checker{
+		owner:   owner,
+		repo:    repo,
+		current: normalizeVersion(currentVersion),
+		client: &http.Client{
+			Timeout: requestTimeout,
+		},
+	}
+}
+
+// CheckForUpdate checks if a newer version is available.
+// Returns nil if current version is up to date or if check fails.
+// This is designed to fail silently - network errors should not impact the user.
+func (c *Checker) CheckForUpdate(ctx context.Context) *UpdateInfo {
+	release, err := c.fetchLatestRelease(ctx)
+	if err != nil {
+		return nil
+	}
+
+	latestVersion := normalizeVersion(release.TagName)
+	if isNewerVersion(latestVersion, c.current) {
+		return &UpdateInfo{
+			CurrentVersion: c.current,
+			LatestVersion:  latestVersion,
+			ReleaseURL:     release.HTMLURL,
+			ReleaseName:    release.Name,
+		}
+	}
+
+	return nil
+}
+
+// fetchLatestRelease fetches the latest release info from GitHub API
+func (c *Checker) fetchLatestRelease(ctx context.Context) (*ReleaseInfo, error) {
+	url := fmt.Sprintf(githubReleasesURL, c.owner, c.repo)
+
+	req, err := http.NewRequestWithContext(ctx, http.MethodGet, url, nil)
+	if err != nil {
+		return nil, err
+	}
+
+	req.Header.Set("Accept", "application/vnd.github.v3+json")
+	req.Header.Set("User-Agent", "kportal-version-checker")
+
+	resp, err := c.client.Do(req)
+	if err != nil {
+		return nil, err
+	}
+	defer resp.Body.Close()
+
+	if resp.StatusCode != http.StatusOK {
+		return nil, fmt.Errorf("GitHub API returned status %d", resp.StatusCode)
+	}
+
+	var release ReleaseInfo
+	if err := json.NewDecoder(resp.Body).Decode(&release); err != nil {
+		return nil, err
+	}
+
+	return &release, nil
+}
+
+// normalizeVersion removes 'v' or 'V' prefix and trims whitespace
+func normalizeVersion(v string) string {
+	v = strings.TrimSpace(v)
+	v = strings.TrimPrefix(v, "v")
+	v = strings.TrimPrefix(v, "V")
+	return v
+}
+
+// isNewerVersion compares two semver-like versions.
+// Returns true if latest is newer than current.
+func isNewerVersion(latest, current string) bool {
+	latestParts := parseVersion(latest)
+	currentParts := parseVersion(current)
+
+	// Compare each part
+	for i := 0; i < len(latestParts) && i < len(currentParts); i++ {
+		if latestParts[i] > currentParts[i] {
+			return true
+		}
+		if latestParts[i] < currentParts[i] {
+			return false
+		}
+	}
+
+	// If all compared parts are equal, longer version is newer
+	// e.g., 1.0.1 > 1.0
+	return len(latestParts) > len(currentParts)
+}
+
+// parseVersion splits a version string into numeric parts
+func parseVersion(v string) []int {
+	// Remove any suffix like -beta, -rc1, etc.
+	if idx := strings.IndexAny(v, "-+"); idx != -1 {
+		v = v[:idx]
+	}
+
+	parts := strings.Split(v, ".")
+	result := make([]int, 0, len(parts))
+
+	for _, p := range parts {
+		var num int
+		fmt.Sscanf(p, "%d", &num)
+		result = append(result, num)
+	}
+
+	return result
+}
+
+// FormatUpdateMessage formats a user-friendly update notification
+func (u *UpdateInfo) FormatUpdateMessage() string {
+	return fmt.Sprintf("New version available: %s (current: %s) - %s",
+		u.LatestVersion, u.CurrentVersion, u.ReleaseURL)
+}

internal/version/checker_test.go

@@ -0,0 +1,90 @@
+package version
+
+import (
+	"testing"
+
+	"github.com/stretchr/testify/assert"
+)
+
+func TestNormalizeVersion(t *testing.T) {
+	tests := []struct {
+		input    string
+		expected string
+	}{
+		{"v1.0.0", "1.0.0"},
+		{"1.0.0", "1.0.0"},
+		{"  v2.1.3  ", "2.1.3"},
+		{"V1.0.0", "1.0.0"},
+	}
+
+	for _, tt := range tests {
+		t.Run(tt.input, func(t *testing.T) {
+			result := normalizeVersion(tt.input)
+			assert.Equal(t, tt.expected, result)
+		})
+	}
+}
+
+func TestParseVersion(t *testing.T) {
+	tests := []struct {
+		input    string
+		expected []int
+	}{
+		{"1.0.0", []int{1, 0, 0}},
+		{"2.1.3", []int{2, 1, 3}},
+		{"1.0", []int{1, 0}},
+		{"10.20.30", []int{10, 20, 30}},
+		{"1.0.0-beta", []int{1, 0, 0}},
+		{"1.0.0-rc1", []int{1, 0, 0}},
+		{"1.0.0+build123", []int{1, 0, 0}},
+	}
+
+	for _, tt := range tests {
+		t.Run(tt.input, func(t *testing.T) {
+			result := parseVersion(tt.input)
+			assert.Equal(t, tt.expected, result)
+		})
+	}
+}
+
+func TestIsNewerVersion(t *testing.T) {
+	tests := []struct {
+		name     string
+		latest   string
+		current  string
+		expected bool
+	}{
+		{"major version bump", "2.0.0", "1.0.0", true},
+		{"minor version bump", "1.1.0", "1.0.0", true},
+		{"patch version bump", "1.0.1", "1.0.0", true},
+		{"same version", "1.0.0", "1.0.0", false},
+		{"current is newer major", "1.0.0", "2.0.0", false},
+		{"current is newer minor", "1.0.0", "1.1.0", false},
+		{"current is newer patch", "1.0.0", "1.0.1", false},
+		{"multi-digit versions", "1.10.0", "1.9.0", true},
+		{"longer version is newer", "1.0.1", "1.0", true},
+		{"shorter version is older", "1.0", "1.0.1", false},
+		{"complex comparison", "2.1.3", "2.1.2", true},
+		{"real world example", "0.2.0", "0.1.0", true},
+	}
+
+	for _, tt := range tests {
+		t.Run(tt.name, func(t *testing.T) {
+			result := isNewerVersion(tt.latest, tt.current)
+			assert.Equal(t, tt.expected, result)
+		})
+	}
+}
+
+func TestUpdateInfo_FormatUpdateMessage(t *testing.T) {
+	info := &UpdateInfo{
+		CurrentVersion: "0.1.0",
+		LatestVersion:  "0.2.0",
+		ReleaseURL:     "https://github.com/nvm/kportal/releases/tag/v0.2.0",
+	}
+
+	msg := info.FormatUpdateMessage()
+	assert.Contains(t, msg, "0.2.0")
+	assert.Contains(t, msg, "0.1.0")
+	assert.Contains(t, msg, "https://github.com/nvm/kportal/releases/tag/v0.2.0")
+}