* fix(refresh): wire RefreshCoordinator into the live refresh path
The RefreshCoordinator existed but was never instantiated. The actual
refresh path used only session.refreshMutex, which is per-SessionData
instance - and SessionData is pulled from a sync.Pool per request -
so concurrent requests sharing a refresh token had ZERO coordination.
Symptom: when access_token expired (e.g. 5min Zitadel default), every
in-flight request from a polling client (Grafana panels) entered the
refresh path simultaneously and POSTed the same refresh_token to the
IdP. With refresh-token rotation enabled (Zitadel/Authentik default),
only one grant succeeded; the rest got invalid_grant and each cleared
the entire session. Subsequent requests then thrashed in re-auth loops.
This commit:
- adds refreshCoordinator field on TraefikOidc
- instantiates it in NewWithContext with DefaultRefreshCoordinatorConfig
- shuts it down in Close() under shutdownOnce
- routes refreshToken() through the coordinator via coordinatedTokenRefresh,
which collapses concurrent grants to a single upstream call per
refresh_token hash
- exports refreshCoordinatorSessionID for both internal hashing and the
middleware-level wireup so dedup keys stay aligned
Behavioural notes:
- nil-coordinator fallback preserves existing tests that build TraefikOidc
literals without going through the constructor
- followers receive the same TokenResponse/error as the leader, so no
per-instance code paths change
- existing TestGetNewTokenWithRefreshToken_Concurrency still passes
because it hits GetNewTokenWithRefreshToken directly, below the
coordinator boundary
Tests:
- refresh_coordinator_wireup_test.go: 50 concurrent refreshes coalesce
to <=2 upstream calls; distinct tokens still run in parallel; nil
coordinator falls back cleanly
* perf(cache): bound L1 backfill goroutines in HybridBackend
Get() and GetMany() previously spawned a goroutine per L2 hit to write
the value through to L1. Under sustained polling traffic (e.g. a Grafana
dashboard refreshing every 30s with N panels) this minted thousands of
goroutines, each running in Yaegi - directly contributing to the
~1000% CPU spike that pairs with the refresh-token herd.
Replace the per-hit goroutines with a single l1BackfillWorker fed by
l1BackfillBuffer, mirroring the existing asyncWriteBuffer/asyncWriteWorker
pattern for L2 writes. Buffer overflow drops the backfill (counted via
l1BackfillDrops) - a dropped backfill just means the next L2 hit for
that key re-queues it, which is safe.
Tests:
- TestHybridBackend_L1BackfillBounded: 1000 distinct L2 hits keep
goroutine count within +20 of baseline (pre-fix it grew by ~1000)
- TestHybridBackend_L1BackfillFullDrops: drops are accounted for when
the buffer is saturated and the worker is stopped
* feat(refresh): implement isRefreshTokenExpired heuristic
Replace the placeholder `return false` with a real check based on the
issued_at timestamp that SetRefreshToken already stamps into the session.
Gated by a new MaxRefreshTokenAgeSeconds config field (default 21600 =
6h, matching the existing comment). 0 disables the check.
This wires the previously-dead refreshTokenExpired branch in middleware.go,
which short-circuits AJAX requests with a 401 instead of letting them
hammer the IdP for a refresh token that's almost certainly stale - the
classic Grafana-after-long-pause failure mode.
Behaviour:
- maxRefreshTokenAge=0 disables the check (preserves prior behaviour)
- legacy sessions without issued_at still attempt one refresh; the IdP
remains the source of truth on first try
- nil-receiver and nil-session guards keep test code that builds
TraefikOidc literals safe
Tests:
- TestIsRefreshTokenExpired_DisabledWhenAgeZero
- TestIsRefreshTokenExpired_LegacySessionWithoutTimestamp
- TestIsRefreshTokenExpired_WithinWindow
- TestIsRefreshTokenExpired_BeyondWindow
- TestIsRefreshTokenExpired_NilGuards
* perf(token): skip parseJWT on cache hit in VerifyToken
The token cache fast-return existed but ran AFTER parseJWT, so every
validation paid for base64 + JSON unmarshal even on a hit. Under bursty
traffic (e.g. 10+ concurrent panel requests on every Grafana dashboard
refresh, each calling validateStandardTokens which verifies BOTH the
access token and the ID token), this is two redundant parses per
request multiplied by the panel count.
Move the cache lookup ahead of parseJWT. On a hit the function returns
nil immediately. On a miss the original flow runs unchanged.
Also nil-guard t.tokenCache to keep partial-literal test instances safe
(matches the same pattern we already use for tokenBlacklist).
Tests:
- TestVerifyToken_CacheHitSkipsParse: cache pre-populated with claims
for a token whose body would fail parseJWT - returns nil iff the
fast-path bypasses the parse
- TestVerifyToken_CacheMissStillParses: a syntactically valid but
unsigned token still errors past parseJWT on cache miss
* feat(refresh): cross-replica refresh-grant dedup via shared cache
The in-process RefreshCoordinator added in 9f96d8c already collapses
concurrent refresh-token grants on a single Traefik replica. With the
plugin's existing Redis (Dragonfly) cache infrastructure available, we
can extend that dedup across replicas: if pod A refreshes a token at
T+0 and pod B receives a request for the same session at T+1, pod B
should reuse pod A's result rather than POSTing the now-rotated refresh
token to the IdP.
Implementation:
- Add a refreshResultCache to UniversalCacheManager (memory-only when
Redis is disabled, Redis-backed in production via the existing
hybrid/Redis-only mode selection)
- Expose it through CacheManager.GetSharedRefreshResultCache and on the
TraefikOidc struct as refreshResultCache (CacheInterface)
- Inside the closure passed to RefreshCoordinator.CoordinateRefresh,
consult the cache first; on hit return immediately, on miss exchange
with the IdP and populate the cache for peers
- 5s TTL: long enough for siblings to observe, short enough that a
rotated refresh token cannot be re-supplied after the IdP has moved on
- Errors are intentionally NOT cached - peers must always be able to
retry on their own
Pragmatic choice: optimistic cache rather than a hard distributed lock.
- A hard lock (SET NX + poll) doubles Redis RTT and risks dead-locks
if a Traefik pod dies mid-grant.
- The user's BGP+Local externalTrafficPolicy already pins ingress for
a session to one node in steady state, so cross-pod racing is rare.
- This optimistic path catches the rare failover case without adding
failure modes.
Tests:
- TestCoordinatedTokenRefresh_CrossReplicaCacheHit: pre-populated cache
short-circuits the upstream call entirely (0 IdP calls)
- TestCoordinatedTokenRefresh_PopulatesCrossReplicaCache: leader stores
a successful result for peers to find
- TestCoordinatedTokenRefresh_ErrorIsNotCached: invalid_grant must not
poison the dedup cache - peers must retry independently
Traefik OIDC Middleware
OpenID Connect authentication middleware for Traefik. Replaces forward-auth + oauth2-proxy. Auto-detects all major OIDC providers, validates ID tokens, manages sessions, and forwards user identity to downstream services.
Documentation
- Configuration reference — every parameter
- Provider guide — Google, Azure, Auth0, Okta, Keycloak, Cognito, GitLab, GitHub, generic
- Auth0 audience guide — custom APIs, opaque tokens, token confusion
- Redis cache — multi-replica deployments
- Dynamic Client Registration — RFC 7591
- Development · Testing
Provider support
| Provider | OIDC | Refresh | Auto-detected by |
|---|---|---|---|
| Full | Yes | accounts.google.com |
|
| Azure AD | Full | Yes | login.microsoftonline.com, sts.windows.net |
| Auth0 | Full | Yes | *.auth0.com |
| Okta | Full | Yes | *.okta.com, *.oktapreview.com, *.okta-emea.com |
| Keycloak | Full | Yes | host containing keycloak, or /realms/ in path (covers KC <17 /auth/realms/ and 17+ /realms/) |
| AWS Cognito | Full | Yes | cognito-idp.*.amazonaws.com |
| GitLab | Full | Yes | gitlab.com |
| GitHub | OAuth 2.0 only — no ID token, no refresh | No | github.com |
| Generic | Full | Yes | any RFC-compliant .well-known/openid-configuration |
Authentication and claim extraction use the ID token. Ensure your provider includes required claims (email, roles, groups) in the ID token, not just the access token or UserInfo endpoint.
Install
Enable the plugin in Traefik's static configuration:
# traefik.yml
experimental:
plugins:
traefikoidc:
moduleName: github.com/lukaszraczylo/traefikoidc
version: v0.7.10
Then attach the middleware in your dynamic configuration (see Quickstart below).
This middleware tracks the current Traefik helm chart release. If it fails to load, update Traefik first.
Verify release signatures
Release checksums are signed with cosign keyless signing:
cosign verify-blob \
--certificate-identity-regexp "https://github.com/lukaszraczylo/traefikoidc/.*" \
--certificate-oidc-issuer "https://token.actions.githubusercontent.com" \
--bundle "traefikoidc_v<version>_checksums.txt.sigstore.json" \
traefikoidc_v<version>_checksums.txt
Quickstart
apiVersion: traefik.io/v1alpha1
kind: Middleware
metadata:
name: oidc-auth
namespace: traefik
spec:
plugin:
traefikoidc:
providerURL: https://accounts.google.com
clientID: 1234567890.apps.googleusercontent.com
clientSecret: urn:k8s:secret:traefik-oidc:CLIENT_SECRET
sessionEncryptionKey: urn:k8s:secret:traefik-oidc:SESSION_KEY
callbackURL: /oauth2/callback
logoutURL: /oauth2/logout
postLogoutRedirectURI: /
# forceHTTPS defaults to true (secure-by-default). Only set false if you
# serve OIDC over plaintext HTTP for local dev.
allowedUserDomains: [company.com]
allowedRolesAndGroups: [admin, developer]
excludedURLs: [/health, /metrics]
More example configs in examples/.
Required parameters
| Parameter | Description |
|---|---|
providerURL |
Issuer URL (used for OIDC discovery). |
clientID |
OAuth 2.0 client ID. |
clientSecret |
OAuth 2.0 client secret. Supports urn:k8s:secret:ns:name:key. |
sessionEncryptionKey |
Cookie encryption key, min 32 bytes. |
callbackURL |
Callback path, e.g. /oauth2/callback. |
Common optional parameters
Full reference in docs/CONFIGURATION.md.
| Parameter | Default | Purpose |
|---|---|---|
forceHTTPS |
true |
Forces https:// in redirect URIs. Leave at default behind any TLS-terminating LB (AWS ALB, GCP LB, Azure App Gateway). Set false only for plaintext HTTP local dev. |
logoutURL |
callbackURL + "/logout" |
RP-initiated logout path. |
postLogoutRedirectURI |
/ |
Where to send users after logout. |
scopes |
appended to openid profile email |
Extra OAuth scopes. Set overrideScopes: true to replace defaults. |
excludedURLs |
none | Prefix-matched paths that bypass auth. |
allowedUserDomains |
none | Restrict to email domains. |
allowedUsers |
none | Restrict to specific addresses (or claim values when userIdentifierClaim != email). |
allowedRolesAndGroups |
none | Require any of these roles/groups from ID-token claims. |
roleClaimName / groupClaimName |
roles / groups |
For namespaced claims (Auth0). |
userIdentifierClaim |
email |
Use sub, oid, upn, or preferred_username for users without email. |
enablePKCE |
false |
PKCE on the auth code flow. |
cookieDomain |
auto | Set explicitly for multi-subdomain setups (.example.com). |
cookiePrefix |
_oidc_raczylo_ |
Unique prefix per middleware instance to isolate sessions. |
sessionMaxAge |
86400 |
Session lifetime in seconds. |
refreshGracePeriodSeconds |
60 |
Proactively refresh tokens this many seconds before expiry. |
rateLimit |
100 |
Requests/sec. Min 10. |
logLevel |
info |
debug, info, error. |
audience |
clientID |
Custom access-token audience (Auth0 custom APIs). |
strictAudienceValidation |
false |
Reject mismatched audiences. Set true in production. |
allowOpaqueTokens / requireTokenIntrospection |
false |
Accept opaque access tokens via RFC 7662. |
disableReplayDetection |
false |
Disable JTI cache. Use Redis instead for multi-replica. |
allowPrivateIPAddresses |
false |
Permit private-IP providerURL (internal Keycloak, etc.). |
minimalHeaders |
false |
Reduce forwarded headers (mitigates HTTP 431). |
stripAuthCookies |
false |
Strip OIDC cookies from backend hop (mitigates HTTP 431). |
caCertPath / caCertPEM |
none | Trust an internal CA for the provider's TLS. |
insecureSkipVerify |
false |
Local dev only. Disables TLS verification, logs a security warning. |
enableBackchannelLogout / backchannelLogoutURL |
false / none |
OIDC Back-Channel Logout (server-to-server). |
enableFrontchannelLogout / frontchannelLogoutURL |
false / none |
OIDC Front-Channel Logout (iframe). |
redis |
disabled | See docs/REDIS.md. |
dynamicClientRegistration |
disabled | See docs/DCR.md. |
Production gotchas
TLS termination at a load balancer
forceHTTPS defaults to true, so redirect URIs always use https://. This is
the right default behind AWS ALB, GCP LB, Azure App Gateway, or any LB that
terminates TLS — X-Forwarded-Proto is unreliable (ALB may overwrite it).
Only set forceHTTPS: false when you actually serve OIDC over plaintext HTTP
(local dev). See issue #82.
Multi-replica deployments
Each replica keeps its own in-memory JTI cache → false positive "token replay detected" when the same token hits different replicas. Two options:
- Set
disableReplayDetection: true(loses replay protection). - Enable Redis for shared state (recommended) — see docs/REDIS.md.
For IdP-initiated logout (back/front-channel) in multi-replica setups, Redis is required so a logout on one instance invalidates sessions on the others.
Multiple middleware instances on the same host
Each instance must use a unique cookiePrefix and sessionEncryptionKey,
otherwise a session minted by one instance can grant access through another.
See issue #87.
HTTP 431 from backends
Either the ID token or the chunked OIDC cookies overflow your backend's header buffer. Combine these as needed:
minimalHeaders: true # drop X-Auth-Request-Token et al.
stripAuthCookies: true # strip _oidc_raczylo_* cookies on the backend hop
Cookies remain in the browser; only the Traefik→backend hop is affected. See #64, #122.
Internal CA for the provider
If the provider's TLS cert is signed by a private CA (self-hosted GitLab, internal Keycloak, ADFS):
caCertPath: /etc/ssl/certs/internal-ca.pem
# or, inline:
caCertPEM: |
-----BEGIN CERTIFICATE-----
...
-----END CERTIFICATE-----
Both can be combined. An unparseable bundle fails the plugin at startup. See #125.
Environment variable names containing API
Traefik reserves TRAEFIK_API_*. User vars whose name contains API (e.g.
OIDC_ENCRYPTION_SECRET_API) make the plugin fail with
invalid handler type: <nil>. Rename to anything without the literal API
substring. See #98.
Templated headers
Forward identity to backends via Go templates over ID-token claims and tokens:
headers:
- name: X-User-Email
value: "{{{{.Claims.email}}}}"
- name: Authorization
value: "Bearer {{{{.AccessToken}}}}"
- name: X-User-Roles
value: "{{{{range $i, $e := .Claims.roles}}}}{{{{if $i}}}},{{{{end}}}}{{{{$e}}}}{{{{end}}}}"
Available bindings: .Claims.<field>, .AccessToken, .IdToken,
.RefreshToken. Names are case-sensitive (.Claims, not .claims).
Escape with quadruple braces. If you see
can't evaluate field AccessToken in type bool, Traefik's YAML parser ate your{{ }}. The fix that actually works is{{{{ }}}}— the YAML pass turns it into{{ }}for the Go template engine. Other escaping tricks (literal blocks, single quotes) do not work reliably.
Default downstream headers
When a request is authenticated, the middleware sets:
| Header | Notes |
|---|---|
X-Forwarded-User |
User's email (always). |
X-User-Groups |
Comma-separated. |
X-User-Roles |
Comma-separated. |
X-Auth-Request-User |
User's email. |
X-Auth-Request-Redirect |
Original request URI. |
X-Auth-Request-Token |
Full ID token — the largest header; suppressed by minimalHeaders. |
Plus security headers (CSP, HSTS, X-Frame-Options, X-Content-Type-Options,
X-XSS-Protection, Referrer-Policy) controlled by the securityHeaders
section — see docs/CONFIGURATION.md.
Common errors
| Symptom | Cause |
|---|---|
Token verification failed |
Wrong/unreachable providerURL, or clock skew. |
Session encryption key too short |
sessionEncryptionKey is < 32 bytes. |
No matching public key found |
JWKS endpoint down, or kid mismatch. |
Access denied: Your email domain is not allowed |
User's domain not in allowedUserDomains. |
Access denied: You do not have any of the allowed roles or groups |
Claims missing or not in allowedRolesAndGroups. |
can't evaluate field AccessToken in type bool |
Template not escaped — use {{{{ }}}}. |
tls: failed to verify certificate: x509: certificate signed by unknown authority |
Internal CA — set caCertPath / caCertPEM. |
invalid handler type: <nil> |
Env var name contains API — rename it. |
false positive replay detected |
Multi-replica without Redis — see Multi-replica deployments. |
| Google sessions expire after ~1h | Consent screen still in "Testing" mode. Do not add offline_access — Google rejects it; the middleware sets access_type=offline automatically. |
Provider-specific issues (Keycloak mappers, Azure AD group overage, Auth0 namespaced claims, Cognito regions, GitLab self-hosted) live in docs/PROVIDERS.md.
Set logLevel: debug to surface detail.
License
See LICENSE.