# Tekton v1.11.0

2026-03-30 · [Original release notes](https://github.com/tektoncd/pipeline/releases/tag/v1.11.0) · [Full summary](https://ratatosk.io/en/releases/tekton/v1.11.0)

Security 2 · Breaking changes 3 · Important fixes 11 · Enhancements 4 · Dependency updates 45 · Other changes 27

## Security (2)

### Cluster resolver trims namespace-list entries before access-control matching

- Affects: Cluster resolver configurations using comma-separated `allowed-namespaces` or `blocked-namespaces` values with whitespace around entries

Previously, entries such as `prod, staging` left a leading space on `staging`, so the access-control match could fail silently. Entries are now trimmed before comparison, so surrounding whitespace no longer prevents a match.

> * :bug: fix: cluster resolver namespace access control whitespace and wildcard bugs (#9592)
> ix cluster resolver namespace access control: trim whitespace in allowed/blocked namespace lists, fix wildcard (*) handling when combined with explicit entries, and reject empty default-namespace values.

### Cluster resolver wildcard blocking fixed

- Affects: cluster resolver configurations with `*` and explicit entries in `blocked-namespaces`

Previously, combined forms such as `"*,kube-system"` silently failed to block namespaces as intended.

> * :bug: fix: cluster resolver namespace access control whitespace and wildcard bugs (#9592)
> ix cluster resolver namespace access control: trim whitespace in allowed/blocked namespace lists, fix wildcard (*) handling when combined with explicit entries, and reject empty default-namespace values.

## Breaking changes (3)

### Cluster resolver rejects empty `default-namespace` values

- Affects: cluster resolver configurations with an empty `default-namespace` value

The shipped `default-namespace: ""` value previously passed through namespace checks without reporting a missing parameter.

Evidence: “`default-namespace: ""` (the shipped default) passes an empty string through all namespace checks instead of reporting a missing parameter” https://github.com/tektoncd/pipeline/pull/9592

> * :bug: fix: cluster resolver namespace access control whitespace and wildcard bugs (#9592)
> ix cluster resolver namespace access control: trim whitespace in allowed/blocked namespace lists, fix wildcard (*) handling when combined with explicit entries, and reject empty default-namespace values.

### TaskRun pod latency metric is now a Histogram without the `pod` label

- Affects: setups with dashboards or alerts referencing `tekton_pipelines_controller_taskruns_pod_latency_milliseconds`
- Action required: Dashboards or alerts referencing this metric need `histogram_quantile()` instead of direct-value queries.

On large clusters, the former per-TaskRun `pod` series could exceed 1.2M active time series, causing Prometheus scrape timeouts and loss of Tekton metrics. The replacement exposes histogram bucket, sum, and count series.

> * :bug: fix: convert pod latency metric to histogram and remove pod label (#9530)
> ction required: The `tekton_pipelines_controller_taskruns_pod_latency_milliseconds` metric has been converted from a Gauge to a Histogram and the `pod` label has been removed. Dashboards or alerts referencing this metric will need to be updated to use `histogram_quantile()` instead of direct value queries.

### Default Jaeger tracing endpoint changed

- Affects: tracing configured to use the default endpoint in `config-tracing.yaml`

Previously, the default endpoint was `http://jaeger-collector.jaeger.svc.cluster.local:14268/api/traces`; after the upgrade, it is `http://jaeger-collector.jaeger.svc.cluster.local:4318/v1/traces`. The note does not specify whether explicitly configured endpoints change.

Evidence: “-* endpoint: API endpoint for jaeger collector to send the traces. By default the endpoint is configured to be `http://jaeger-collector.jaeger.svc.cluster.local:14268/api/traces`. +* endpoint: API endpoint for jaeger collector to send the traces. By default the endpoint is configured to be `http://jaeger-collector.jaeger.svc.cluster.local:4318/v1/traces`.” https://github.com/tektoncd/pipeline/pull/9141

> * :bug: fix: update default tracing endpoint to http protobuf endpoint (#9141)

## Important fixes (11)

### [Silent failure] Explicit error replaces `GetNameAndNamespace` namespace fallback

- Affects: resolution requests without a namespace on the request or owning object (Inferred)

When neither the request nor its owning object provided a namespace, the method silently used `default`, potentially creating a `ResolutionRequest` in the wrong namespace with an OwnerReference that Kubernetes garbage collection ignores. It now returns an explicit error instead.

> * :bug: fix: replace silent default namespace fallback with explicit error in GetNameAndNamespace (#9594)
> eplace silent "default" namespace fallback in GetNameAndNamespace with an explicit error, preventing potential ResolutionRequest creation in wrong namespace.

### [Silent failure] Resolution framework request-name lookup fixed

- Affects: resolution-framework code that injects and reads request namespace and name context values

When the namespace was injected before the request name, the colliding context keys caused name injection to fail silently and `RequestName()` to return the namespace string.

> * :bug: fix: resolve context key collision and ownerRef nil panic in resolution framework (#9593)
> ix context key collision in resolution framework where RequestName() silently returned the namespace value, and fix nil pointer panic in ownerRefsAreEqual when both Controller fields are nil.

### Overcounting fixed in the `running_taskruns` metric

- Affects: setups using the `running_taskruns` metric

Before their first reconcile, newly created `TaskRun`s with no condition set could be counted as running.

> * :bug: Fix running_taskruns metric overcounting TaskRuns with no condition (#9485)
> Fixed overcounting in the `running_taskruns` metric for `TaskRun`s with no condition set yet.

### `PipelineRun` duration and count metrics now recorded on cancellation

- Affects: setups that consume duration and count metrics for cancelled `PipelineRuns` (Inferred)

When cancellation took the `cancelPipelineRun()` short-circuit without calling `reconcile()`, duration and count observations were omitted without an error; this path now records them.

> * :bug: fix: record metrics for cancelled PipelineRuns (#9658)

### Repository-specific SSH key selection fixed

- Affects: multiple SSH auth secrets for different repositories on one Git server, with repository paths in the secret annotation URLs

Previously, SSH used the first key for every repository, causing deploy-key authentication failures. It now selects the corresponding key for each repository.

> * :bug: Fix: Add SSH Host aliases to support multiple SSH credentials on same host (#9643)
> Fixed SSH credential matching to support multiple repositories on the same host with different SSH keys. Previously, when using multiple SSH auth secrets for different repositories on the same Git server (e.g., github.com/org/repo1 and github.com/org/repo2), SSH would use the first key for all repositories, causing authentication failures with deploy keys. SSH Host aliases and Git `url.*.insteadOf` rewriting now enable per-repository SSH key selection when the secret annotation URL includes a repo path.

### `step-init` repeat-run symlink failure fixed

- Affects: pods where `step-init` symlinks persist on a shared volume

When a container restarted within a pod while symlinks from its previous run persisted on the shared volume (for example, after OOM or eviction), `step-init` failed with `symlink: file exists`. It now tolerates those existing links when run again.

> * :bug: fix: make step-init symlink creation idempotent (#9600)
> ix entrypoint step-init to handle container restarts gracefully. Previously, if a container restarted within a pod (e.g. due to OOM or eviction), the init process would fail with "symlink: file exists" because symlinks from the previous run persisted on the shared volume.

### Nil-pointer panic fixed in resolution-framework owner-reference comparisons

- Affects: Resolution-framework owner-reference comparisons

When both `a.Controller` and `b.Controller` were nil, `ownerRefsAreEqual` dereferenced them and panicked the controller. It now checks the both-nil case before dereferencing.

> * :bug: fix: resolve context key collision and ownerRef nil panic in resolution framework (#9593)
> ix context key collision in resolution framework where RequestName() silently returned the namespace value, and fix nil pointer panic in ownerRefsAreEqual when both Controller fields are nil.

### Tekton credential volume name collisions fixed

- Affects: Tekton credential injection from annotated secrets

In namespaces with many (118+) annotated secrets, names could collide, particularly for secrets with similar prefixes. Volume names now use deterministic SHA-256 hashing of the full secret name instead of truncation with a random suffix.

> * :bug: fix: use hashed volume names to prevent credential volume name collisions (#9528)
> ix credential volume name collisions when namespaces have many (118+)
> annotated secrets. Volume names now use deterministic SHA-256 hashing
> instead of truncation with random suffix.

### Longer `PipelineRun` timeouts now reach child TaskRuns

- Affects: `PipelineRun` task or finally timeouts above the global default, with child TaskRuns that have no explicit per-task timeout

Previously, the webhook applied the global default (typically 1h) to those child TaskRuns, so they could be canceled before the longer `PipelineRun` timeout elapsed. They now receive the longer `PipelineRun` timeout; an explicit per-task timeout still takes precedence.

> * :bug: fix: propagate PipelineRun tasks/finally timeout to child TaskRuns (#9419)
> When `spec.timeouts.tasks` or `spec.timeouts.finally` on a PipelineRun exceeds the global default timeout, the value is now propagated to individual child TaskRuns that do not have an explicit per-task timeout. This prevents TaskRuns from being prematurely canceled at the global default (e.g., 1h) when the PipelineRun allows a longer duration.

### Git credential matching by repository path fixed

- Affects: multiple Git credential secrets for repositories on the same Git server, with repository paths in annotation URLs

Git previously matched only by hostname and could use the first credential for every repository, causing authentication failures. It now selects credentials by full repository URL, with host-only credentials kept as fallbacks.

> * :bug: Fix: Add useHttpPath to support multiple Git credentials on same host (#9143)
> Fixed Git credential matching to support multiple repositories on the same host with different credentials. Previously, when using multiple secrets for different repositories on the same Git server (e.g., github.com/org/repo1 and github.com/org/repo2), it incorrectly use the first credential for all repositories, causing authentication failures. Git credential contexts now include `useHttpPath = true`, enabling proper per-repository credential selection.

### Resolver observability metadata restored

- Affects: the resolver deployment's `config/resolvers/config-observability.yaml` configuration

During the OTel migration, its namespace changed to `tekton-pipelines` and the `app.kubernetes.io/component: resolvers` label was removed. The namespace is now `tekton-pipelines-resolvers`, the label is restored, and the OTel configuration content is unchanged.

> * :bug: fix: revert mistaken metadata changes in resolvers config-observability (#9468)

## Enhancements (4)

- Webhook inherits configured TLS policy (To use: Applicable `WEBHOOK_TLS_MIN_VERSION`, `WEBHOOK_TLS_MAX_VERSION`, `WEBHOOK_TLS_CIPHER_SUITES`, and/or `WEBHOOK_TLS_CURVE_PREFERENCES` environment variables)
- Hub Resolver per-request URLs and ordered hub fallback (To use: `artifact-hub-urls` or `tekton-hub-urls` ConfigMap keys, or a per-resolution `url` parameter)
- TaskRun execution can be deferred
- Automatic cleanup for `volumeClaimTemplate` workspace PVCs (To use: Set `tekton.dev/auto-cleanup-pvc` to exactly `"true"` on a PipelineRun using `coschedule: workspaces`.)

## Dependency updates (45)

45 more, none mentioning security

- `* :hammer: chore: bump knative.dev/pkg to main and k8s libs to 0.35.1 (#9470)`
- `* :hammer: build(deps): bump the all group in /tekton with 4 updates (#9652)`
- `* :hammer: build(deps): bump github/codeql-action from 4.33.0 to 4.34.1 (#9651)`
- `* :hammer: build(deps): bump actions/cache from 5.0.3 to 5.0.4 (#9650)`
- `* :hammer: build(deps): bump chainguard-dev/actions from 1.6.8 to 1.6.9 (#9649)`
- `* :hammer: build(deps): bump github.com/spiffe/spire-api-sdk from 1.14.3 to 1.14.4 (#9648)`
- `* :hammer: build(deps): bump k8s.io/apimachinery from 0.35.2 to 0.35.3 (#9639)`
- `* :hammer: build(deps): bump k8s.io/client-go from 0.35.2 to 0.35.3 (#9638)`
- `* :hammer: build(deps): bump k8s.io/api from 0.34.5 to 0.34.6 in /test/custom-task-ctrls/wait-task-beta (#9637)`
- `* :hammer: build(deps): bump k8s.io/client-go from 0.34.5 to 0.34.6 in /test/custom-task-ctrls/wait-task-beta (#9634)`
- `* :hammer: build(deps): bump github.com/spiffe/spire-api-sdk from 1.14.1 to 1.14.3 (#9629)`
- `* :hammer: build(deps): bump google.golang.org/grpc from 1.79.2 to 1.79.3 (#9628)`
- `* :hammer: build(deps): bump github.com/google/go-containerregistry from 0.21.2 to 0.21.3 (#9627)`
- `* :hammer: build(deps): bump github.com/tektoncd/pipeline from 1.10.0 to 1.10.2 in /test/custom-task-ctrls/wait-task-beta (#9626)`
- `* :hammer: build(deps): bump golang.org/x/sync from 0.19.0 to 0.20.0 (#9611)`
- `* :hammer: build(deps): bump the all group in /tekton with 4 updates (#9587)`
- `* :hammer: build(deps): bump github/codeql-action from 4.32.6 to 4.33.0 (#9586)`
- `* :hammer: build(deps): bump fgrosse/go-coverage-report from 1.2.0 to 1.3.0 (#9585)`
- `* :hammer: build(deps): bump step-security/harden-runner from 2.15.1 to 2.16.0 (#9584)`
- `* :hammer: build(deps): bump chainguard-dev/actions from 1.6.7 to 1.6.8 (#9583)`
- `* :hammer: build(deps): bump go.opentelemetry.io/otel/trace from 1.41.0 to 1.42.0 (#9549)`
- `* :hammer: build(deps): bump github.com/google/go-containerregistry from 0.21.1 to 0.21.2 (#9548)`
- `* :hammer: build(deps): bump google.golang.org/grpc from 1.79.1 to 1.79.2 (#9547)`
- `* :hammer: build(deps): bump step-security/harden-runner from 2.15.0 to 2.15.1 (#9542)`
- `* :hammer: build(deps): bump the all group in /tekton with 4 updates (#9541)`
- `* :hammer: build(deps): bump tj-actions/changed-files from 47.0.4 to 47.0.5 (#9540)`
- `* :hammer: build(deps): bump chainguard-dev/actions from 1.6.5 to 1.6.7 (#9539)`
- `* :hammer: build(deps): bump github/codeql-action from 4.32.5 to 4.32.6 (#9538)`
- `* :hammer: build(deps): bump actions/dependency-review-action from 4.8.3 to 4.9.0 (#9536)`
- `* :hammer: build(deps): bump k8s.io/apiextensions-apiserver from 0.35.1 to 0.35.2 (#9487)`
- `* :hammer: build(deps): bump the all group in /tekton with 4 updates (#9483)`
- `* :hammer: build(deps): bump github/codeql-action from 4.32.4 to 4.32.5 (#9482)`
- `* :hammer: build(deps): bump step-security/harden-runner from 2.14.2 to 2.15.0 (#9481)`
- `* :hammer: build(deps): bump actions/setup-go from 6.2.0 to 6.3.0 (#9480)`
- `* :hammer: build(deps): bump chainguard-dev/actions from 1.6.4 to 1.6.5 (#9479)`
- `* :hammer: build(deps): bump actions/upload-artifact from 6.0.0 to 7.0.0 (#9478)`
- `* :hammer: build(deps): bump go.opentelemetry.io/otel/metric from 1.40.0 to 1.41.0 (#9477)`
- `* :hammer: build(deps): bump k8s.io/apimachinery from 0.35.1 to 0.35.2 (#9476)`
- `* :hammer: build(deps): bump k8s.io/client-go from 0.34.3 to 0.34.5 in /test/custom-task-ctrls/wait-task-beta (#9475)`
- `* :hammer: build(deps): bump k8s.io/code-generator from 0.35.1 to 0.35.2 (#9473)`
- `* :hammer: build(deps): bump k8s.io/api from 0.34.3 to 0.34.5 in /test/custom-task-ctrls/wait-task-beta (#9472)`
- `* :hammer: build(deps): bump k8s.io/apiextensions-apiserver from 0.34.3 to 0.34.5 (#9455)`
- `* :hammer: build(deps): bump github.com/tektoncd/pipeline from 1.9.1 to 1.10.0 in /test/custom-task-ctrls/wait-task-beta (#9453)`
- `* :hammer: build(deps): bump k8s.io/client-go from 0.34.3 to 0.34.4 (#9447)`
- `* :hammer: build(deps): bump go.opentelemetry.io/otel/trace from 1.39.0 to 1.40.0 (#9445)`

## Other changes (27)

- Gitea end-to-end tests added to CI and memcached fixture startup arguments fixed
- Name-generation errors now identify the item as a `resource`
- Resolver cache deduplicates concurrent resolutions
- Explicit top-level permissions blocks added to three GitHub Actions workflows
- Image references pinned in the `step-script.yaml` task-run example
- Example TaskRun image references pinned to digests
- Bash, Alpine, and Busybox image references in `examples/v1/taskruns` pinned to digests
- Ubuntu image references pinned to digests in `examples/v1/taskruns`
- `VerificationPolicy` list fetched once per pipeline-state resolution
- GitHub Actions security fixes and zizmor CI check added
- PipelineRun child-resource label lookup moved to `memberOfLookup`
- Legacy GCS release URLs replaced in documentation and scripts
- Gitea test infrastructure upgraded to 1.25.4
- Resolved TaskSpec storage comment clarified
- `opencensus` dependency removed from test files
- `tj-actions/changed-files` version comment updated to v47.0.5
- Release cheat sheet and pipeline use `release.notags.yaml`
- Pipeline API documentation generation restored and published docs updated
- Stale TODO removed from `auth.md`
- Authentication docs distinguish Step credentials from `imagePullSecrets`
- `releases.md` updated with security patch releases
- Metrics documentation now lists four previously undocumented metrics
- Broken internal Markdown links fixed
- README files added for `PipelineRun` and `TaskRun` examples
- Broken Tekton Bundles example link fixed in `taskruns.md`
- `releases.md` updated for v1.10.0
- TaskRun documentation links to `tkn bundle` CLI documentation

---

Summaries are produced by AI, so they can be wrong, and they are provided without warranty.
