Job teardown ran every step on one shared minute. When removing the job container took longer, the network removal ran on an expired context and the network stayed behind until the orphan reclaim, which only takes networks older than `workdir_cleanup_age`, 24h by default. All that time it held a subnet of the daemon's address pool.
The job and service containers are now removed in parallel, and the docker proxy sweep, the network and the volumes get a fresh budget after them. The idle reclaim takes unused networks and volumes once they are 10 minutes old, and network removal treats a network that is already gone as removed.
Reviewed-on: https://gitea.com/gitea/runner/pulls/1318
Reviewed-by: bircni <[email protected]>
Co-authored-by: silverwind <[email protected]>
Each step refreshed its own clone of an action, so a composite using one action repository ten times fetched it ten times per job, and a single network error failed the job.
Like actions/runner, a job now downloads each action repository and ref once, as an archive extracted into a directory per commit. Each action still gets its own copy in the job container. This restores the per-job dedupe of https://gitea.com/gitea/runner/pulls/1178, which only reached a since-deleted branch.
How a download works:
1. The action URL is taken after `insteadOf` rewriting, as git would use it.
2. The archive comes from `<url>/archive/<ref>.tar.gz`, or from codeload.github.com for github.com, as for actions/runner.
3. A cached tag or branch is revalidated with its ETag, which GitHub and Gitea 28+ answer with a 304. A cached commit pin needs no request.
4. Transient failures (5xx, 408, 429 honouring `Retry-After`, dropped transfers) are retried three times with a 10 to 30 second backoff.
5. Git takes over, cloning and exporting the commit into the same directory layout, when the URL is not HTTP(S), the ref is a short commit, the host serves no usable archive, the host cannot be connected to, or the retries run out.
Median of 3 runs, git (shallow clone, then fetch) against archives, on a cold and a warm cache:
| Action | git cold | git warm | archive cold | archive warm |
|---|---|---|---|---|
| github.com actions/checkout@v4 | 937ms | 726ms | 286ms | 171ms |
| github.com actions/cache@v4 | 1006ms | 801ms | 516ms | 161ms |
| github.com actions/setup-node@v4 | 1046ms | 786ms | 463ms | 156ms |
| gitea.com actions/checkout@v4 | 2007ms | 1165ms | 667ms | 222ms |
The action cache layout changes, so existing caches download once more.
*Written by Claude.*
Fixes: https://gitea.com/gitea/runner/issues/1220
Reviewed-on: https://gitea.com/gitea/runner/pulls/1316
Reviewed-by: bircni <[email protected]>
Co-authored-by: silverwind <[email protected]>
The runner's own messages between steps (not step output) were uploaded to the job log at `log.level`, so raising `log.level` for local debugging also changed what workflow authors see in Gitea, and lowering it hid messages from them.
Add `runner.log_report_level` to set that independently. Unset, it still follows `log.level`, so existing setups behave as before. An invalid value is rejected when the config loads.
Assisted-by: Codet
Assisted-by: Claude Code:claude-opus-5-5
Co-authored-by: silverwind <[email protected]>
Co-authored-by: silverwind <[email protected]>
Reviewed-on: https://gitea.com/gitea/runner/pulls/1310
Co-authored-by: Lunny Xiao <[email protected]>
The action cache refresh fetched with `--prune` and the unqualified refspec `+<ref>:refs/runner/<hash>`. Git cannot map that ref back to a remote ref, so it deleted it on every other refresh. The first job after a branch or tag moved then failed with `cannot lock ref`, and a moved tag in a full clone could silently run its old commit. `--no-prune` fixes this and also overrides an inherited `fetch.prune`.
Fixes: https://gitea.com/gitea/runner/issues/1311
Reviewed-on: https://gitea.com/gitea/runner/pulls/1312
Reviewed-by: Lunny Xiao <[email protected]>
Reviewed-by: Zettat123 <[email protected]>
actionslib v1.4.1 no longer logs through logrus, so the workflow model no longer needs the logrus bridge. The bridge stays for docker/cli and moby/go-archive, which still log through logrus's global logger; the comment now says so.
actionslib now logs through slog's global logger, which the runner did not capture. A new slog bridge forwards it to the process-wide logger, so its messages keep the runner's level, format and secret masking.
Assisted-by: Codet
Co-authored-by: silverwind <[email protected]>
Co-authored-by: bircni <[email protected]>
Reviewed-on: https://gitea.com/gitea/runner/pulls/1313
Reviewed-by: silverwind <[email protected]>
Co-authored-by: Lunny Xiao <[email protected]>
A runner nested behind another runner's docker proxy, as in gitea.com CI, had its `com.gitea.runner.job` label overwritten by the outer proxy, so its job cleanup removed nothing. This failed `TestDockerProxyWithDaemon` depending on the CI host, for example on https://gitea.com/gitea/runner/pulls/1312.
Each proxy now sets `com.gitea.runner.job.<job>` and cleans up by that key, so nested proxies keep each other's. This replaces the `com.gitea.runner.job=<job>` label.
Reviewed-on: https://gitea.com/gitea/runner/pulls/1314
Reviewed-by: bircni <[email protected]>
Co-authored-by: silverwind <[email protected]>
Gitea derives the steps of a job from the workflow file, so the `pre` and `post` steps of actions only show up inside "Set up job" and "Complete job", see https://github.com/go-gitea/gitea/issues/24604.
The runner now also reports the steps as it runs them (https://gitea.com/gitea/actions-proto-def/pulls/19): "Set up job", a "Pre" and "Post" step for each action that has one, the workflow steps with their evaluated names, and "Complete job", as actions/runner shows them. The per-step state Gitea reads today is unchanged, and the job log only moves "Complete job name" and "Matrix:" ahead of the pre steps, so older Gitea versions keep showing everything as before. Job hooks and containers stay part of "Set up job" and "Complete job".
The `replace` of `gitea.dev/actionslib` is swapped for the release once the proto change is merged.
Reviewed-on: https://gitea.com/gitea/runner/pulls/1293
Reviewed-by: Zettat123 <[email protected]>
Co-authored-by: bircni <[email protected]>
Replace go-git in `exec --use-gitignore` filtering with `git ls-files`. The git CLI is already required for action clones since https://gitea.com/gitea/runner/pulls/1277, so this drops `go-git` and 14 other modules, and filtering now follows git's own ignore rules.
Also stop deleting a remote action's `.gitignore` before copying it. Tracked files are never filtered, so the deletion only made copied actions differ from GitHub's action tarballs.
Reviewed-on: https://gitea.com/gitea/runner/pulls/1309
Reviewed-by: bircni <[email protected]>
Co-authored-by: silverwind <[email protected]>
Log through `log/slog`, with a console `slog.Handler` that writes the text the runner has always written (sorted `key=value` fields on a plain stream, a colored prefix on a terminal), so nothing changes for users. I compared its output byte for byte with logrus's `TextFormatter` over plain, colored, timestamped and caller cases.
Why not `slog.TextHandler`: its output differs from what users and their log parsers see today, and `ReplaceAttr` cannot close the gap.
- level is `WARN` where logrus wrote `warning`, and time has milliseconds and no quotes
- fields keep call order instead of being sorted
- the caller is one `source=` key, where the runner writes `func=` and `file=` (reshaped by `callPrettyfier`), and `ReplaceAttr` returns a single attribute
- no colored prefix on a terminal
- no hook to rewrite a whole formatted line, which the process-wide secret masking relies on
Libraries that log through logrus's global logger (the workflow model, the Docker client) are forwarded to the process-wide logger, so they keep following its level, format and secret masking. This is why logrus stays in `go.mod`.
Stacked on https://gitea.com/gitea/runner/pulls/1303.
Assisted-by: Codet
Co-authored-by: bircni <[email protected]>
Reviewed-on: https://gitea.com/gitea/runner/pulls/1304
Reviewed-by: bircni <[email protected]>
Move the log calls in `internal` onto `logging`, and add what the process-wide logger setup and its tests need: `SetTextFormat` with a line filter for secret masking (replacing `report.MaskingFormatter`), `SetReportCaller`, `SetDefault` and `logtest.SetDefault`. After this only the `logging` package imports logrus.
Stacked on https://gitea.com/gitea/runner/pulls/1301.
Assisted-by: Codet
Co-authored-by: bircni <[email protected]>
Reviewed-on: https://gitea.com/gitea/runner/pulls/1302
Reviewed-by: bircni <[email protected]>
Add `act/common/logging`: a printf-style `Logger` on a slog-shaped `Handler`, with a logrus-backed handler as the only backend, plus a `logtest` handler for tests. Move the job logger onto it: `common.Logger` returns `*logging.Logger`, the job log formatters and the `Reporter` hook become handlers, and composite step IDs travel in the context.
Logging is tightly coupled to logrus types (hooks, formatters, `Entry.Data`). Going through our own interface first lets the remaining call sites move over with unchanged behavior, and lets the backend be swapped for `log/slog` in one place afterwards. The global process logger and its masking formatter stay on logrus for now.
Assisted-by: Codet
Co-authored-by: bircni <[email protected]>
Reviewed-on: https://gitea.com/gitea/runner/pulls/1295
Reviewed-by: bircni <[email protected]>
- Keep `GITHUB_ENV`, `GITHUB_OUTPUT`, `GITHUB_STATE` and `GITHUB_PATH` when a job times out or is cancelled, so post steps still get their state
- Retry the startup Declare while Gitea is unreachable instead of exiting, still failing fast on rejected registrations, auth and misconfiguration
- Reuse a local image when a platform variant such as `linux/arm/v7` is requested instead of pulling again
- Read `PATH` from the created job container so cross-architecture jobs get tool paths for the right architecture
- Extract action copies relative to the act volume so they never cross an image's `/var/run` symlink, which rootless Podman rejects
- Keep the job container's entrypoint when container options set `--entrypoint`, warning instead of appending it to the keep-alive command
- Lock action cache entries across runner processes that share a cache directory
- `exec`: omit `GITHUB_SERVER_URL` and `GITHUB_API_URL` without `--gitea-instance` instead of exporting a bare `https://`, and read owner and repository from any origin remote
Reviewed-on: https://gitea.com/gitea/runner/pulls/1294
Reviewed-by: bircni <[email protected]>
Co-authored-by: silverwind <[email protected]>
Sends the OTLP metrics under [OpenTelemetry semantic convention](https://opentelemetry.io/docs/specs/semconv/) names where the runner's data matches the definition, like `cicd.worker.count`, `cicd.system.errors` and the `go.*` and `process.*` runtime metrics, and under `gitea.runner.*` otherwise. OTel dashboards then work without remapping, as they already do for the traces from https://gitea.com/gitea/runner/pulls/1206.
Adds a `gitea_runner_state` gauge (`busy`, `idle`, `unavailable`), and `/readyz` now fails once the runner shuts down.
**Breaking**
1. OTLP metric names, units, types and attribute values change, and the info and start time metrics move to resource attributes.
1. Prometheus `gitea_runner_client_errors_total` gains a `code` label, and `gitea_runner_job_total` counts jobs stopped by `runner.timeout` as `timeout` instead of `failure`.
---------
Co-authored-by: silverwind <[email protected]>
Co-authored-by: silverwind <[email protected]>
Co-authored-by: bircni <[email protected]>
Reviewed-on: https://gitea.com/gitea/runner/pulls/1282
Reviewed-by: silverwind <[email protected]>
Co-authored-by: wmTJc9IK0Q <[email protected]>
The Docker proxy labels every container, network and volume a job creates and removes them when the job ends, which breaks jobs that deploy long-running services to the runner's own Docker daemon. Document that behavior and add `container.sweep` (default `true`) so operators who deploy to the same daemon can opt out. Kept resources stay labeled with `com.gitea.runner.job`.
Fixes: https://gitea.com/gitea/runner/issues/1243
Co-authored-by: silverwind <[email protected]>
Reviewed-on: https://gitea.com/gitea/runner/pulls/1275
Reviewed-by: silverwind <[email protected]>
Co-authored-by: bircni <[email protected]>
Add `runner.client_cert_file` and `runner.client_key_file` for Gitea instances that require mutual TLS. The runner presents the certificate on its RPCs, job summary uploads, and the action, reusable workflow and `builtin:checkout` clones from the instance. The files are read on use, so renewed certificates apply without a restart. Job steps do not get them. An incomplete or mismatched pair fails startup.
Clones now decide the token, client certificate and `runner.insecure` with one host check, which also gives reusable workflow clones the TLS settings they lacked. `runner.insecure` still only covers the registered address.
Fixes: https://gitea.com/gitea/runner/issues/321
Co-authored-by: silverwind <[email protected]>
Co-authored-by: silverwind <[email protected]>
Reviewed-on: https://gitea.com/gitea/runner/pulls/1273
Reviewed-by: silverwind <[email protected]>
Co-authored-by: bircni <[email protected]>
Config values can reference the runner's environment as `${NAME}`, for example `capacity: ${RUNNER_CAPACITY}`, so container deployments can configure the runner without templating the file. An unset `NAME` fails the load instead of silently becoming empty, and `$${NAME}` writes a literal `${NAME}`. Keys are never expanded.
Breaking: an existing value containing `${NAME}`, such as a shell command in a Kubernetes `pod_template`, now needs `$${NAME}`.
Closes https://gitea.com/gitea/runner/issues/551
Co-authored-by: silverwind <[email protected]>
Co-authored-by: silverwind <[email protected]>
Reviewed-on: https://gitea.com/gitea/runner/pulls/1274
Reviewed-by: silverwind <[email protected]>
Co-authored-by: bircni <[email protected]>
Add `runner.extra_headers` for reverse proxies that require request headers on the runner's API requests to Gitea, like registration, polling and log and status reports.
Headers the runner sets itself (authentication, User-Agent, protocol headers) stay authoritative. Extra headers are dropped when a redirect changes the hostname or leaves HTTPS, while the runner's own authentication headers follow redirects as before. Invalid or case-insensitively duplicated header names are rejected at config load.
Fixes: https://gitea.com/gitea/runner/issues/171
Co-authored-by: silverwind <[email protected]>
Reviewed-on: https://gitea.com/gitea/runner/pulls/1269
Co-authored-by: bircni <[email protected]>
Allow `valid_volumes` entries ending in `:ro` to permit only read-only mounts, so a workflow can no longer mount writable a host path the runner only shares read-only, like a certificate store. `**:ro` allows any volume read-only. An entry without `:ro` that matches too still allows writable mounts.
A named volume allowed only through `:ro` is mounted with `nocopy`, because Docker otherwise fills an empty volume from the image even when it is mounted read-only.
Fixes: https://gitea.com/gitea/runner/issues/407
Co-authored-by: silverwind <[email protected]>
Reviewed-on: https://gitea.com/gitea/runner/pulls/1272
Reviewed-by: silverwind <[email protected]>
Co-authored-by: bircni <[email protected]>
The runner now uses the `git` command line instead of go-git to download actions and reusable workflows. go-git had recurring bugs and does not support SHA-256 repositories, and full clones of actions are several times faster.
`builtin:checkout` now runs `git` where the job runs and behaves like `actions/checkout`:
1. A branch is checked out as a local branch tracking `origin`, tag and pull request runs keep their ref, and a triggering tag that moved fails the step.
1. Later steps can fetch and push with the persisted token or SSH key.
1. It supports the `ssh-*`, `lfs`, `submodules`, `sparse-checkout`, `filter`, `fetch-tags`, `clean`, `persist-credentials`, `show-progress` and `set-safe-directory` inputs and the `ref` and `commit` outputs.
1. SHA-256 repositories work for actions, reusable workflows and `builtin:checkout`.
Git keeps the token off its command line, ignores `GIT_DIR` and similar variables inherited by the runner, and keeps action files LF regardless of `core.autocrlf`. A cancelled job lets git remove its lock files.
This is a breaking change: the runner needs Git 2.34.1 or newer on its `PATH`, and so do job images that use `builtin:checkout`. The official runner images include it.
Fixes: https://gitea.com/gitea/runner/issues/1077
Fixes: https://gitea.com/gitea/runner/issues/1267
Fixes: https://gitea.com/gitea/runner/issues/1279
Fixes: https://gitea.com/gitea/runner/issues/1284
Fixes: https://gitea.com/gitea/runner/issues/1285
Co-authored-by: silverwind <[email protected]>
Co-authored-by: silverwind <[email protected]>
Co-authored-by: Lunny Xiao <[email protected]>
Reviewed-on: https://gitea.com/gitea/runner/pulls/1277
Reviewed-by: Lunny Xiao <[email protected]>
Reviewed-by: silverwind <[email protected]>
Co-authored-by: bircni <[email protected]>
Fail the workflow before any of its jobs run when a regular job omits `runs-on` or leaves it empty, as GitHub does, instead of running the job in the fallback image. Gitea rejects such workflows itself since https://github.com/go-gitea/gitea/pull/39480. This covers released Gitea versions, which hand a job without labels to any runner, and `exec`. Reusable workflow jobs and the stepless placeholders the runner builds for a job's `needs` are exempt.
Closes https://gitea.com/gitea/runner/issues/836
---------
Co-authored-by: silverwind <[email protected]>
Reviewed-on: https://gitea.com/gitea/runner/pulls/1271
Reviewed-by: silverwind <[email protected]>
Support tmpfs in container `options` via `--mount type=tmpfs,...`, like GitHub does. The `valid_volumes` check silently dropped these mounts because they have no source. A tmpfs is backed by memory and exposes nothing from the host, same as `--tmpfs`, which already passes. Every other mount still goes through the allowlist.
Both forms now work:
```yaml
services:
mysql:
image: mysql:8.4
options: --tmpfs /var/lib/mysql # already worked
```
```yaml
services:
mysql:
image: mysql:8.4
options: --mount type=tmpfs,destination=/var/lib/mysql # was dropped, now works
```
Reviewed-on: https://gitea.com/gitea/runner/pulls/1289
Reviewed-by: bircni <[email protected]>
Co-authored-by: silverwind <[email protected]>
The Kubernetes backend extracted CopyDir archives at "/" with entries like
"var/run/act/actions/<sha>/...". Ubuntu's patched tar refuses to extract
through an absolute symlink, failing with "Invalid cross-device link" on
images that link /var/run to /run, so actions never reached the job
container. Extract at the destination with relative entries instead, like
Copy already does.
This is a similar issue to #1128, but for k8s instead of Docker.
I built the image and deployed it to my k8s cluster to verify the fix.
---------
Co-authored-by: bircni <[email protected]>
Co-authored-by: silverwind <[email protected]>
Reviewed-on: https://gitea.com/gitea/runner/pulls/1286
Reviewed-by: silverwind <[email protected]>
Co-authored-by: Drew Inglis <[email protected]>
Since https://github.com/go-gitea/gitea/pull/39578 (backported to v28), Gitea answers an unregistered runner with a Connect `Unauthenticated` error instead of a gRPC status wrapped as `Unknown`. `IsRegistrationRejected` missed it, so an ephemeral runner kept its spent registration file and failed on every restart. This also fails the `gitea/gitea:main-nightly` e2e job.
Match on the `unregistered runner` message suffix, which both shapes share. The code can't distinguish Gitea from a proxy, because a proxy's 401 with a JSON body also decodes as a wire `Unauthenticated` error.
Written by Claude Code.
Reviewed-on: https://gitea.com/gitea/runner/pulls/1287
Reviewed-by: bircni <[email protected]>
Co-authored-by: silverwind <[email protected]>
- Docker actions evaluate `runs.args` and `runs.env` with their own `inputs`, so a same-named `workflow_dispatch` input no longer replaces the step's `with:` value
- Dockerfile action images are tagged per step and job and removed afterwards, so changed Dockerfiles rebuild and concurrent jobs never share an image
- Workflow and job `env` are evaluated before job and service container credentials and service `env` use them
- Docker actions in host jobs get the host workspace and file-command directory mounted, and no longer inherit the runner's own `PATH`
- Cache server address detection accepts a single interface address
Fixes: https://gitea.com/gitea/runner/issues/390
Fixes: https://gitea.com/gitea/runner/issues/406
Fixes: https://gitea.com/gitea/runner/issues/578
Related: https://gitea.com/gitea/runner/issues/64
Reviewed-on: https://gitea.com/gitea/runner/pulls/1268
Reviewed-by: bircni <[email protected]>
Co-authored-by: silverwind <[email protected]>
`gitea-runner register` pinged the instance in an unbounded loop, so a permanent failure like an untrusted certificate or a wrong URL never returned. It now pings once and returns the error, like other one-shot commands.
The image entrypoint keeps retrying registration indefinitely by default, like the daemon, so a container without a restart policy still recovers once Gitea is reachable. `GITEA_MAX_REG_ATTEMPTS` had no effect because its counter was incremented outside the loop. When set, it now caps the attempts, after which the entrypoint exits non-zero.
Fixes: https://gitea.com/gitea/runner/issues/1195
Co-authored-by: silverwind <[email protected]>
Reviewed-on: https://gitea.com/gitea/runner/pulls/1197
Reviewed-by: Lunny Xiao <[email protected]>
Reviewed-by: silverwind <[email protected]>
Co-authored-by: bircni <[email protected]>
Fixes https://gitea.com/gitea/runner/issues/1263
- Wait for dockerd without a check limit. s6 readiness gave up after 7 checks, so the runner never started on slow hosts, also seen in https://gitea.com/gitea/runner/issues/712, https://gitea.com/gitea/runner/issues/652 and https://gitea.com/gitea/runner/issues/758.
- Let the runner run the docker service's readiness check itself instead of reading s6 status, which can be missing at startup or stale after an unclean stop.
- Exit the container with the first failing service's exit code instead of always 0, so `restart: on-failure` works. A requested stop still exits 0.
- Stop the runner before dockerd, so running jobs can finish within `shutdown_timeout`. The stop grace period has to cover that drain.
- Check the supervision exit codes in `make test-dind`.
---------
Co-authored-by: bircni <[email protected]>
Reviewed-on: https://gitea.com/gitea/runner/pulls/1265
Reviewed-by: bircni <[email protected]>
Push the runner's Prometheus metrics to the OTLP endpoint next to the job traces, so both land in the same backend.
The `/metrics` registry is converted as the [Prometheus compatibility spec](https://opentelemetry.io/docs/specs/otel/compatibility/prometheus_and_openmetrics/) describes and sent through the existing gRPC-free client, so names stay unchanged and no second metrics stack is linked. Metrics are on by default and pushed every interval and once on shutdown, so ephemeral runners report too.
New standard variables, each `OTEL_EXPORTER_OTLP_*` one with `TRACES` and `METRICS` variants:
- `OTEL_METRICS_EXPORTER`, `none` turns metrics off
- `OTEL_METRIC_EXPORT_INTERVAL`, `OTEL_METRIC_EXPORT_TIMEOUT`
- `OTEL_EXPORTER_OTLP_TIMEOUT`, `OTEL_EXPORTER_OTLP_COMPRESSION`
- `OTEL_EXPORTER_OTLP_CERTIFICATE`, `OTEL_EXPORTER_OTLP_CLIENT_CERTIFICATE`, `OTEL_EXPORTER_OTLP_CLIENT_KEY`
---------
Co-authored-by: silverwind <[email protected]>
Co-authored-by: silverwind <[email protected]>
Co-authored-by: bircni <[email protected]>
Reviewed-on: https://gitea.com/gitea/runner/pulls/1248
Reviewed-by: silverwind <[email protected]>
Reviewed-by: bircni <[email protected]>
Co-authored-by: wmTJc9IK0Q <[email protected]>
Ephemeral registrations are single-use, but the daemon kept `.runner` after the task, so a restarted container failed `Declare` with `unregistered runner` forever.
An ephemeral runner now removes its registration file as soon as it fetches a task, so a container killed mid-job registers afresh on restart. It also removes the file when Gitea answers `unregistered runner` at startup or while polling. A proxy's 401 does not count. Persistent runners keep their file, because Gitea sends the same reply for a token mismatch.
Closes https://gitea.com/gitea/runner/issues/818
---------
Co-authored-by: silverwind <[email protected]>
Reviewed-on: https://gitea.com/gitea/runner/pulls/1256
Reviewed-by: silverwind <[email protected]>
Adds a Kubernetes backend. A `name:kubernetes://image` label runs the job as one pod through the Kubernetes API, with services as containers of the same pod. Step env is exported by a script sent over exec stdin, keeping it out of the pod spec and process arguments, and service env goes into a per-job Secret. Docker actions and container options are rejected. See `docs/kubernetes.md`.
Operators shape job pods with `kubernetes.pod_template`, with per-label `pod_templates` merged in `runs-on` order and a `$services` entry for service containers. The idle cleanup removes pods and Secrets a runner left behind.
The client calls the Kubernetes API directly and runs step commands over `coder/websocket` (same dependency used in gitea) instead of pulling in the heavyweight `kubernetes/client-go`. Includes an e2e test against k3s.
Replaces: https://gitea.com/gitea/runner/pulls/1000
Fixes: https://gitea.com/gitea/runner/issues/31
Assisted-by: Claude Code:claude-opus-5-5
Co-authored-by: Zettat123 <[email protected]>
Co-authored-by: bircni <[email protected]>
Reviewed-on: https://gitea.com/gitea/runner/pulls/1260
Reviewed-by: Lunny Xiao <[email protected]>
Reviewed-by: Zettat123 <[email protected]>
Co-authored-by: silverwind <[email protected]>
Replace the `task N repo is ...` line with a structured `task started` entry carrying `task_id`, `job_id`, `run_id`, `run_number`, `repository` and the workflow's `job` key, so external supervisors can match runner activity to the job in Gitea. `job_id` is read from the runtime token's claims, tokens and secrets are never logged.
Closes https://gitea.com/gitea/runner/issues/752
---------
Co-authored-by: silverwind <[email protected]>
Co-authored-by: silverwind <[email protected]>
Reviewed-on: https://gitea.com/gitea/runner/pulls/1257
Reviewed-by: silverwind <[email protected]>
Make Docker-in-Docker the recommended and first-shown setup. Mounting the host's Docker socket gives every job root-equivalent access to the host, while DinD gives jobs their own daemon. The default `latest` tag is unchanged.
Also shorten the image flavour docs to what users need to pick an image, and correct that `dind-rootless` needs `--privileged` too.
Closes https://gitea.com/gitea/runner/issues/1227
---------
Co-authored-by: silverwind <[email protected]>
Co-authored-by: silverwind <[email protected]>
Reviewed-on: https://gitea.com/gitea/runner/pulls/1258
Reviewed-by: silverwind <[email protected]>
Co-authored-by: bircni <[email protected]>
The Docker entrypoint always checked `.runner` in `/data` and ignored `runner.file` from `CONFIG_FILE`. When `runner.file` pointed elsewhere, it re-registered on every start, or skipped registration because of a stale `.runner`. It now reads the path from a new `config runner-file` subcommand, which resolves `runner.file` the same way as `register` and `daemon`.
Closes https://gitea.com/gitea/runner/issues/585
---------
Co-authored-by: silverwind <[email protected]>
Reviewed-on: https://gitea.com/gitea/runner/pulls/1259
Reviewed-by: silverwind <[email protected]>
1. Host mode is selected only through `runs-on` again, as in upstream act.
1. A workflow volume that `valid_volumes` drops no longer displaces the runner's own socket, workspace, tool cache or `/var/run/act` mount, so `/var/run/docker.sock:/var/run/docker.sock` works on hosts whose daemon socket lives elsewhere.
1. `docker://` is stripped from job and service images, as on GitHub.
1. Each created service's log is printed at job end, and service log dumps no longer run workflow commands, as on GitHub.
1. A `--privileged` dropped because privileged mode is off now logs a warning.
1. README: the proxy variables change per step through `env:` or for the rest of a job through `$GITHUB_ENV`, `container.env` does not override them.
Reviewed-on: https://gitea.com/gitea/runner/pulls/1253
Reviewed-by: Zettat123 <[email protected]>
Co-authored-by: silverwind <[email protected]>
Fixes https://github.com/go-gitea/gitea/issues/38705
Gitea side: https://github.com/go-gitea/gitea/pull/39452
Jobs of a called workflow currently see `gitea.event_name == "workflow_call"`, and their `gitea.event.inputs` is replaced by the caller's `with:`. On GitHub they keep the caller's event, and their `inputs` are the run's `workflow_dispatch` inputs overlaid with the caller's `with:`.
For compatibility with older runners, Gitea keeps sending that form and adds a `gitea_workflow_call` key with the original trigger event name and inputs of the run, and the `inputs` Gitea resolved for the job. When the key is present, the runner restores `gitea.event_name` and `gitea.event.inputs` from it, and uses the resolved `inputs` for the job's expressions and the setup log.
Without that key (older Gitea, or jobs outside called workflows), nothing changes.
Reviewed-on: https://gitea.com/gitea/runner/pulls/1250
Reviewed-by: bircni <[email protected]>
Reviewed-by: silverwind <[email protected]>
Co-authored-by: Zettat123 <[email protected]>
Store the cache used by `actions/cache` in an S3-compatible bucket such as AWS S3 or RustFS, with no `cache-server`. Runners with the same settings use the same cache.
A cache saved on one runner can be restored on the others within 30 seconds. The runner talks to the bucket over the S3 REST API with SigV4, without an SDK. Lifecycle rules on the bucket expire old caches.
*Description rewritten by Claude.*
---------
Co-authored-by: silverwind <[email protected]>
Reviewed-on: https://gitea.com/gitea/runner/pulls/1244
Reviewed-by: silverwind <[email protected]>
Co-authored-by: bircni <[email protected]>
Make cache requests go through the runner and remove outbound traffic from the cache server. The runner forwards them to `cache.external_server` when one is set, and attaches itself to each job network so jobs can reach it.
Benefits:
1. Jobs need no http proxy rules for the cache server, which `@actions/http-client` mishandles for CIDR ranges in `no_proxy` (https://github.com/actions/toolkit/pull/2495).
2. Only runners reach a shared cache server, which needs no route to Gitea.
3. A runner container sharing the host Docker socket now caches, where jobs previously could not reach it.
Breaks:
1. Jobs that reach `external_server` but not their runner, such as a remote `DOCKER_HOST`. The built-in cache has always needed that reachability, so both modes now agree. Set `cache.host` and `cache.port` to an address those jobs reach.
2. Jobs on `container.network: bridge`, where Docker rejects the alias, so the runner cannot attach and their cache fails as it does today. Leave `container.network` empty, or name a user-defined network.
*Written by Claude.*
---------
Co-authored-by: bircni <[email protected]>
Reviewed-on: https://gitea.com/gitea/runner/pulls/1229
Reviewed-by: bircni <[email protected]>
Co-authored-by: silverwind <[email protected]>
`valid_volumes` patterns now match with doublestar instead of gobwas/glob, which was compiled without separators so `*` crossed slashes. doublestar is already in the module graph, keeps `*` within one path segment and matches host paths with the OS separator, which fixes Windows patterns where a backslash was read as an escape.
Breaking: a host pattern that relied on `*` matching nested paths needs `**`, e.g. `/data/**` instead of `/data/*`. Named volumes and the bare `**` are unaffected.
Co-authored-by: bircni <[email protected]>
Reviewed-on: https://gitea.com/gitea/runner/pulls/1214
Reviewed-by: bircni <[email protected]>
Co-authored-by: silverwind <[email protected]>
Adds built-in actions, native Go actions shipped with the runner and selected with `uses: builtin:<name>`. Jobs need no action download and no Node in their image.
The first one is `builtin:checkout`, a subset of `actions/checkout` with `repository`, `ref`, `token`, `path` and `fetch-depth`. Other inputs fail the step. The runner clones shallowly on its host and copies the tree into the job, while `gitea-runner exec` copies the local working directory for the workflow's own ref.
`actions/checkout` and other remote actions are unchanged. A clone ref of `HEAD` now shallow-clones the remote's default branch, which checkout uses for other repositories.
---------
Co-authored-by: silverwind <[email protected]>
Co-authored-by: silverwind <[email protected]>
Reviewed-on: https://gitea.com/gitea/runner/pulls/871
Reviewed-by: silverwind <[email protected]>
Co-authored-by: bircni <[email protected]>
Updates actionslib to https://gitea.com/gitea/actionslib/releases/tag/v1.2.1.
1. A job whose `if:` fails to evaluate, like `fromJSON` of an invalid job output, is skipped with the error in its log instead of failing, like on GitHub Actions.
1. Workflows the runner reads itself, like with `exec`, are rejected when an expression does not parse, like an unclosed `${{` or an unknown context, instead of failing later.
1. Task payloads from Gitea are read without that check, because they hold evaluated text that may contain `${{`.
1. `if: False`, `if: ~` and other boolean, null or number conditions work.
Gitea PR: https://github.com/go-gitea/gitea/pull/39358
Reviewed-on: https://gitea.com/gitea/runner/pulls/1247
Reviewed-by: bircni <[email protected]>
Co-authored-by: silverwind <[email protected]>
The dind images run `s6-svscan` as PID 1, so the `docker-init` that the upstream entrypoint puts in front of `dockerd` cannot reap and warns on every start. `TINI_SUBREAPER` makes it a subreaper for the daemon's subtree, which is what tini recommends when it is not PID 1.
The warning was cosmetic. Orphaned `containerd-shim` processes re-parented to `s6-svscan`, which reaps them just as well, so no zombies accumulated. With this change they re-parent to `docker-init` instead, and the warning is gone.
Seen in https://gitea.com/gitea/runner/issues/1240
---------
Co-authored-by: bircni <[email protected]>
Reviewed-on: https://gitea.com/gitea/runner/pulls/1241
Reviewed-by: bircni <[email protected]>
Co-authored-by: silverwind <[email protected]>
A composite action's child run context starts with empty step results, so a nested action's `post-if: success()` saw the composite's own result and still ran after the job had failed, which let `actions/cache` save caches from failed jobs.
The `job` context now always comes from the job's run context. Only `success()` and `failure()` in a composite's own main steps read the composite's result, matching GitHub Actions.
Assisted-by: Claude Code:claude-fable-5-1
Reviewed-on: https://gitea.com/gitea/runner/pulls/1242
Reviewed-by: bircni <[email protected]>
Co-authored-by: silverwind <[email protected]>
Re-sync `act/container/docker_cli.go` and `act/container/docker_cli_test.go` to docker/cli v29.8.1, the version go.mod pins. The copies lacked the `--umask` flag added in v29.8.0, so `container.options` rejected it as unknown. Daemons below API 1.56 ignore the setting.
Reviewed-on: https://gitea.com/gitea/runner/pulls/1238
Reviewed-by: Lunny Xiao <[email protected]>
Co-authored-by: silverwind <[email protected]>
Podman creates a missing bind source instead of rejecting it, so the Docker proxy probe passed even when the daemon could not see the runner's files. Jobs of Podman runners in a container got an empty directory at `/var/run/docker.sock` (https://gitea.com/gitea/runner/issues/1193#issuecomment-1700501).
The probe now binds the directory and stats its marker through the created container. A runner in a container stats the marker through its own container, so it no longer creates probe containers, which cost up to 2s per job. `make test-dind TARGET=podman` runs the probe against Podman in CI.
Reviewed-on: https://gitea.com/gitea/runner/pulls/1231
Reviewed-by: bircni <[email protected]>
Co-authored-by: silverwind <[email protected]>
`waitForRunningJobLog` gave one minute to run creation, scheduling, the container start and the first log flush, then blamed the missing log line, which is how https://gitea.com/gitea/runner/actions/runs/927661 failed. It now gets the suite's `runTimeout` and reports the job's status with the run logs on timeout, and the job holds ten seconds instead of two, so neither the marker nor the cancellation arrives after it has exited.
Against gitea 1.28+ this costs nothing, since the cancel ends the job as soon as the marker appears. Against 1.27, which has no cancel route and waits the job out, the suite goes from about 20s to about 23s, as the tests run in parallel.
*Written by Claude.*
Reviewed-on: https://gitea.com/gitea/runner/pulls/1230
Reviewed-by: bircni <[email protected]>
Co-authored-by: silverwind <[email protected]>
Containers a job starts through its Docker socket, for example `docker run -v "$PWD:/src"`, `./data:/data` in docker compose, or actions like dockerfile-roast, can now bind-mount the workspace and other paths the job sees, as on a host, without `bind_workdir`.
The per-job Docker proxy rewrites container and volume create requests. A bind source, or the device of a `local` volume with `o: bind`, that lies under one of the job container's mounts is pointed at that mount's path on the daemon, read from inspecting the job container. Paths that already name a daemon path, like `GITEA_DOCKER_WORKSPACE`, and paths outside the job's mounts pass through unchanged.
The proxy now also starts when the runner runs in a container given the host's Docker socket, by placing its socket in the runner's working directory, and in rootless dind, by granting the daemon socket's group through an ACL.
Fixes https://gitea.com/gitea/runner/issues/1219
Fixes https://gitea.com/gitea/runner/issues/1193
Reviewed-on: https://gitea.com/gitea/runner/pulls/1226
Reviewed-by: bircni <[email protected]>
Co-authored-by: silverwind <[email protected]>
Mount the job workspace volume, or the `bind_workdir` directory, above `<owner>/<repo>` instead of at the repository, like GitHub's work directory.
With the repository as a mount point, pnpm 12 puts its store in `node_modules/.pnpm-store` (https://github.com/pnpm/pnpm/pull/13536), so a restored `setup-node` pnpm cache creates `node_modules` before install. Renames into `$RUNNER_WORKSPACE` fail with `EXDEV` and removing the workspace fails with `EBUSY`.
Verified against GitHub hosted and container jobs with cold and warm caches, on volume and `bind_workdir` runners, covering checkout, cache, artifacts, github-script, the node, pnpm, bun, python, uv, go, rust, java, gradle, dotnet, ruby and terraform setup actions, docker and Dockerfile actions, buildx, compose via `GITEA_DOCKER_WORKSPACE`, services, host mode, `container.volumes` on the workspace and `exec`.
---------
Co-authored-by: bircni <[email protected]>
Reviewed-on: https://gitea.com/gitea/runner/pulls/1224
Co-authored-by: silverwind <[email protected]>
Every cache hit rewrote the entry's access time, costing a bolt write and fsync per restore. Completed entries accessed within the last minute are no longer rewritten, and the in-use and retention windows grow by that minute so nothing is evicted earlier. Same-second rewrites are skipped, reservations are inserted in one transaction, and prefix lookups use the repository index instead of scanning every entry.
The store is still opened per request. Keeping it open, as this commit first did, is faster under concurrency but stops a second process sharing the cache directory from starting its cache server.
The README caching section is condensed and corrected.
| Benchmark, v2 | main | store kept open | this PR |
| --- | ---: | ---: | ---: |
| Exact hit, 1 client | 16.42 | 0.15 | 0.24 |
| Exact hit, 8 clients | 13.70 | 0.08 | 1.28 |
| Exact hit, 32 clients | 16.01 | 0.07 | 4.26 |
| Prefix hit, 8 clients | 18.34 | 0.12 | 1.28 |
| Miss, 8 clients | 3.21 | 0.11 | 1.28 |
| Save and restore 1 MiB, 8 clients | 74.4 | 22.9 | 34.9 |
| Geomean, all 22 benchmarks | 14.03 | 0.31 | 2.11 |
Mean ms per request, medians of 4 rounds of 300 requests over 1,000 entries on an Apple M3 Max.
---------
Co-authored-by: silverwind <[email protected]>
Co-authored-by: silverwind <[email protected]>
Reviewed-on: https://gitea.com/gitea/runner/pulls/1222
Reviewed-by: silverwind <[email protected]>
Implement speedups to job start and shutdown.
- Create the job container while services are still becoming healthy, and poll their health at a flat one second instead of a 2s to 32s doubling backoff
- Pull each service image once instead of twice, and fetch a warm action cache once instead of twice
- Report the job result before reclaiming its volumes, and reap volumes stranded by a runner that died mid-job
| Step | Scenario | Before | After |
| --- | --- | --- | --- |
| Complete job | Large workspace volume | 4.2s | 0.4s |
| Set up job | One service, 2s health interval | 7.08s | 3.26s |
| Set up job | Two cached actions from github.com | 1.81s | 1.34s |
| Set up job | Two cached actions from gitea.com | 2.42s | 1.95s |
| Set up job | Two actions, cold action cache | 6.62s | unchanged |
| Set up job | Minimal job, no services or actions | 0.62s | unchanged |
Assisted-by: Claude Code:Opus 5
Reviewed-on: https://gitea.com/gitea/runner/pulls/1218
Reviewed-by: bircni <[email protected]>
Co-authored-by: silverwind <[email protected]>
Three independent fixes, one commit each.
1. Hosts and ports were interpolated directly when building URLs, so a literal IPv6 address produced an unbracketed authority, making `ACTIONS_CACHE_URL`, `ACTIONS_RUNTIME_URL` and the artifact server listener unusable. Hosts are expected bare, as documented for `cache.host`, so an address that already carries brackets is no longer accepted.
2. `cache-server` waited on its own `os.Interrupt` channel and ignored SIGTERM, so service managers and container runtimes had to kill it.
3. Task admission used a separate `Load` and `Store`, so two concurrent dispatches of the same task id could both be admitted.
Assisted-by: Claude Code:Opus 5
Co-authored-by: bircni <[email protected]>
Reviewed-on: https://gitea.com/gitea/runner/pulls/1217
Reviewed-by: bircni <[email protected]>
Co-authored-by: silverwind <[email protected]>
Cache v2 makes the cache server the `ACTIONS_RESULTS_URL` origin, so artifact calls arrived there and were proxied on to Gitea, failing whenever it could not reach the instance.
- Artifact calls are answered with a redirect, so the cache server opens no connection to Gitea. A scheme change or an untrusted instance is still proxied, but there the cache server is the runner itself, which already reaches Gitea.
- Failures answer in twirp, not an empty `502` that clients report as `Unexpected end of JSON input`.
- `cache.v2: false` really points artifacts at Gitea now.
- Cache reservations are bound to the job that made them, so two jobs saving one key cannot commit against each other's upload, and a retry after a lost answer no longer fails a saved entry.
- The toolkit patch, which edits the GitHub-host check out of an action's bundle, was left in the shared checkout where a job running with `runner.patch_actions: false` could inherit it. It is put back after the job's copy.
- `exec` names an origin for the cache v2 it advertises, and masks its runtime token.
Behaviour changes: `no_proxy` no longer exempts `cache.external_server`, and `cache.enabled: false` also stops external registration.
Fixes https://gitea.com/gitea/runner/issues/1208
Fixes https://gitea.com/gitea/runner/issues/1211
Assisted by Claude (Opus 5).
Reviewed-on: https://gitea.com/gitea/runner/pulls/1216
Reviewed-by: bircni <[email protected]>
Co-authored-by: silverwind <[email protected]>
Fixes https://gitea.com/gitea/runner/issues/1213
Fix the DooD regression that mounts `/var/run/docker.sock` as a directory. Keep the Docker proxy available through job and post steps. Clean stale resources before opening it, then remove containers before their networks and volumes during teardown.
Use a unique filesystem probe and preserve socket ownership. Fall back to direct access when proxying is unsupported. Preserve exec output and clean up active streams and failed starts.
Add a real Docker job test for mounted socket access, post steps and resource cleanup.
---------
Co-authored-by: silverwind <[email protected]>
Reviewed-on: https://gitea.com/gitea/runner/pulls/1215
Reviewed-by: silverwind <[email protected]>
Co-authored-by: Zettat123 <[email protected]>
1. Dependency updates: gobwas/glob v1.0.0, which replaced its `Glob` interface with `*Pattern`, logrus v1.10.2, prometheus/client_model v0.6.3, golangci-lint v2.13.2, Go 1.27.1, docker/setup-qemu-action and nginx digests.
2. `make fmt-check` and `make tidy-check` report only what the formatter or `go mod tidy` would change, so unrelated uncommitted work no longer fails `make checks`.
---------
Co-authored-by: silverwind <[email protected]>
Reviewed-on: https://gitea.com/gitea/runner/pulls/1212
Reviewed-by: silverwind <[email protected]>
Co-authored-by: Renovate Bot <[email protected]>
2026-09-07 08:20:15 +00:00
214 changed files with 15119 additions and 4821 deletions
Docker Engine Community version is required for docker mode. To install Docker CE, follow the official [install instructions](https://docs.docker.com/engine/install/).
On the runner:
-`git` 2.34.1 or newer on `PATH`, to clone reusable workflows, and actions whose host serves no archive of them.
-`docker` or `podman` with `podman.socket` enabled, for docker mode, Docker actions and `docker://` steps. To install Docker CE, follow the official [install instructions](https://docs.docker.com/engine/install/).
Where steps run, in the job image or on the runner in host mode:
-`bash` or `sh` for `run` steps, `pwsh` or `powershell` on a Windows host.
-`node` for JavaScript actions and `hashFiles()`.
-`git` 2.34.1 or newer for [`builtin:checkout`](docs/builtin-actions.md).
-`git-lfs` and `ssh` for `builtin:checkout`, only when its `lfs` or `ssh-key` input is used.
Kubernetes mode needs [a few more tools](docs/kubernetes.md) in job images.
### Download pre-built binary
@@ -79,53 +91,24 @@ If the registry succeed, it will run immediately. Next time, you could run the r
Mount a volume on `/data` if you want the registration file and optional config to survive container recreation (see [scripts/run.sh](scripts/run.sh)).
> **`/data` does not hold the image cache.** It is the runner's working directory and contains only the `.runner` registration file and, optionally, your config file. Images pulled for jobs live in the *Docker daemon's* data root, which for the `dind` flavours is inside the container (`/var/lib/docker`, or `/home/rootless/.local/share/docker` for `dind-rootless`). To keep the image cache across restarts, give that path its own volume as well — otherwise every new container re-pulls the job images. With the `basic` flavour the images live on whichever daemon you point the runner at, so there is nothing extra to persist.
### Image flavours
The image is published in three flavours, all built from the single multi-stage [Dockerfile](Dockerfile) in this repository. They differ only in how a Docker daemon is made available to the jobs the runner executes; the `gitea-runner` binary inside them is identical.
| Tag | Build target | Base image | Docker daemon | Process supervisor | Runs as |
| --- | --- | --- | --- | --- | --- |
| `latest` (and `<version>`) | `basic` | `alpine` | none — uses an external daemon you provide | [`tini`](https://github.com/krallin/tini) | `root` |
| `latest-dind` | `dind` | `docker:dind` | bundled, started inside the container | [`s6`](https://skarnet.org/software/s6/) | `root` (privileged) |
| `latest-dind-rootless` | `dind-rootless` | `docker:dind-rootless` | bundled, started rootless inside the container | [`s6`](https://skarnet.org/software/s6/) | `rootless` (UID 1000) |
#### `latest` — basic
The default flavour ships only the runner on a minimal Alpine base. It contains **no Docker daemon of its own**: jobs that use `docker://` images need a daemon supplied from outside the container, typically by bind-mounting the host's socket:
```bash
docker run -e GITEA_INSTANCE_URL=https://your_gitea.com -e GITEA_RUNNER_REGISTRATION_TOKEN=<your_token> \
`tini` is the entrypoint (it reaps zombie processes), and it just runs [`scripts/run.sh`](scripts/run.sh), which registers the runner on first start and then execs `gitea-runner daemon`. This flavour does not need `--privileged`. The trade-off is that jobs share the host's daemon, so they can see other containers and images on that daemon.
#### `latest-dind` — Docker-in-Docker
This flavour is based on the official `docker:dind` image and bundles its own Docker daemon, so it needs no external socket — only the `--privileged` flag that Docker-in-Docker requires:
The recommended Docker-in-Docker image runs its own Docker daemon for jobs. Jobs never get the host's Docker socket, which is equivalent to root access on the host.
```bash
docker run --privileged -e GITEA_INSTANCE_URL=https://your_gitea.com -e GITEA_RUNNER_REGISTRATION_TOKEN=<your_token> \
Two processes have to run side by side here (the Docker daemon and the runner), so the entrypoint is the [`s6`](https://skarnet.org/software/s6/) supervision tree under [`scripts/s6`](scripts/s6) instead of `tini`. `s6` starts `dockerd`, and the runner service waits for the daemon to come up (`s6-svwait`) before launching [`run.sh`](scripts/run.sh). Each container has a private daemon isolated from the host's, at the cost of running privileged.
The volumes keep the runner's registration and the job image cache across container recreation.
Same idea as `dind`, but built on `docker:dind-rootless` so the bundled daemon and the runner run as an unprivileged user (`rootless`, UID 1000) rather than `root`. `DOCKER_HOST` is preset to `unix:///run/user/1000/docker.sock` so the runner talks to the rootless daemon. This reduces the blast radius compared to the privileged `dind` flavour, but rootless Docker carries the usual rootless limitations (networking, cgroups, storage drivers, and some operations that need additional host configuration such as `/etc/subuid` / `/etc/subgid` mappings and unprivileged user-namespace support).
| `latest-dind-rootless` | Bundled, runs as a non-root user (UID 1000) for stronger isolation. Needs `--privileged` and [rootless Docker](https://docs.docker.com/engine/security/rootless/) support on the host. Keeps its image cache in `/home/rootless/.local/share/docker`. |
| `latest` | None. Mount the host's socket with `-v /var/run/docker.sock:/var/run/docker.sock`, or set `DOCKER_HOST`. |
> **The UID is fixed at 1000.** It comes from the `rootless` user baked into the upstream `docker:dind-rootless` base image, and the bundled daemon always listens on `/run/user/1000/docker.sock` inside the container, so running this flavour as a different user (`--user 1001`) does not work. If you need the runner to talk to a *host* rootless daemon that runs under some other UID, use the `basic` flavour instead and bind-mount that daemon's socket (see [examples/vm/rootless-docker.md](examples/vm/rootless-docker.md)); pointing `DOCKER_HOST` at a host socket from inside `dind-rootless` will not work. Changing the UID otherwise means rebuilding the image from a base with a different `rootless` user.
> **Note on Podman:** these images target the Docker daemon. The bundled `dind`/`dind-rootless` daemons are `dockerd`, not Podman, and the `basic` flavour expects a Docker-compatible socket. Running them under rootless Podman is not a supported configuration, though pointing the `basic` flavour at a Podman socket that emulates the Docker API may work for some workloads.
Versioned and `nightly` tags carry the same suffixes, for example `nightly-dind`. The images target Docker, not Podman.
### Configuration
@@ -186,7 +169,7 @@ Of the [runner images](https://gitea.com/gitea/runner-images), the `-full` flavo
#### Environment variables
Earlier releases let a few environment variables (`GITEA_DEBUG`, `GITEA_TRACE`, `GITEA_RUNNER_CAPACITY`, `GITEA_RUNNER_FILE`, `GITEA_RUNNER_ENVIRON`, `GITEA_RUNNER_ENV_FILE`) override parts of the config. They are gone, use the YAML file for all settings. The Docker images still read their own variables, such as `RUNNER_STATE_FILE`, see [scripts/run.sh](scripts/run.sh) and the container documentation below.
Earlier releases let a few environment variables (`GITEA_DEBUG`, `GITEA_TRACE`, `GITEA_RUNNER_CAPACITY`, `GITEA_RUNNER_FILE`, `GITEA_RUNNER_ENVIRON`, `GITEA_RUNNER_ENV_FILE`) override parts of the config. They are gone, use the YAML file for all settings. Config values can read the runner's environment as `${NAME}`, for example `capacity: ${RUNNER_CAPACITY}`. An unset `NAME` fails the load, and `$${NAME}` is a literal `${NAME}`. The Docker images still read their own variables, such as `RUNNER_STATE_FILE`, see [scripts/run.sh](scripts/run.sh) and the container documentation below.
### Labels
@@ -201,10 +184,10 @@ A label is written as:
| Part | Meaning |
| --- | --- |
| `name` | The name a workflow refers to in `runs-on`, e.g. `ubuntu-latest`. |
| `schema` | Either `docker` or `host`. Defaults to `host` when omitted. |
| `args` | Only used by the `docker` schema: the image to run the job in. |
| `schema` | `docker`, `kubernetes` or `host`. Defaults to `host` when omitted. |
| `args` | Only used by the `docker` and `kubernetes` schemas: the image to run the job in. |
Two schemas are supported:
Three schemas are supported:
- **`docker://<image>`** — the job runs inside a container created from `<image>`:
a workflow with `runs-on: ubuntu-latest` is executed in the `runner-images:ubuntu-latest` container, and one with `runs-on: macos` is executed directly on the host.
Names may themselves contain a colon (for example `pool:e57e18d4-10d4-406f-93bf-60f127221bdd`); only `host` and `docker` are treated as schemas.
Names may themselves contain a colon, for example `pool:e57e18d4-10d4-406f-93bf-60f127221bdd`. Only `host`, `docker` and `kubernetes` are treated as schemas.
If a job's `runs-on` matches none of the runner's labels, or sets no `runs-on` at all, it still runs: in `runner.default_image` where docker is available, on the host where it is not. Images maintained for this purpose are listed at [gitea/runner-images](https://gitea.com/gitea/runner-images).
Jobs run on `runner.default_image` when no runner label matches their `runs-on`: as a pod on runners with `kubernetes` labels, otherwise in a container, or on the host when docker is unavailable. Suitable images are listed at [gitea/runner-images](https://gitea.com/gitea/runner-images).
Labels are chosen at registration time (`--labels`, or the interactive prompt) and can be changed afterwards by editing `runner.labels` in the config file, or in the Gitea UI under the runner's settings.
@@ -268,14 +257,14 @@ Unlike GitHub, a job whose steps run on the host (a `host` label without `contai
#### Docker from a job (`GITEA_DOCKER_WORKSPACE`)
A container a job starts through the Docker socket cannot bind-mount the workspace by the job's own path, the daemon does not have it. `GITEA_DOCKER_WORKSPACE` holds the path the daemon sees. Use it as the prefix of workspace binds, with `.` as the fallback for local use, here in a `docker-compose.yaml`:
Containers a job starts through the Docker socket can bind-mount the workspace by the job's own path, as on a host, for example `docker run -v "$PWD:/src"`. Without the Docker proxy, for example with a remote `DOCKER_HOST`, use `GITEA_DOCKER_WORKSPACE`, the path the daemon sees, as the prefix of workspace binds, with `.` as the fallback for local use:
```yaml
volumes:
- ${GITEA_DOCKER_WORKSPACE:-.}/data:/app/data
```
Containers, networks and volumes a job creates through the socket are removed when the job ends. Both apply to jobs in containers, on the host `.` already works.
Linux container jobs use a Docker proxy when socket sharing and permissions allow it. Other setups use the daemon socket directly. Host jobs use their existing Docker access, where `.` works. The proxy removes the containers, networks and volumes a job creates, `docker compose` services included, after post steps and the completed hook. Set `container.sweep: false` when jobs deploy services to the runner's Docker daemon.
#### Proxy
@@ -291,14 +280,14 @@ The runner uses them for its own requests and gives them to every job, in lower
These hosts are added to `no_proxy` for jobs, so they are always reached directly:
- the cache server
- the built-in cache server, whose address is assigned at startup
- `localhost`, `127.0.0.1` and `::1`
- the job's service containers
- the Docker daemon, when it is reached over `tcp://`
Gitea is not added. Add it to `no_proxy` yourself if it should be reached directly.
To change a value for one job, set it in a step's `env:` or in the job's `container.env`. Setting it at workflow or job level has no effect. To change it for the whole runner, set it in `runner.envs`. A `no_proxy` set there is added to the list above instead of replacing it.
For one step, set it in the step's `env:`, for the rest of a job, write it to `$GITHUB_ENV`. Setting it at workflow or job level or in `container.env` has no effect. To change it for the whole runner, set it in `runner.envs`. A `no_proxy` set there is added to the list above instead of replacing it.
Images are pulled by the Docker daemon, which needs its own proxy setting. In the `dind` images the daemon runs in the same container and reads the variables above. For any other daemon, see [the Docker documentation](https://docs.docker.com/engine/daemon/proxy/). The runner logs a warning at startup if it has a proxy and the daemon does not.
@@ -306,63 +295,68 @@ Dockerfile actions are built with these variables as build arguments, so their `
A password in a proxy URL is hidden in job logs. Any step can still read it, because the step is given the proxy URL in its environment.
#### Built-in actions (`uses: builtin:<name>`)
Use `uses: builtin:checkout` to check out a repository without downloading `actions/checkout` or needing Node in the job image, see **[docs/builtin-actions.md](docs/builtin-actions.md)**.
#### Caching (`actions/cache`)
Each runner starts its own cache server automatically. Cache entries are local to that runner — runners do not share a cache by default.
Each runner starts its own cache server, so runners do not share cached entries. A runner in a container on the jobs' Docker daemon attaches itself to each job network, so jobs reach its cache without extra setup.
**Eviction**
**Sharing a cache between runners**
An entry nothing has read or written for `retention` is removed, and a repository past `repo_size_limit` loses its least recently accessed entries until it fits; `size_limit` caps the whole cache the same way. Age alone never retires an entry still in use, and whatever these allow, the cache keeps free space above `health_check.min_free_disk_space_mb` when health checks are enabled.
Run one `gitea-runner cache-server` and point every runner at it, using the same secret everywhere, for example from `openssl rand -hex 32`:
These apply where the cache server runs, so on a shared server they belong in *its* config, not the runners'. See `retention`, `repo_size_limit`, `size_limit` and `sweep_interval` in [config.example.yaml](internal/pkg/config/config.example.yaml) for units and defaults.
```yaml
# cache-server.yaml
cache:
dir: /data/actcache
port: 8088
external_secret: "<secret>" # or external_secret_file: /path/to/secret
```
**Cache service v2**
```bash
gitea-runner -c cache-server.yaml cache-server
```
`actions/cache@v4.2` and later can use the *cache service v2* API. The runner serves it from the same store as v1, on by default, and it works with `external_server`. Turn it off with:
`--dir`, `--host` and `--port` override the matching `cache` keys. Eviction settings take effect on the cache server, not on the runners.
Runners can also share one `cache.dir` on a file system with working file locks, at the cost of slower cache requests.
**Storing the cache in an S3-compatible bucket**
The cache can also be stored in an S3-compatible bucket such as AWS S3, MinIO or RustFS, with no cache server. Runners with the same settings use the same cache:
Those actions refuse any host they do not take for GitHub. Rather than misreport the server URL, the runner edits that check out of the action's own bundle on its way into the job, undone whenever the action is downloaded again. A bundle it does not recognise is left alone and keeps to v1. The same edit lets the stock `actions/upload-artifact` and `actions/download-artifact` work from `v4.4.0` on, without the `gitea-upload-artifact` fork, so it is made whatever `v2` says: that setting only governs the API the runner advertises. Set `runner.patch_actions: false` to leave every bundle exactly as shipped, an escape hatch for an action the edit breaks. The artifact actions then refuse again and the cache client keeps to v1.
A cache saved on one runner can be restored on the others within 30 seconds. The eviction settings below do not apply to the bucket, so add lifecycle rules to it that expire old caches and abort incomplete multipart uploads.
**Shared cache across multiple runners**
**Eviction**
Run one dedicated `gitea-runner cache-server` that all runners point at.
Entries not read or written within `retention` (default `168h`) are removed. A repository over `repo_size_limit` (default `10GB`) loses its least recently used entries, and `size_limit` (off by default) caps the whole cache the same way. Entries in use are never removed. The cache also keeps 1024 MiB free on its volume, or `health_check.min_free_disk_space_mb` when health checks are enabled. See [config.example.yaml](internal/pkg/config/config.example.yaml) for all options.
1. Create a config file for the cache server host:
# external_secret_file: /path/to/secret # secret can also be passed via a file
```
`actions/cache` v3.4.0, v4.2.0 and later use the cache service v2 API, which the runner serves by default. These actions fall back to v1 on hosts they do not recognize as GitHub, so the runner removes that check from them while a job runs. This also lets the stock `actions/upload-artifact` v4.4.0 and `actions/download-artifact` v4.1.8 and later work without the `gitea-upload-artifact` fork. With `runner.patch_actions: false`, the cache stays on v1 and the artifact actions fail.
external_secret: "replace-with-a-strong-random-secret" # must match the server
# external_secret_file: /path/to/secret # secret can also be passed via a file
```
Jobs reach the cache server at `external_server`, so when a reverse proxy fronts the server, point `external_server` at the proxy. The cache server itself needs no extra configuration.
Alternatively, mount the same NFS/CIFS share on every runner and point `cache.dir` at it — simpler, but with weaker isolation between repositories.
**S3 / MinIO** — mount object storage as a FUSE filesystem (e.g. [s3fs](https://github.com/s3fs-fuse/s3fs-fuse) or [goofys](https://github.com/kahing/goofys)) and set `cache.dir` to the mount point.
Flags `--dir`, `--host`, and `--port` on `cache-server` override the corresponding `cache.*` YAML keys; all other settings, including `external_secret`, require the config file.
With v2, artifact calls also go through the cache server, so jobs must be able to reach it. `cache.v2: false` sends them to Gitea directly, unless the Gitea URL has a path or `runner.insecure` is set with HTTPS.
#### Official Docker image
@@ -374,7 +368,8 @@ While the runner is idle it cleans up after earlier jobs:
- when `container.bind_workdir` is enabled, stale task workspace directories older than `runner.workdir_cleanup_age` are removed (default: `24h`; set `0` to disable)
- only purely numeric subdirectories under `container.workdir_parent` are treated as task workspaces and may be removed
- cleanup assumes `container.workdir_parent` is not shared across multiple runners
- on runners that use docker, per-job networks left behind by jobs the runner did not live to tear down are removed, identified by the `com.gitea.runner.uuid` label carrying this runner's uuid
- on runners that use docker, per-job networks and volumes left behind by jobs the runner did not live to tear down are removed once unused and 10 minutes old, identified by the `com.gitea.runner.uuid` label carrying this runner's uuid
- on runners with `kubernetes` labels, job pods with that label older than `runner.workdir_cleanup_age` are removed the same way, together with their Secrets
- cleanup runs every `runner.idle_cleanup_interval` (default: `10m`; set `0` to disable), and setting either knob to `0` disables all of the above
#### Post-task script (`runner.post_task_script`)
@@ -397,6 +392,10 @@ Both hooks are synchronous and block the job while they run. Either one exiting
See **[docs/job-hooks.md](docs/job-hooks.md)** for the execution order, environment, and platform notes.
#### OpenTelemetry (`OTEL_*`)
Set `OTEL_EXPORTER_OTLP_ENDPOINT` in the runner's environment to send a trace of every job, and the runner's metrics, to an OpenTelemetry collector. See **[docs/telemetry.md](docs/telemetry.md)**.
#### Local job logs (`log.job.dir`)
Set `log.job.dir` to a path and the runner writes a copy of every task's log there as `<start time>-task-<id>.log`: the rows exactly as Gitea received them, with the same secrets masked and the job's result on the last line. Off by default, and what Gitea shows does not change.
// insecureTransport is shared, because a transport per request would pool no connections.
varinsecureTransport=&http.Transport{TLSClientConfig:&tls.Config{InsecureSkipVerify:true}}//nolint:gosec // the runner reaches its instance on the operator's say-so
varinsecureTransport=&http.Transport{Proxy:http.ProxyFromEnvironment,TLSClientConfig:&tls.Config{InsecureSkipVerify:true}}//nolint:gosec // the runner reaches its instance on the operator's say-so
for_,data:=range[]string{date[:8],s.opts.Region,"s3","aws4_request",fmt.Sprintf("AWS4-HMAC-SHA256\n%s\n%s\n%x",date,scope,sha256.Sum256([]byte(strings.Join(canonical,"\n"))))}{// key derivation, the last round signs
funcFetchAction(ctxcontext.Context,inputNewGitCloneExecutorInput)(string,string,error){// input.Dir is one clone URL and ref's, holding a directory per commit
iferr!=nil||source.Scheme!="http"&&source.Scheme!="https"||hexRefRegex.MatchString(input.Ref)&&!isFullObjectID(input.Ref){// only git tells a short commit from a branch
funcextractTree(archiveio.Reader,dirstring,stripbool)(string,error){// into dir/<commit of the pax header>, strip dropping the directory wrapping the files
t.Run("keeps an unprefixed archive whole, downloads where insteadOf points and exports a clone without an archive or from an unreachable host",func(t*testing.T){
returnfmt.Errorf("docker has no address pool left for this job's network, lower runner.capacity or widen default-address-pools in the docker daemon config: %w",err)
}
returnfmt.Errorf("docker has no address pool left for this job's network, lower runner.capacity or widen default-address-pools in the docker daemon config: %w",err)
common.Logger(ctx).Warnf("Refusing to remove network %s because it still has active endpoints, the orphan cleanup reclaims it once they are gone",name)
Loaded 50 of 214 files, more files were not shown because too many files have changed in this diff.
Show more
Reference in new issue
Block a user
Blocking a user prevents them from interacting with repositories, such as opening or commenting on pull requests or issues. Learn more about blocking a user.
silverwind
silverwind
Renovate Botandsilverwind
