Add the maintenance procedure

.github/workflows/main.yaml

@@ -49,13 +49,53 @@ jobs:
           export SLACK_APP_TOKEN=${{ secrets.SLACK_APP_TOKEN }}
           export SLACK_BOT_TOKEN=${{ secrets.SLACK_BOT_TOKEN }}
           EOF
+      - name: Clean up unnecessary files
+        run: |
+          docker image prune -af
+          sudo rm -rf /home/linuxbrew || true
+          sudo rm -rf /opt/az || true
+          sudo rm -rf /opt/microsoft || true
+          sudo rm -rf /opt/pipx || true
+          sudo rm -rf /opt/google || true
+          sudo rm -rf /usr/share/dotnet || true
+          sudo rm -rf /usr/local/lib/android || true
+          sudo rm -rf /usr/local/share/boost || true
+          sudo rm -rf /usr/lib/jvm || true
+          sudo rm -rf /usr/share/swift || true
+          sudo rm -rf /usr/local/julia* || true
+          sudo rm -rf /usr/local/n || true
+          sudo rm -rf /usr/share/kotlinc || true
+          sudo rm -rf /usr/local/share/edge_driver || true
+          sudo rm -rf /usr/local/share/chromedriver-linux64 || true
+          sudo rm -rf /usr/local/share/gecko_driver || true
+          sudo rm -rf /usr/share/miniconda || true
+          sudo rm -rf /usr/local/share/phantomjs* || true
+          sudo rm -rf /usr/share/sbt || true
+          sudo rm -rf /usr/local/aws-cli || true
+          sudo rm -rf /usr/local/aws-sam-cli || true
+          sudo rm -rf /usr/local/sqlpackage || true
+          sudo rm -rf /usr/local/bin/minikube || true
+          sudo rm -rf /usr/local/bin/kustomize || true
+          sudo rm -rf /usr/local/bin/kubectl || true
+          sudo rm -rf /usr/local/bin/kind || true
+          sudo rm -rf /usr/local/bin/helm || true
       - run: make setup KINDTEST_K8S_VERSION=${{ matrix.k8s-version }}
       - run: make -C kindtest start KINDTEST_K8S_VERSION=${{ matrix.k8s-version }}
       - run: make -C kindtest test
         env:
           GIT_SSH_COMMAND: "ssh -i /tmp/deploy-key.pem"
       - run: make -C kindtest logs
         if: always()
+
+      - name: Check disk usage and docker images
+        run: |
+          df -h
+          docker images
+        if: always()
+      - name: Check directory sizes
+        run: sudo du -sh /usr/local/* /home/* /opt/* /tmp/* /usr/* /var/* | sort -hr
+        if: always()
+
       - uses: actions/upload-artifact@v4
         if: always()
         with:

CHANGELOG.md

@@ -25,8 +25,8 @@ This project adheres to [Semantic Versioning](http://semver.org/).
 We migrated the image repositories of meows to `ghcr.io`.
 From meows v0.14.0, please use the following images.
 
-- https://github.com/cybozu-go/meows/pkgs/container/meows-controller
-- https://github.com/cybozu-go/meows/pkgs/container/meows-runner
+- <https://github.com/cybozu-go/meows/pkgs/container/meows-controller>
+- <https://github.com/cybozu-go/meows/pkgs/container/meows-runner>
 
 The images on Quay.io ([meows-controller](https://quay.io/repository/cybozu/meows-controller), [meows-runner](https://quay.io/repository/cybozu/meows-runner)) will not be updated in the future.
 
@@ -41,7 +41,6 @@ The images on Quay.io ([meows-controller](https://quay.io/repository/cybozu/meow
 - Support Kubernetes 1.27 ([#178](https://github.com/cybozu-go/meows/pull/1781))
 - Build with go 1.21 ([#178](https://github.com/cybozu-go/meows/pull/178))
 
-
 ## [0.12.0] - 2023-07-05
 
 ### Changed

Dockerfile

@@ -1,10 +1,10 @@
-FROM ghcr.io/cybozu/golang:1.23-jammy as builder
+FROM ghcr.io/cybozu/golang:1.23-jammy AS builder
 
 WORKDIR /workspace
 COPY . .
 RUN make build
 
-FROM ghcr.io/cybozu/ubuntu:22.04 as controller
+FROM ghcr.io/cybozu/ubuntu:22.04 AS controller
 LABEL org.opencontainers.image.source="https://github.com/cybozu-go/meows"
 
 COPY --from=builder /workspace/tmp/bin/controller /usr/local/bin
@@ -14,14 +14,15 @@ COPY --from=builder /workspace/tmp/bin/meows /usr/local/bin
 USER 10000:10000
 ENTRYPOINT ["controller"]
 
-FROM ghcr.io/cybozu/ubuntu:22.04 as runner
+FROM ghcr.io/cybozu/ubuntu:22.04 AS runner
 LABEL org.opencontainers.image.source="https://github.com/cybozu-go/meows"
 
 # Even if the version of the runner is out of date, it will self-update at job execution time. So there is no problem to update it when you notice.
 # TODO: Until https://github.com/cybozu-go/meows/issues/137 is fixed, update it manually.
-ARG RUNNER_VERSION=2.319.1
+ARG RUNNER_VERSION=2.320.0
 
 ENV DEBIAN_FRONTEND=noninteractive
+# hadolint ignore=DL3015
 RUN apt-get update -y \
   && apt-get install -y software-properties-common \
   && add-apt-repository -y ppa:git-core/ppa \

README.md

@@ -42,6 +42,7 @@ You can run jobs in your GitHub Actions workflows on your Kubernetes cluster wit
 ## Docker images
 
 Docker images are available on [ghcr.io](https://github.com/orgs/cybozu-go/packages?repo_name=meows)
+
 - [Controller](https://github.com/cybozu-go/meows/pkgs/container/meows-controller)
 - [Runner](https://github.com/cybozu-go/meows/pkgs/container/meows-runner)
 

RELEASE.md

@@ -1,15 +1,12 @@
-Release procedure
-=================
+# Release procedure
 
 This document describes how to release a new version of meows.
 
-Versioning
-----------
+## Versioning
 
 Follow [semantic versioning 2.0.0][semver] to choose the new version number.
 
-Prepare change log entries
---------------------------
+## Prepare change log entries
 
 Add notable changes since the last release to [CHANGELOG.md](CHANGELOG.md).
 It should look like:
@@ -19,65 +16,67 @@ It should look like:
 ## [Unreleased]
 
 ### Added
+
 - Implement ... (#35)
 
 ### Changed
+
 - Fix a bug in ... (#33)
 
 ### Removed
+
 - Deprecated `-option` is removed ... (#39)
 
 (snip)
 ```
 
-Bump version
-------------
+## Bump version
 
 1. Determine a new version number. Then set `VERSION` variable.
 
-    ```console
+    ```bash
     # Set VERSION and confirm it. It should not have "v" prefix.
-    $ VERSION=x.y.z
-    $ echo $VERSION
+    VERSION=x.y.z
+    echo $VERSION
     ```
 
 2. Make a branch to release
 
-    ```console
-    $ git neco dev "bump-$VERSION"
+    ```bash
+    git switch -c "bump-$VERSION"
     ```
 
 3. Edit `CHANGELOG.md` for the new version ([example][]).
 4. Bump image version.
 
-    ```console
-    $ sed -i -E "s/(.*newTag: ).*/\1${VERSION}/" config/controller/kustomization.yaml config/agent/kustomization.yaml
-    $ sed -i -E "s/(.*Version = ).*/\1\"${VERSION}\"/" constants.go
+    ```bash
+    sed -i -E "s/(.*newTag: ).*/\1${VERSION}/" config/controller/kustomization.yaml config/agent/kustomization.yaml
+    sed -i -E "s/(.*Version = ).*/\1\"${VERSION}\"/" constants.go
     ```
 
 5. Commit the change and push it.
 
-    ```console
-    $ git commit -a -m "Bump version to $VERSION"
-    $ git neco review
+    ```bash
+    git commit -a -m "Bump version to $VERSION"
+    git push origin "bump-$VERSION"
     ```
 
 6. Merge this branch.
 7. Add a git tag to the main HEAD, then push it.
 
-    ```console
+    ```bash
     # Set VERSION again.
-    $ VERSION=x.y.z
-    $ echo $VERSION
+    VERSION=x.y.z
+    echo $VERSION
 
-    $ git checkout main
-    $ git pull
-    $ git tag -a -m "Release v$VERSION" "v$VERSION"
+    git checkout main
+    git pull
+    git tag -a -m "Release v$VERSION" "v$VERSION"
 
     # Make sure the release tag exists.
-    $ git tag -ln | grep $VERSION
+    git tag -ln | grep $VERSION
 
-    $ git push origin "v$VERSION"
+    git push origin "v$VERSION"
     ```
 
 GitHub actions will build and push artifacts such as container images and

docs/commands.md

@@ -1,12 +1,10 @@
-CLI options
-===========
+# CLI options
 
-`controller`
------------
+## `controller`
 
 The CLI allows you to use the following options:
 
-```bash
+```console
 $ controller -h
 Kubernetes controller for GitHub Actions self-hosted runner
 
@@ -41,9 +39,7 @@ Flags:
       --zap-stacktrace-level level         Zap Level at and above which stacktraces are captured (one of 'info', 'error', 'panic').
 ```
 
-
-`slack-agent`
--------------
+## `slack-agent`
 
 The Slack agent is a server program.
 This notifies CI results and accepts requests for extending Pods' lifecycles
@@ -68,9 +64,7 @@ Flags:
   -v, --verbose              Verbose.
 ```
 
-
-`meows`
-------
+## `meows`
 
 This is a tool command to do some operations.
 It enables to send requests to the slack-agent, or to control the GitHub runners.

docs/design.md

@@ -135,7 +135,7 @@ Runner has the `status` and `busy` state as written [here](https://docs.github.c
 If the `--ephemeral` option is given to `config.sh` does not repeat the
 long polling again, and never gets `online` after the assigned job is done.
 This behavior is useful for ensuring to make a clean environment for each job.
-ref: https://docs.github.com/en/actions/hosting-your-own-runners/autoscaling-with-self-hosted-runners#using-ephemeral-runners-for-autoscaling
+ref: <https://docs.github.com/en/actions/hosting-your-own-runners/autoscaling-with-self-hosted-runners#using-ephemeral-runners-for-autoscaling>
 
 #### A job is scheduled only on a `online` runner
 
@@ -189,7 +189,7 @@ meows sets the namespaced name of a `RunnerPool` as a custom label.
       command when the job is failed.  The `if: failure()` syntax allows users
       to run the step only when one of previous steps exit with non-zero code.
    1. Publish the timestamp of when to delete this pod in the `/deletion_time` endpoint.
-      If the job is succeeded or canceled, the `Pod` publishes the current time for 
+      If the job is succeeded or canceled, the `Pod` publishes the current time for
       delete itself. If the job is failed, the `Pod` publishes the future time for
       delete itself, for example 20 min later.
 1. The Slack agent notifies the result of the job on a Slack channel.
@@ -208,15 +208,15 @@ A Runner `Pod` has the following state as a GitHub Actions job runner.
     for example, booting a couple of VMs needed in a job before the job is assigned.
 - `running`: `Pod` is running. Registered in GitHub Actions.
 - `debugging`: The job has finished with failure and Users can enter `Pod` to debug.
-- `stale`: The environment in the `Pod` is dirty. If a runner restarts before completing a job, 
+- `stale`: The environment in the `Pod` is dirty. If a runner restarts before completing a job,
     the environment in the `Pod` may be dirty. This state means waiting for the Pod
     to be removed to prevent Job execution with that stale Pod.
 
 In addition, it has the following states as the exit state of the execution result of `Runner.Listener`.
 
 - `retryable_error`: If execution fails due to a factor other than a job, restart `Runner.Listener`.
-- `updating`: When a new `Runner.Listener` is released, it updates itself and restarts` Runner.Listener`.
-- `undefined`: When the exit code of `Runner.Listener` is undefined. It restarts` Runner.Listener`.
+- `updating`: When a new `Runner.Listener` is released, it updates itself and restarts `Runner.Listener`.
+- `undefined`: When the exit code of `Runner.Listener` is undefined. It restarts `Runner.Listener`.
 
 The above states are exposed from `/metrics` endpoint as Prometheus metrics. See [metrics.md](metrics.md).