-
Notifications
You must be signed in to change notification settings - Fork 14.3k
Upstream libc++ buildbot restarter. #93582
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Conversation
I've been running a cronjob on my local machine to restart preempted libc++ CI runs. This is bad and brittle. This upstreams a much better version of the restarter. It works by matching on check run annotations looking for mention of the machine being shutdown. If there are both preempted jobs and failing jobs, we don't restart the workflow. Maybe we should change that?
@llvm/pr-subscribers-github-workflow Author: Eric (EricWF) ChangesI've been running a cronjob on my local machine to restart preempted It works by matching on check run annotations looking for mention If there are both preempted jobs and failing jobs, we don't restart Full diff: https://github.com/llvm/llvm-project/pull/93582.diff 1 Files Affected:
diff --git a/.github/workflows/restart-preempted-libcxx-jobs.yaml b/.github/workflows/restart-preempted-libcxx-jobs.yaml
new file mode 100644
index 0000000000000..3da17b9f85544
--- /dev/null
+++ b/.github/workflows/restart-preempted-libcxx-jobs.yaml
@@ -0,0 +1,108 @@
+name: Restart Preempted Libc++ Workflow
+
+# The libc++ builders run on preemptable VMs, which can be shutdown at any time.
+# This workflow identifies when a workflow run was canceled due to the VM being preempted,
+# and restarts the workflow run.
+
+# We identify a canceled workflow run by checking the annotations of the check runs in the check suite,
+# which should contain the message "The runner has received a shutdown signal."
+
+# Note: If a job is both preempted and also contains a non-preemption failure, we do not restart the workflow.
+
+on:
+ workflow_run:
+ workflows:
+ - "Build and Test libc\+\+"
+ types:
+ - failure
+ - canceled
+
+permissions:
+ contents: read
+
+jobs:
+ restart:
+ name: "Restart Job"
+ permissions:
+ statuses: read
+ checks: read
+ actions: write
+ runs-on: ubuntu-latest
+ steps:
+ - name: "Restart Job"
+ uses: actions/github-script@60a0d83039c74a4aee543508d2ffcb1c3799cdea #v7.0.1
+ with:
+ script: |
+ const failure_regex = /Process completed with exit code 1./
+ const preemption_regex = /The runner has received a shutdown signal/
+
+ console.log('Listing check runs for suite')
+ const check_suites = await github.rest.checks.listForSuite({
+ owner: context.repo.owner,
+ repo: context.repo.repo,
+ check_suite_id: context.payload.workflow_run.check_suite_id
+ })
+
+ check_run_ids = [];
+ for (check_run of check_suites.data.check_runs) {
+ console.log('Checking check run: ' + check_run.id);
+ console.log(check_run);
+ if (check_run.status != 'completed') {
+ console.log('Check run was not completed. Skipping.');
+ continue;
+ }
+ if (check_run.conclusion != 'failure' && check_run.conclusion != 'cancelled') {
+ console.log('Check run had conclusion: ' + check_run.conclusion + '. Skipping.');
+ continue;
+ }
+ check_run_ids.push(check_run.id);
+ }
+
+ has_preempted_job = false;
+
+ for (check_run_id of check_run_ids) {
+ console.log('Listing annotations for check run: ' + check_run_id);
+
+ annotations = await github.rest.checks.listAnnotations({
+ owner: context.repo.owner,
+ repo: context.repo.repo,
+ check_run_id: check_run_id
+ })
+
+ console.log(annotations);
+ for (annotation of annotations.data) {
+ if (annotation.annotation_level != 'failure') {
+ continue;
+ }
+
+ const preemption_match = annotation.message.match(preemption_regex);
+
+ if (preemption_match != null) {
+ console.log('Found preemption message: ' + annotation.message);
+ has_preempted_job = true;
+ }
+
+ const failure_match = annotation.message.match(failure_regex);
+ if (failure_match != null) {
+ // We only want to restart the workflow if all of the failures were due to preemption.
+ // We don't want to restart the workflow if there were other failures.
+ console.log('Choosing not to rerun workflow because we found a non-preemption failure');
+ console.log('Failure message: ' + annotation.message);
+ return;
+ }
+ }
+ }
+
+ if (!has_preempted_job) {
+ console.log('No preempted jobs found. Not restarting workflow.');
+ return;
+ }
+
+ console.log("Restarted workflow: " + context.payload.workflow_run.id);
+ await github.rest.actions.reRunWorkflowFailedJobs({
+ owner: context.repo.owner,
+ repo: context.repo.repo,
+ run_id: context.payload.workflow_run.id
+ })
+
+
|
@tstellar Once this lands we can revoke the permissions for the libcxx-buildbot-restarter app. |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
LGTM.
I've deleted the restarter app. |
I've been running a cronjob on my local machine to restart preempted libc++ CI runs. This is bad and brittle. This upstreams a much better version of the restarter. It works by matching on check run annotations looking for mention of the machine being shutdown. If there are both preempted jobs and failing jobs, we don't restart the workflow. Maybe we should change that?
I've been running a cronjob on my local machine to restart preempted
libc++ CI runs. This is bad and brittle. This upstreams a much better
version of the restarter.
It works by matching on check run annotations looking for mention
of the machine being shutdown.
If there are both preempted jobs and failing jobs, we don't restart
the workflow. Maybe we should change that?