Debugging Failed Builds

Diagnose and fix failed Buildkite builds.

Load skills: using-buildkite

Workflow

1. Watch the Build

Wait for the build to finish (or start failing):

bk build watch -p pipeline -b my-branch

2. Find Failed Jobs

bk build list -p org/pipeline --branch my-branch --limit 1 -o json | jq '.[0].jobs[] | select(.state == "failed" or .state == "timed_out") | {id, name, web_url}'

3. Read Job Logs

For each failed job, fetch the log and look for errors:

bk job log <job-id> -p org/pipeline -b <build-number> --no-timestamps | tail -n 200

4. Search for Errors

bk job log <job-id> -p org/pipeline -b <build-number> --no-timestamps | grep -i error -C 3

5. Check for Test Failures

If the failed job ran tests, get the test run IDs and hand off to the debugging-failed-tests skill:

ruby scripts/test_runs.rb org/pipeline <build_number>