The integration job discovered its Postgres via docker ps --filter name=integration, which matches every concurrent integration run's service container. A dev push and the main-merge run overlapping on the shared act_runner daemon produced 2 candidates → the "expected exactly 1" guard aborted. That's a concurrency race in the discovery, not a code defect — the released v2026.05.19.0 code is unaffected (a clean, non-overlapping re-run of that job passes).
Fix: scope discovery by network intersection — act_runner attaches the job container and its service container to a shared per-job network, so select the postgres on a network this job container is also on. Dev-compose container still skipped explicitly. Now concurrency-safe (each run finds only its own postgres).
No application/code change; .forgejo/workflows/test-go.yml only.
CI-only hotfix on top of v2026.05.19.0.
The integration job discovered its Postgres via `docker ps --filter name=integration`, which matches **every concurrent integration run's** service container. A dev push and the main-merge run overlapping on the shared act_runner daemon produced 2 candidates → the "expected exactly 1" guard aborted. That's a concurrency race in the discovery, not a code defect — the released v2026.05.19.0 code is unaffected (a clean, non-overlapping re-run of that job passes).
Fix: scope discovery by network intersection — act_runner attaches the job container and its service container to a shared per-job network, so select the postgres on a network this job container is also on. Dev-compose container still skipped explicitly. Now concurrency-safe (each run finds only its own postgres).
No application/code change; `.forgejo/workflows/test-go.yml` only.
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.
CI-only hotfix on top of v2026.05.19.0.
The integration job discovered its Postgres via
docker ps --filter name=integration, which matches every concurrent integration run's service container. A dev push and the main-merge run overlapping on the shared act_runner daemon produced 2 candidates → the "expected exactly 1" guard aborted. That's a concurrency race in the discovery, not a code defect — the released v2026.05.19.0 code is unaffected (a clean, non-overlapping re-run of that job passes).Fix: scope discovery by network intersection — act_runner attaches the job container and its service container to a shared per-job network, so select the postgres on a network this job container is also on. Dev-compose container still skipped explicitly. Now concurrency-safe (each run finds only its own postgres).
No application/code change;
.forgejo/workflows/test-go.ymlonly.