Darwin gitlab-runner occasionally receives SIGQUIT and dies
Seems to be happening about once a month starting in August 2022. It usually happens when the runner is uploading artifacts at the end of a run. Once, though, it happened while uploading the cache at the end of a run.
I guess this is probably a bug in gitlab-runner. My plan is just to treat it as spurious and retry such jobs.
I am opening this ticket just so we have a discussion spot for this topic, should it ever need discussion.
NB gitlab-runner apparently uses SIGQUIT as a signal for a "graceful shutdown", which makes me sad. That's what SIGTERM is already for...?
json->>'web_url' | date(json->>'created_at') | json->>'runner.description' |
---|---|---|
https://gitlab.haskell.org/ghc/ghc/-/jobs/1397503 | 2023-03-08 | aarch64-darwin-1_Mini18-Beta_f87b8a204711 |
https://gitlab.haskell.org/ghc/ghc/-/jobs/1393354 | 2023-03-03 | aarch64-darwin-2_aarch64-darwin-2_a1a3a7139d97 |
https://gitlab.haskell.org/ghc/ghc/-/jobs/1381660 | 2023-02-18 | aarch64-darwin-1_Mini18-Beta_f87b8a204711 |
https://gitlab.haskell.org/ghc/ghc/-/jobs/1348404 | 2023-01-31 | m1-delta_Mini18-Delta_f87b8a204711 |
https://gitlab.haskell.org/ghc/ghc/-/jobs/1303051 | 2023-01-12 | aarch64-darwin-1_Mini18-Epsilon_004e0928d81c |
https://gitlab.haskell.org/ghc/ghc/-/jobs/1275162 | 2022-12-22 | aarch64-darwin-1_Mini18-Epsilon_004e0928d81c |
https://gitlab.haskell.org/ghc/ghc/-/jobs/1261884 | 2022-12-14 | aarch64-darwin-1_Mini18-Beta_f87b8a204711 |
https://gitlab.haskell.org/ghc/ghc/-/jobs/1175376 | 2022-09-19 | aarch64-darwin-1_Mini18-Epsilon_004e0928d81c |
https://gitlab.haskell.org/ghc/ghc/-/jobs/1132694 | 2022-08-06 | aarch64-darwin-1_Mini18-Beta_f87b8a204711 |