| I have a pipeline with parallel sequential stages. In one of the sequential stages (that has multiple 'stages'), there was a stage that had a timeout of 20 minutes. That stage took too long, but instead of it being aborted – a different stage had a process killed – a maven build running in a different parallel sequential stream abruptly aborted. Something in parallel stages isn't being tracked properly in the pipeline, I think. once my pipeline gets to the "testing" phase, there are a few parallel stages (some of which have multiple stages), like this: a--b--c
-d-e-
---g---
---h---
---i---- In my case, b had a timeout of 20 minutes. But it ran for 24 with nothing stopping it – the stage is a shell script step running a suite of npm/node.js tests. At the 24 minute mark when it finished normally (but should have been aborted at 20 minutes!!), stage e abruptly had it's maven surefire unit tests abort mid-suite. It's at exactly the same time that stage b finished (which should have aborted earlier) It's as if the code set to kill the processes due to the timeout was delayed ignored by stage "b", and then stage "e" ended up being the victim. |
