prod-202104-27: Fix stall problem in pipeliner runner in environments with nice>0
sawenzel
released this
27 Apr 06:41
·
2390 commits
to master
since this release
This is fixing a stall problem in the pipeline runner, which was observed on systems (GRID), where the runner was launched with an increased nice value (>0). In this scenario, allocated resources per task were not correctly "un-accounted' and so the runner stalled at some moment since it couldn't find free resources. This commit is also adding some debugging mechanism with webhooks. Later integration into the standard logging system would be nice.