Actions
action #114769
openHave jobs fail if screen checks take too long, e.g. if there are "two many needles" after warning about it
Start date:
Due date:
% Done:
0%
Estimated time:
Description
Motivation¶
We already have warnings in autoinst-log.txt if screen checks using needles takes too long, e.g. if there are many needles, and screen checks in os-autoinst run into timeouts or stalls but obviously many people do not see or ignore those warnings until jobs fail which in this case might be weird issues until people understand what the actual reason is. So it is likely better to be explicit and fail if there are two many needles after warning about it, e.g. in case we warn about too many needles based on a configurable number/time then if we have 1.5x this number/time (again configurable) we should fail with an explicit error.
Acceptance criteria¶
- AC1: Similar as for warning if a higher threshold than the warning threshold is exceeded we fail the job with an explicit option
- AC2: The setting has a sane default and is configurable
Suggestions¶
- Understand how the current check works in https://github.com/os-autoinst/os-autoinst/blob/1a85ca45aa07515589af919eb93e9fe1429f324b/backend/baseclass.pm#L980
- Extend to die the test with a configurable value with good default
- Optional: Make not simply die based on time in case the running environment has a sporadic slow-down but rather include the list of considered needles, e.g. only fail if "time > X && nr_needles > Y"
Actions