| Summary: | Git polling manually triggered; and history | ||
|---|---|---|---|
| Product: | [Technology] Hudson | Reporter: | Steve Powell <zteve.powell> |
| Component: | GIT | Assignee: | Eclipse Webmaster <webmaster> |
| Status: | CLOSED WONTFIX | QA Contact: | |
| Severity: | enhancement | ||
| Priority: | P3 | CC: | eclipse, mikael.barbero |
| Version: | unspecified | ||
| Target Milestone: | --- | ||
| Hardware: | PC | ||
| OS: | Mac OS X - Carbon (unsup.) | ||
| Whiteboard: | |||
|
Description
Steve Powell
Well the polling seems to be working based on the logs for some of the other Virgo jobs this morning. This may be related to bug 323977, where we suspect the culprit is xinetd service limits. We're in the process of correcting that so let us know if that 'resolves' this problem. -M. The Git Polling log is overwritten each time, and the Job logs do not reveal if the polling works or not. We have been clearing the workspace in order to ensure a clean checkout for our proxy testing, so polling hasn't been kicking off the jobs very much, though some may be. Is there a way to force a poll? Also can we have the last few polling log records saved?? (In reply to comment #2) Apologies -- the log does indicate if a poll has kicked off the job -- though this isn't conclusive that the polling has 'worked' correctly. I'll change this to an enhancement request. Enhancement: please provide a way to trigger a poll -- and to see the polling record history. Thank you. If the purpose of polling Git is to launch a build, instead of implementing a way to 'manually trigger a poll' why not just hit the Build Now button? Alternatively, you could commit a blank line... The purpose of polling Git is not to trigger a build, but to check to see if a build needs to be triggered. Moreover, the polling mechanism may be broken, or misconfigured, and being able to 'manually trigger a poll' is a good way to test it. Finally, the polling history is useful when you come in 'in the morning' and discover that the polling you expected to run overnight didn't seem to produce the expected build pattern. Ok that makes sense. However this kind of enhancement sounds like something that should be directed at the Hudson dev team. Or you could try cooking up a groovy script via Hudsons script console to test this. If polling seems to have failed let us know and we'll restart hudson, as that seems to be the only real fix(even if temporary). -M. Moving to hudson as it is probably an enhancement request for them. Since Jenkins has superseded Hudson, and 7ish years have passed I'm going to close this as 'wontfix'. -M. |