| Summary: | Buildhistory fails to write to repo on new cluster | ||
|---|---|---|---|
| Product: | [Infrastructure] AutoBuilder | Reporter: | Stephano Cetola <stephano> |
| Component: | autobuilder | Assignee: | Stephano Cetola <stephano> |
| Status: | RESOLVED FIXED | QA Contact: | |
| Severity: | normal | ||
| Priority: | Medium+ | CC: | infras.ab.watcher, Infras.watcher, joshuagloe, leonardo.sandoval.gonzalez, pidge |
| Version: | unspecified | ||
| Target Milestone: | 2.4 M2 | ||
| Hardware: | x86 | ||
| OS: | Multiple | ||
| Whiteboard: | |||
| OS type for building Yocto: | --- | Type of Regression: | --- |
| Verified: | Documentation change: | No (bug/feature does not impact docs) | |
| Bug Depends on: | |||
| Bug Blocks: | 11814 | ||
|
Description
Stephano Cetola
2017-01-14 00:20:52 UTC
I believe there are two issues here: 1) there's a configuration issue in that the autobuilder configuration points to a non-existent repository as the buildhistory remote 2) when the buildhistory push fails the buildstep isn't marked as failed *** Bug 11334 has been marked as a duplicate of this bug. *** Too late to patch AB for 2.3 at this point, let's punt to 2.4 M1 I changed the configs on all the slaves and for some reason had to restart the controller to get the right variables to populate. I'm not sure what was going on there, but I'll keep an eye on the CreateAutoConf steps to be sure this is fixed. So far that step has completed on debian-testing just fine with nightly-arm. The bit that was failing was the git ls-remote: https://autobuilder.yocto.io/builders/nightly-arm/builds/400/steps/CreateAutoConf/logs/stdio (git ls-remote --exit-code file:////home/pokybuild/buildhistory) Note that BUILD_HISTORY_DIR was set correctly there after I updated the config, but was being ignored for some reason. I restarted the controller and it seems to work now: https://autobuilder.yocto.io/builders/nightly-arm/builds/401/steps/CreateAutoConf/logs/stdio (git ls-remote --exit-code ssh://git@git.yoctoproject.org/poky-buildhistory) One thing we have to decide is if we want this to be a hard error in the future. Currently it does not fail because, as mentioned, the git push is buried in an if statement with no error handling around it, so it simply never falls into the "then" clause. See CreateAutoConf.py around line 266. I'll leave this as "in progress review" until we have time to discuss how we want to deal with these types of failures. |