Currently only oe-core project automatically updates patches state to Accepted when they're merged. This feature should be enabled for all projects in patchwork.
OE's git repositories are using a git-hook older than the asynchronous one that is used only for the oe-core project. All repositories will be configured to use the same script, since this script seems to be causing the lack of updates for those projects. All of other project configurations have been verified and no difference was found between them. Even manual state updates can be made.
All of the OE's git repositories were configured to use the asynchronous githook in Dec 13. Waiting for confirmation that the automatic updates are working as expected.
Logs show that the githook is working for "oe" and "bitbake" projects as well as "oe-core". The other projects should also be updated upon future merging. Please let me know if other behavior is detected.
Today I was pushing some new changes to master and master-next, everything looked OK (push was much faster probably because of asynchronous hook you mentioned before). It wasn't showing any output from the hook, so I assumed everything is OK, but now some people on IRC are reporting missing branches and indeed master and master-next branches are gone: OE @ ~/meta-openembedded $ git remote prune origin Pruning origin URL: git://git.openembedded.org/meta-openembedded * [pruned] origin/master * [pruned] origin/master-next refs/remotes/origin/HEAD has become dangling! And I cannot just repush them: OE @ ~/meta-openembedded $ git push origin master Everything up-to-date OE @ ~/meta-openembedded $ git push origin master-next Everything up-to-date Can you please check what went wrong on the server and if it's related to the new hook or the recent infra update?
interesting the same happened in -contrib repository which probably doesn't have this hook The push looked normal: OE @ ~/meta-openembedded $ git push -f contrib jansa/master Counting objects: 38, done. Delta compression using up to 8 threads. Compressing objects: 100% (38/38), done. Writing objects: 100% (38/38), 12.83 KiB | 0 bytes/s, done. Total 38 (delta 22), reused 0 (delta 0) To git+ssh://git.openembedded.org/meta-openembedded-contrib fcb6d27..6cdb9f6 jansa/master -> jansa/master but the branch is gone now: http://git.openembedded.org/meta-openembedded-contrib/log/?h=jansa/master
(In reply to comment #4) > Today I was pushing some new changes to master and master-next, everything > looked OK (push was much faster probably because of asynchronous hook you > mentioned before). > > It wasn't showing any output from the hook, so I assumed everything is OK, > but now some people on IRC are reporting missing branches and indeed master > and master-next branches are gone: > > OE @ ~/meta-openembedded $ git remote prune origin > Pruning origin > URL: git://git.openembedded.org/meta-openembedded > * [pruned] origin/master > * [pruned] origin/master-next > refs/remotes/origin/HEAD has become dangling! > > And I cannot just repush them: > OE @ ~/meta-openembedded $ git push origin master > Everything up-to-date > OE @ ~/meta-openembedded $ git push origin master-next > Everything up-to-date > > Can you please check what went wrong on the server and if it's related to > the new hook or the recent infra update? I believe this is something not related to the script discussed on this bug entry but either Jose or Michael have more insight. Martin, I followed your steps and I saw no pruning on my local branches. ---- snip ---- $ git remote prune origin $ git remote -v origin git://git.openembedded.org/openembedded-core (fetch) origin git://git.openembedded.org/openembedded-core (push) ---- snip ---- Any other step that I am missing?
(In reply to comment #6) > (In reply to comment #4) > > Today I was pushing some new changes to master and master-next, everything > > looked OK (push was much faster probably because of asynchronous hook you > > mentioned before). > > > > It wasn't showing any output from the hook, so I assumed everything is OK, > > but now some people on IRC are reporting missing branches and indeed master > > and master-next branches are gone: > > > > OE @ ~/meta-openembedded $ git remote prune origin > > Pruning origin > > URL: git://git.openembedded.org/meta-openembedded > > * [pruned] origin/master > > * [pruned] origin/master-next > > refs/remotes/origin/HEAD has become dangling! > > > > And I cannot just repush them: > > OE @ ~/meta-openembedded $ git push origin master > > Everything up-to-date > > OE @ ~/meta-openembedded $ git push origin master-next > > Everything up-to-date > > > > Can you please check what went wrong on the server and if it's related to > > the new hook or the recent infra update? > > I believe this is something not related to the script discussed on this bug > entry but either Jose or Michael have more insight. > > Martin, I followed your steps and I saw no pruning on my local branches. > > ---- snip ---- > $ git remote prune origin > $ git remote -v > origin git://git.openembedded.org/openembedded-core (fetch) > origin git://git.openembedded.org/openembedded-core (push) > ---- snip ---- > > Any other step that I am missing? I believe it is very unlike for the githook to mess with branches, however, I'll ask Michael to take a look at the server and see what he can find.
> Any other step that I am missing? different project, try it in meta-openembedded(-contrib)
This hook is not causing the problem with unavailable branches. It is a umask problem with gitolite and the git daemon.
I'm closing this bug since the branches issue is not related. Let me know if I should file a new bug to track that.