git review sometimes causes unpacker error for Missing tree

Bug #1332549 reported by Stéphane Duchesneau
20
This bug affects 4 people
Affects Status Importance Assigned to Milestone
git-review
New
Undecided
Unassigned

Bug Description

This is apparently a bug in JGit (implemented in gerrit), but the simplest workaround requires change on the git-review side.

Behaviour:
* Occasionnal failure on "git review" command can prevent push for revs.

### snip ###
git review
error: unpack failed: error Missing tree 99af9575b7587a755868f5c25398acad1a69ef7d
fatal: Unpack error, check server log
To ssh://<email address hidden>:29418/project
 ! [remote rejected] HEAD -> refs/publish/master/somebranch (n/a (unpacker error))
error: failed to push some refs to 'ssh://<email address hidden>:29418/project'
### snap ###

It is related to buggy support of git "thin" optimization in JGit implementation in gerrit.

ref:
https://code.google.com/p/gerrit/issues/detail?id=1582
https://code.google.com/p/gerrit/issues/detail?id=2296

A simple workaround is to use "git push --no-thin", so allowing that option to be configured as default in git-review.conf would allow a workaround.

Revision history for this message
Stéphane Duchesneau (stephane-duchesneau) wrote :

Created a patch (and pull request to upstream)

See it in action here:
<pre>
$ grep disablethinpush ~/.config/git-review/git-review.conf
disablethinpush = true

$ git review -v
[...]
2014-06-20 11:47:22.483602 Running: git push --no-thin origin HEAD:refs/publish/master
[...]

rm ~/.config/git-review/git-review.conf

$ git review -v --no-thin
[...]
2014-06-20 11:55:57.272656 Running: git push --no-thin origin HEAD:refs/publish/master
[...]

$ git review -v
[...]
2014-06-20 11:57:13.375985 Running: git push origin HEAD:refs/publish/master
[...]
</pre>

Revision history for this message
James Byrne (james-byrne-q) wrote :

Although this issue hasn't been updated in over a year, it still affects the latest version of Gerrit (currently 2.11.3). It would be very useful to have this option in the released version of git-review, so is there any chance this patch will be applied to a forthcoming release?

To post a comment you must log in.
This report contains Public information  
Everyone can see this information.

Other bug subscribers