unpacker error?

135 views
Skip to first unread message

Dan Armbrust

unread,
Jul 13, 2016, 2:41:36 AM7/13/16
to gitblit
Hi, 

I'm not sure if this is a gitblit issues, or somewhere else in the stack involved.

I've recently migrated a few repositories onto gitblit - 1.8.0 - and I also use maven / JGitFlow

When I do a release like this:

mvn jgitflow:release-start jgitflow:release-finish -DuseReleaseProfile=false -DreleaseVersion=1.2.6 -DdevelopmentVersion=1.2.7-SNAPSHOT -DaltDeploymentRepository=vadev::default::https://....

I end up with a failure at the very end - 

... error occurred during unpacking on the remote end: error Missing commit  <commit hash> ...

The end result, is that it doesn't push the master branch updates, nor does it clean up the release branches.  Everything else happens properly - the tag is made / pushed - all of my artifacts successfully make it onto nexus, etc.

Another oddity with this - the commit that it says is missing - is actually present when I go browse gitblit.  

After the failure, I can simply go to eclipse, and push the master branch, and delete the remote release branch - and there are no further errors - everything works properly - as if the release had completely correctly.

I saw your note in the FAQ about the unpacker error, but it doesn't feel like quite the same failure.

This release pattern works for me against github.

I'm waiting for my sysadmin to wake up and check the logs on the server, to see if anything interesting got reported there... but thought I'd post the question in the meantime, and see if this has been seen before.  

Awesome tool / project, by the way.

Thanks,
Dan

James Moger

unread,
Jul 13, 2016, 9:11:55 AM7/13/16
to git...@googlegroups.com
Hi Dan,

I have most often seen the unpacker error with a missing tree message.  But it also manifests as missing delta base and I see no reason why it can't manifest as missing commit.

Since you are testing JGit with three different tools (JGitFlow, Eclipse, & Gitblit) and Eclipse worked it wouldn't surprise me if the EGit team has worked-around the known unpacker error in JGit until a suitable fix is applied.


-J



--
You received this message because you are subscribed to the Google Groups "gitblit" group.
To unsubscribe from this group and stop receiving emails from it, send an email to gitblit+u...@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

Dan Armbrust

unread,
Jul 13, 2016, 10:32:48 AM7/13/16
to gitblit
I'll look at those closer - thanks for the links.  Here are the logs from the server:


2016-07-13 02:37:32 [INFO ] 18 repository models loaded for darmbrust in 14 msecs
org.eclipse.jgit.errors.MissingObjectException: Missing blob 0000000000000000000000000000000000000000
        at org.eclipse.jgit.internal.storage.file.WindowCursor.getObjectSize(WindowCursor.java:167)
        at com.gitblit.models.PathModel$PathChangeModel.from(PathModel.java:181)
        at com.gitblit.utils.JGitUtils.getFilesInCommit(JGitUtils.java:1048)
        at com.gitblit.utils.JGitUtils.getFilesInCommit(JGitUtils.java:985)
        at com.gitblit.service.LuceneService.index(LuceneService.java:648)
        at com.gitblit.service.LuceneService.updateIndex(LuceneService.java:847)
        at com.gitblit.service.LuceneService.index(LuceneService.java:220)
        at com.gitblit.service.LuceneService.run(LuceneService.java:186)
        at java.util.concurrent.Executors$RunnableAdapter.call(Executors.java:511)
        at java.util.concurrent.FutureTask.runAndReset(FutureTask.java:308)
        at java.util.concurrent.ScheduledThreadPoolExecutor$ScheduledFutureTask.access$301(ScheduledThreadPoolExecutor.java:180)
        at java.util.concurrent.ScheduledThreadPoolExecutor$ScheduledFutureTask.run(ScheduledThreadPoolExecutor.java:294)
        at java.util.concurrent.ThreadPoolExecutor.runWorker(ThreadPoolExecutor.java:1142)
        at java.util.concurrent.ThreadPoolExecutor$Worker.run(ThreadPoolExecutor.java:617)
        at java.lang.Thread.run(Thread.java:745)
org.eclipse.jgit.errors.MissingObjectException: Missing blob 0000000000000000000000000000000000000000
        at org.eclipse.jgit.internal.storage.file.WindowCursor.getObjectSize(WindowCursor.java:167)
        at com.gitblit.models.PathModel$PathChangeModel.from(PathModel.java:181)
        at com.gitblit.utils.DiffUtils$DiffStat.addPath(DiffUtils.java:169)
        at com.gitblit.utils.DiffStatFormatter.format(DiffStatFormatter.java:48)
        at com.gitblit.utils.JGitUtils.getFilesInCommit(JGitUtils.java:1052)
        at com.gitblit.utils.JGitUtils.getFilesInCommit(JGitUtils.java:985)
        at com.gitblit.service.LuceneService.index(LuceneService.java:648)
        at com.gitblit.service.LuceneService.updateIndex(LuceneService.java:847)
        at com.gitblit.service.LuceneService.index(LuceneService.java:220)
        at com.gitblit.service.LuceneService.run(LuceneService.java:186)
        at java.util.concurrent.Executors$RunnableAdapter.call(Executors.java:511)
        at java.util.concurrent.FutureTask.runAndReset(FutureTask.java:308)
        at java.util.concurrent.ScheduledThreadPoolExecutor$ScheduledFutureTask.access$301(ScheduledThreadPoolExecutor.java:180)

James Moger

unread,
Jul 13, 2016, 11:09:27 AM7/13/16
to git...@googlegroups.com
Those stacktraces are not from pushing.  Those are from the Lucene indexer.  It is not happy with trying to grab the 0 commit which is usually the parent id of the initial commit of a branch with no history. e.g. a brand new repository or an orphan branch.

-J

--

Dan Armbrust

unread,
Jul 13, 2016, 11:21:45 AM7/13/16
to gitblit
Ok, thanks.  Thought maybe the timezone was off on the server, but I guess it just didn't log any error.

Thanks, 

Dan

Dan Armbrust

unread,
Jul 15, 2016, 8:19:16 PM7/15/16
to gitblit
FYI, I retested this with a newer version of JGitFlow (which uses a newer version of GIT) and the error went away.

Thanks, 

Dan
Reply all
Reply to author
Forward
0 new messages