RE: svnsync fails with unknown error

> svnsync is failing on a 9.2 G repository. The source repo is SVN
> 1.4.5....
>
> Originally, a sync was set up with another repo, also running SVN 1.4.5.
> It eventually failed quietly at rev 12531. This went unnoticed until I
> tried to build an upgrade environment in lenny running 1.5.6 and tried
> to sync the source. It failed at rev 12531. It doesn't give me a lot
> of information to work with:
>
> ".svnsync: Unknown command 'failure' "
>
> The sync has been broken 3 times now on rev 12531.
>
> Subsequent syncs return this result:
>
> svnsync: Path 'QA/Releases/Q2 2.6.3/Q2 Adaptors
> 2.6.3.1617/ProgramBins/Sample/QuorumChangeOfAddress' not present Failure
> synchronizing SVN repo: 1
>
> Figuring that there was something anomalous in the repo, I did a
> hot-copy and verified all the revisions, and then attempted to make the
> hot-copy a sync destination. It did would not establish a sync with it.
>
> So I created an empty repo, did a dump of the orig repo, and a load into
> a repo where the link had been successfully established between the
> source and sync repos. After doing the load, the sync complained that I
> committed data outside of the sync operation.
>
> So what I have is this: A repo that won't sync correctly, and no way to
> force a sync repo to be created from data from a pre-existing repo.
>
> I really need to get this migration done. Does anyone know how to:
>
> 1) determine why svnsync is choking? ( I don't seem to have log files
> to work with... )
> 2) force a pre-existing repo to be used for sync'ing?
> 3) unstick the svnsync operation so that it can complete syncing the
> repo?

I went through the entire repo, committing an empty random folder and incrementing the properties that determined the next rev to sync. There are two sets of revs that seem to trigger the unknown error. The first set included a directory that ended with characters identified by Tortoise as %E2%96%80. As I was unable to rename this directory from Tortoise, I assumed that this was the issue and started looking into how I could clean the data. I think I found a way, unfortunately, I don't find the same kind of 'extra' characters in the second second of revs.

Does anyone have *any* experience that is even remotely similar to this? *at all* ?

I'm literally sitting on the migration to the SVN 1.5.6 because I'd like to make sure the data is as clean as it can be... ... and because I want to know how to prevent svnsync from just failing quietly from similar user interaction in the future...

... please help...

Thanks!

--J.

------------------------------------------------------
http://subversion.tigris.org/ds/viewMessage.do?dsForumId=1065&dsMessageId=2018841

To unsubscribe from this discussion, e-mail: [users-unsubscribe_at_subversion.tigris.org].
Received on 2009-05-01 22:00:47 CEST