On Fri, Apr 16, 2010 at 08:19:26PM -0400, Andersen, Krista wrote:
> I have run into the same "first" sync issue again today. This time it was on a commit that involved over 8000 paths. When I gave the sync command myself I got this error output:
> Transmitting file data .........<truncated - this really ran for over 6000 polka-dots>............svnsync: REPORT of 'http://serverName/parentDirectory/repoName': Could not read response body: connection was closed by server (http://serverName)
>
Do you know what the corresponding log entries on the server look like?
> Does the race condition have more than one way of expressing itself?
>
> Because it has always appeared the same when I find it. Two commits get out of order and this causes svnsync to quit. The rev-prop data is not synced on the two corrupted revs, and this causes any attempt to sync again to return with an error message asking if anyone has committed directly to the mirror. After I remove the corrupt /revs and /revprops files - and edit a couple of the /db files, the sync operation can run again.
>
> This race condition is rare - the fix is a little tricky - but quick enough to be tolerable. Is there any corruption possible from the race condition that I am not yet aware of?
There is no corruption possible other than svnsync meta data getting
out of sync, as you described.
Stefan
--
Sent from my Nokia 1100
Received on 2010-04-17 16:35:01 CEST