[svn.haxx.se] · SVN Dev · SVN Users · SVN Org · TSVN Dev · TSVN Users · Subclipse Dev · Subclipse Users · this month's index

repository went belly up

From: Merlin Moncure <merlin.moncure_at_rcsonline.com>
Date: 2004-08-25 15:12:05 CEST

Our network admin rebooted the raid server last night and we had a
repository crash the following morning. svnadmin recover just said the
db is hosed and to run db_recover or something like this. I was running
with fsync off on the database, which maybe was night a great idea. We
run frequent dumps though.

The system first gave some cryptic errors followed by a busted db

Os win XP running svnserve/firedaemon
Svn: 1.0.5

Well, the good news is that a svnadmin dump seemed to run without any
problems for all revisions. I'm recreating the repo from the dumpfile
as we speak...and taking the opportunity to update to 1.1 rc2. I
learned my lesson and am running with fsync on, the only reason I turned
it off in the first place was because it made converting the repo take
forever. (aside: it would be nice if svnadmin load did each revision in
a single transaction).

I should be back up in about a hour or so, is there any reason why I
should not trust the dumpfile from the busted repo? If not, I'll just
go back to the previous day's dump and lose a day's worth of commits.


To unsubscribe, e-mail: users-unsubscribe@subversion.tigris.org
For additional commands, e-mail: users-help@subversion.tigris.org
Received on Wed Aug 25 15:13:20 2004

This is an archived mail posted to the Subversion Users mailing list.

This site is subject to the Apache Privacy Policy and the Apache Public Forum Archive Policy.