RE: DB Crash - Wedge
From: Peter Kahn <pkahn_at_connected.com>
Date: 2005-04-13 20:05:43 CEST
Thanks for the information. I have heard this suggestion before. FSFS
Before, I create an implementation/testing plan for this, has anyone
-- Peter Kahn pkahn@connected.com Iron Mountain Digital -----Original Message----- From: kfogel@newton.ch.collab.net [mailto:kfogel@newton.ch.collab.net] On Behalf Of kfogel@collab.net Sent: Wednesday, April 13, 2005 12:28 PM To: Peter Kahn Cc: users@subversion.tigris.org Subject: Re: DB Crash - Wedge "Peter Kahn" <pkahn@connected.com> writes: > Any ideas on what is a reasonable course of action? I was going to > attempt to replicate the situation on a backup copy of my database, > but that does bring in many variables because I cannot get back to the > same state that the db was in prior to the issue. My other > possibility is to write an additional pre-commit hook to analyze the > check in prior to processing and record this information in a log. Well, that would be very helpful for us in debugging, if you spot a pattern. If you just want the problem solved, I suggest switching to FSFS. Subversion's use of Berkeley DB is not quite as the Berkeley docs recommend (I phrase it this way because I don't want to see Berkeley/Sleepycat blamed for what is really Subversion's problem). The bugs are frustratingly idiosyncratic: some people never get them, others get them once in a blue moon, some (like you) apparently get them on a regular basis. For people in that last category, FSFS is probably the way to go. -Karl --------------------------------------------------------------------- To unsubscribe, e-mail: users-unsubscribe@subversion.tigris.org For additional commands, e-mail: users-help@subversion.tigris.orgReceived on Wed Apr 13 20:08:01 2005 |
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.