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

RE: Always reproducible base checksum mismatch error

From: Philipp Wollermann <philipp_at_igowo.de>
Date: 2007-11-22 12:58:51 CET

Hi John,

John Szakmeister wrote on 2007-11-22:
> Can you run 'svnadmin verify' on the repository? You're likely to get
> the same error. If so, then you may be the first person I've met who
> can consistently cause this fsfs corruption error. Changing to mpm-
> prefork makes the issue go away, but I'm not entirely sure why.

Unfortunately (?) I don't get this error on svnadmin verify:

$ svnadmin verify test
* Verified revision 0.
* Verified revision 1.
* Verified revision 2.
* Verified revision 14.
* Verified revision 15.

There is some strange behavior:
I used my stress testing shell-script "exercise". It does a checkout of the repo, appends a line to a file and commits the changes. Somehow the working copy gets corrupted now after the second or third commit. If try to repeatedly commit again now, it won't work again ever. "svnadmin verify" on the repository yields no error. *However:* If I delete my working copy, checking it out from the server again and repeating (appending a line, committing), it commits fine, but the working copy breaks again after sometimes one, two or three commits.

It seems like the repository is okay (as long as svnadmin verify can tell) but the working copy gets corrupted.

> Is there anything else special about your setup? A latent or lossy
> link to the repository? What kind of machine are you hosting the
> repository on?

The server is a Core 2 Duo E6600 (2,4 GHz) with 4 GByte DDR2-RAM. It has a fast, low latency connection to the internet (100 MBit/s).
On this server I use OpenVZ to create virtual machines. The repository and its webserver are in one of these vservers (there are at the moment only two vservers running, so no congestion of resources happens).
I test directly on the local machine (kind of http to localhost), but the error also occurs over longer links like from my home or the university. I checked the resource-limits, but there was no exceeding of the resource-limits during this test. The Apache (it's 2.2.6, I remembered it wrong last mail) which serves the SVN also hosts some webspaces with mod_fcgid for PHP and mod_wsgi for Python.
I compiled all software myself, but without using any fancy optimization flags (I guess -g -O2 as is the default).

If you need any more information, please let me know, I'd be glad if we can solve this problem together. :)


To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Received on Thu Nov 22 13:00:20 2007

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

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