You are viewing a plain text version of this content. The canonical link for it is here.
Posted to users@subversion.apache.org by Jack Repenning <jr...@collab.net> on 2009/07/06 18:09:56 UTC

What am I doing wrong? (Base checksum mismatch on commit)

I have a cronjob that runs every two hours: checkout, modify, commit,  
nuke working copy. Around once a day, it encounters something like this:

> svn: Commit failed (details follow):
> svn: Base checksum mismatch on '/trunk/www/overrides/snippets/ 
> ViewIssueTab4Att-Top.html':
>   expected:  6623a0052da1ca124331e936f9eb8c76
>     actual:  cf7c74932d25593d94369e5b15c7080b

The next run (which always uses a clean working copy) always works fine.

There's long history (http://snurl.com/m8c9z) in this list and dev@  
alleging that this is because of some sort of corruption in the  
working copy, which I suppose is consistent with my experience that  
the next co/mod/ci is always OK. But I can't quite imagine what I  
might be doing that would cause such corruption, especially since it's  
the same script every time, on a pristine, for-purpose working copy.

Looking at the script, I'm reminded that I did something modestly  
tricky, back last December, right at the point of the commit that  
always encounters this error, and yes, it does seem rather as if the  
problem began about then, though my time-sense here has some pretty  
large error bars. What I do is this:

	svn stat |
	cut -c5- |
	sort -r |
	xargs -n100 svn commit -F svn-commit.log --depth empty

Is there something wrong with such an approach?



-==-
Jack Repenning
Chief Technology Officer
CollabNet, Inc.
8000 Marina Boulevard, Suite 600
Brisbane, California 94005
office: +1 650.228.2562
twitter: http://twitter.com/jrep

------------------------------------------------------
http://subversion.tigris.org/ds/viewMessage.do?dsForumId=1065&dsMessageId=2368467

To unsubscribe from this discussion, e-mail: [users-unsubscribe@subversion.tigris.org].

RE: What am I doing wrong? (Base checksum mismatch on commit)

Posted by Jack Repenning <jr...@collab.net>.
On July 18, I changed my process to store the working copy in local file space, and I have not seen this problem since. Whereas it was happening once or twice a day before (in a job that runs 12 times a day), there are zero incidents in over a month since.

In my case at least, this seems pretty convincingly to argue that the issue is something about the NFS remote storage of the working copy. Whether that's my workstation's NFS client implementation, or the file server's NFS server implementation, or some known or allowed variation in the NFS protocol itself, who knows, but clearly the first thing to try, if you're seeing this symptom, is moving your working copies local.

------------------------------------------------------
http://subversion.tigris.org/ds/viewMessage.do?dsForumId=1065&dsMessageId=2391153

To unsubscribe from this discussion, e-mail: [users-unsubscribe@subversion.tigris.org].