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

Re: Bug in commiting changes to an fsfs repository?

From: <Mathias.Weinert_at_gfa-net.de>
Date: 2005-08-22 08:55:47 CEST

Philip Martin <philip@codematters.co.uk> wrote:

> Mathias.Weinert@gfa-net.de writes:
> > Guess what happened... Yes, I again had the problem which I mentioned
> > days ago.
> > I have the problem with version 1.2.1 as well as with 1.1.4. But when
> > apply my stuporous
> > little patch I can do the commit without any problem.
> svn_fs_fs__dag_clone_child is documented as allowing a NULL copy_id.
> It passes copy_id to svn_fs_fs__create_successor and that calls
> svn_fs_fs__id_txn_create to get a copy_id if NULL is passed. Your
> change will cause a different code path to get followed but it's far
> from clear that it's correct.

That's exactly what I expected.

> > Next I stepped back to the problematic state of the repository /
> > copy and tried
> > it with yesterday morning's trunk (release 15801) and had the same
> > again.
> >
> > Please don't hesitate to ask me for any additional information you
need to
> > find a solution for the problem.
> Ideally we would want to know how to reproduce the problem on our own
> machines. I suppose one possibility is that your repository already
> has some corruption that is causing the problem. Are you able to dump
> or verify your repository?

Unfortunately I can't tell you when exactly the problem will occur. All I
know is that it happened several times with my repository and that I have
the problem at the moment.
Doing an 'svn up' doesn't help to get rid of the problem. I am able to
dump the repository as well as to verify it.
What could I do if the repository is corrupt (beside exporting it and
importing it again which will lose the history information)?

By the way I am using svnserve with the repository. If you need some
special information I can debug it.

Received on Mon Aug 22 08:56:58 2005

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.