Re: cvs2svn.py corrupting repository

"Peter Howard" <pjh@coastal.net.au> writes:

> Hi all
>
> (Note: this is using ver 0.15)
> I have been trying cvs2svn.py on a work CVS repository. Given the comments
> in the release, plus the mail here, I was not expecting it to do a full
> conversion (_lots_ of branching).

Yes, it should work just fine, many people use it.

> I was only mildly surprised when it dumped core. I _was_ surprised
> that the repository it created was corrupt.

What does "corrupt" mean? You need to be a lot more specific.

> Should I be surprised? I assume it has to do with the fact that
> it's working directly on the repository (not via http). Does this
> mean any of the direct-access tools (eg. svnadmin) are at risk of
> doing this?

Any process that uses libsvn_fs directly (svnadmin, svnlook,
mod_dav_svn, cvs2svn.py, or libsvn_ra_local) is making Berkeley DB
calls, and thus at risk of leaving a lockfile sitting in the Berkeley
DB area if suddenly interrupted (control-C, segfault, etc.) If this
happens, you run 'svnadmin recover' to have Berkeley DB return the
database to a consistent state. Is that what you mean by "corrupt"?
Did you recover the datbase after cvs2svn coredumped?

---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Received on Wed Nov 27 13:57:33 2002