Tim Moloney <moloney@mrsl.com> writes:
> I use a couple of scripts (see below) to backup my subversion
> repositories. Occassionally, I get messages like the following...
>
> Dumping /var/svn/vp-sdw ...
> WARNING: cmp_rev 20 is older than oldest dumped rev 21
> ... loading this dump into an empty repository will fail.
> * Dumped revision 21.
> * Dumped revision 22.
>
> I'm not sure if I should be concerned about the warning. I would
> think that rev 20 would always be older than rev 21 so I don't
> understand what that part of the warning is telling me.
It's warning you (in a highly unhelpful fashion) that something in
revision 21 was a copy of something in revision 20, and since you
started your dump at revision 21, loading that dumpfile into an empty
repository will fail because it will try to perform a copy of
something that doesn't exist.
> Loading all of the incremental backups into an empty repository
> appears to work just fine. I don't see a need to load an individual
> incremental backup into an empty repository but when I did test this
> I got the following...
>
> bash> svnadmin load blah < /var/backup/svn/vp-sdw-r21-r22.inc_dump
> <<< Started new txn, based on original revision 21
> * adding path : trunk/calc_sdw_stats.c ...svn: Filesystem has no
> item
> svn: file not found: revision `0', path
> `branches/rawson/calc_sdw_stats.c'
Yep. I betcha that if you look at the dump record for that addition,
it has:
Node-action: add
Copyfrom-path: branches/rawson/calc_sdw_stats.c
Copyfrom-revision: 20
I think (but am not sure) that the reason your error message shows
revision 0 is because 'svnadmin load' does revision renumbering such
that if your repository is at revision X and you dumpfile starts with
a dump of revision Y, copyfrom revisions (Y - C) are translated to
copies from (X - C) at load time. Since your dumpfile started at 21,
and your next repository revision would have been 1, then a copyfrom
revision of 20 mapped to a copy from revision 0 -- hence the failure.
And, ultimately, hence the warning at dumptime, though it really
should be more clear. I think I might have accidentally broken that
when I (long ago) did a query-replace of copyfrom_rev to cmp_rev
... I might have gotten that printf() unintentionally.
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Received on Fri Aug 29 20:02:59 2003