db corruption after an incorrect 'svn copy'

Hi all,

I accidentally tried to make a copy between two different repositories
(I stayed in an incorrect directory), and after that the target repo's
db stayed in an erroneous state, so I consider it as a bug. (And I hope
it is not a coincidence. :)

The offending cmd was:

svn copy . https://cirmi:8188/another-repo/tags/2005-05-13_v1.1rev422 -m
'Tagging v1.1rev422'

Where the current directory belonged not to the repository what the URL
specified.

The apache log said:

[Tue May 17 10:03:22 2005] [error] [client 100.0.0.218] Could not
MOVE/COPY /svn/!svn/bc/442/orzomain/trunk. [500, #0]
[Tue May 17 10:03:22 2005] [error] [client 100.0.0.218] Unable to make a
filesystem copy. [500, #160012]
[Tue May 17 10:03:22 2005] [error] [client 100.0.0.218] (2)No such file
or directory: Cannot modify transaction named '3g' in filesystem '/
var/local/lib/svn/db' [500, #160012]

After this cmd, every access to this repository resulted in:

[Tue May 17 10:04:37 2005] [error] [client 100.0.0.218] (20014)Error
string not specified yet: Berkeley DB error while opening environment
for filesystem /var/local/lib/svn-kisorzo-bin/db:\nDB_RUNRECOVERY: Fatal
error, run database recovery

After a db4.2_recovery the db got repaired, but for sake, I have loaded
back the one-day old backup.

I didn't try it again, but if you wish, I can manage it.

Bye:
Ferenc

---------------------------------------------------------------------
To unsubscribe, e-mail: users-unsubscribe@subversion.tigris.org
For additional commands, e-mail: users-help@subversion.tigris.org
Received on Tue May 17 12:05:13 2005