Brett Wooldridge wrote:
> I need some help, I think.
>
> We had a user initiate a large import into subversion (when they
> intended to perform
> a checkout). One of the engineers killed the import, which left the DB
> in an inconsistent
> state.
>
> Our build guy is running 'svnadmin recover' on the repository, but I am
> concerned, because
> even though it says it can take quite a few minutes, it's been running
> for over 6 hours now.
> I have no idea whether this is normal or not, and whether we can expect
> it to complete in
> our lifetimes or not. Our repository is ~4.5G in size and has several
> hundred meg of logs.
> Is it replaying *all* of the logs, or just the last few transactions?
> Does subversion 'checkpoint'
> the berkeley DB periodically (or regularly?).
>
> Thanks in advance for your help.
>
>
>
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: users-unsubscribe@subversion.tigris.org
> For additional commands, e-mail: users-help@subversion.tigris.org
>
>
I encountered similar situation several days ago, my repository is
about 100MB, the 'svnadmin recover' command hangs, I use 'top' and
find that it just idled, takes no cpu time. As there is a backup on
my hand, I kill the process. Then I run db_recover in the db directory
of the repository, after that the 'svnadmin recover' finished in 1
minute. And my data did not lost.
But I have to say that I have a backup so I'm dare to kill the recover
process. You should not do that if not.
Could our dear developer confirm if we can kill 'svnadmin recover' in
such situation?
---------------------------------------------------------------------
To unsubscribe, e-mail: users-unsubscribe@subversion.tigris.org
For additional commands, e-mail: users-help@subversion.tigris.org
Received on Sat Feb 28 08:59:47 2004