This is the mail archive of the
gcc@gcc.gnu.org
mailing list for the GCC project.
Re: CVS server deadlocked?
Diego Novillo <dnovillo@redhat.com> writes:
> On Sun, 16 Jun 2002, Jeff Law wrote:
>
> > In message <m3r8j7lggt.fsf@merlin.nerim.net>, Gabriel Dos Reis writes:
> > >
> > > Hi been seing this since this morning, early:
> > >
> > > [...]
> > > cvs server: [02:27:39] waiting for dnovillo's lock in /cvs/gcc/egcs
> > > cvs server: [02:28:09] waiting for dnovillo's lock in /cvs/gcc/egcs
> > > cvs server: [02:28:39] waiting for dnovillo's lock in /cvs/gcc/egcs
> > > cvs server: [02:29:09] waiting for dnovillo's lock in /cvs/gcc/egcs
> > > cvs server: [02:29:39] waiting for dnovillo's lock in /cvs/gcc/egcs
> > > cvs server: [02:30:09] waiting for dnovillo's lock in /cvs/gcc/egcs
> > > cvs server: [02:30:39] waiting for dnovillo's lock in /cvs/gcc/egcs
> > > [...]
> > >
> > > What is up?
> > Probably just a long running CVS operation. Everything seems fine at
> > 9:35am PDT.
> >
> That was a CVS rtag operation that somehow got stuck.
Probably not. The load on sourceware is highest at 2am, because
that's when everyone's batch operations run. It's normal for a tag of
the whole tree to take 20 minutes even when load is light (say at
10am), and high load can double or triple that. My records show the
lock was held for about 40 minutes.
> The logs I have, show my rtag operations fighiting with anoncvs
> access.
That's normal for any large operation.
> I believe that aldyh killed my lock at some point.
That's dangerous if the operation is actually still running. The
usual result is truncated or otherwise corrupted CVS files on the
server.
--
- Geoffrey Keating <geoffk@geoffk.org> <geoffk@redhat.com>