On Fri, Feb 16, 2007 at 02:13:48PM -0800, David Bustos wrote: > Quoth Peter Kovacs on Thu, Feb 15, 2007 at 12:21:39AM +0100: > > I removed the lockfile and cleared the "maintenance" flag on the > > postgresql service which started up properly at this point. > > I haven't followed this thread closely. Is there anything we could do > to make this sort of thing easier to diagnose?
I did follow the thread, and I looked at code, and didn't understand how this actually happened. The lockfile existed when the service was started, the lockfile no longer existed when the postgresql program got to the bottom of its lock loop where it decides it's safe to remove the old lock file, and when the unlink(2) failed it quit. Unless we figure out exactly what happened it's hard to answer your question. Nico --
