Dear Adrian, Floarian, and Nick, On Tue, Nov 15, 2005 at 08:32:44 +0100, Adrian von Bidder wrote: > On Monday 14 November 2005 22.51, Florian Weimer wrote: > > postgrey seems to use proper locking and DB_RECOVERY, so it *should* > > work. But postgrey disables durable transactions: > > > > my $setflags = DB_TXN_NOSYNC; > > > > On some file systems, this adversely impacts the consistency after > > system crashes. It would be interesting to know what happens if the > > line is changed to: > > > > my $setflags = 0; > > David - might that be be a solution? Although the documentation indicates > that setting DB_TXN_NOSYNC shouldn't be dangerous... > > Do you have any tool to drive postgrey with some test input so that I can > quickly get a database of some size by testing? Or perhaps Nick can test > this?
Since it is not so easy to reproduce the problem, the best would be if Nick would test that. I will integrate the change if it has shown to help. Anyway, performance is not such a big issue. Cheers David -- David Schweikert | phone: +41 44 632 7019 System manager ISG.EE | walk: ETH Zentrum, ETL F24.1 ETH Zurich, Switzerland | web: http://people.ee.ethz.ch/dws -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]