On Wed, Dec 28, 2011 at 01:35:23PM +0100, Wolfgang Breyha wrote: > On 2011-12-28 01:59, Andrew Morgan wrote: > > A quick read of the code makes me think that this only crashes a backend > > imapd process that is performing the mailbox move between partitions. Is > > that correct? > > That is correct. > > > Does this have any effect on the mupdate master (running 2.4.12)? > > mupdate master wont crash if you mean that. But you end up with different > mailbox entries on backend and mupdate. "ctl_mboxlist -m" is needed to get > it in sync again. Even more repairs are required on the backend to fix the > moved mailbox IIRC. And ACLs need to be checked. I'm not sure anymore what > exactly had to be done to repair it completely. But it was PITA enough to > fix it fast;-) Don't know why Bron missed my report/fix for 2.4.11 since it > was already in bugzilla ready for 2.4.12 release. Sorry - I have probably missed a few things along the way! Is there anything still outstanding to add now? Are you happy with the fix that I'm proposing to run with (only push the MUPDATE from the new backend after the DUMP, so that the new backend chooses the servername)? I plan to set up a more complete murder testbed and make sure that all the changes get tested on it as well... part of pulling murder support into Cassandane. Thanks, Bron. ---- Cyrus Home Page: http://www.cyrusimap.org/ List Archives/Info: http://lists.andrew.cmu.edu/pipermail/info-cyrus/