On Thu, 2007-03-08 at 11:02 -0800, Dan Price wrote:
On Wed 07 Mar 2007 at 02:28AM, Timo Sirainen wrote:
http://dovecot.org/releases/dovecot-1.0.rc26.tar.gz http://dovecot.org/releases/dovecot-1.0.rc26.tar.gz.sig
Most importantly this should fix mbox problems in recent RCs.
(Sorry for the long mail)
I've got rc26 up and running on our production box, and we are now migrating users onto it. In a few days we'll see higher user load than we saw in our pilot deployment. So far things look mixed. Some observations:
I'm seeing a lot more imap-login's around than I am used to. Usually there have been just 3; maybe no big deal...? Maybe I made a settings mistake?
# pgrep -x imap-login | wc -l 27 # pgrep -x imap | wc -l 30
Do you have login_process_per_connection=yes? If so, they're probably proxying SSL connections.
#6 0x08087b2f in mbox_sync_rewrite (sync_ctx=0x8047800, mail_ctx=0x0, end_offset=138189881, move_diff=742, extra_space=The value of variable 'extra_space' is distributed across several locations, and GDB cannot access its value. ) at mbox-sync-rewrite.c:566 mails = (struct mbox_sync_mail *) 0x811dc48 offset = The value of variable 'offset' is distributed across several locations, and GDB cannot access its value.
Hmm. I don't think I've seen this assert-crash before. If it's happening often, compiling without -O2 might help more in debugging it.
- Dovecot is still *really* noisy on endianness switchover, and also I think it has been crashy, too:
Looks like I didn't finish fixing the endianess conversion problems. http://dovecot.org/list/dovecot-cvs/2007-March/007997.html should fix them so that it only complains "Corrupted index cache dovecot.index.cache: indexid changed"
Error: IMAP(foobar): file client.c: line 412 (_client_input): assertion failed: (!client->handling_input)
Any chance of getting gdb backtrace from this?
Error: IMAP(foobar): Raw backtrace: 0x80b8fd3 -> 0x806c1da -> 0x8068f2d -> 0x8069169 -> 0x8069222 -> 0x806935f -> 0x806bfd7 -> 0x806bf5e -> 0x806c140 -> 0x80be67d -> 0x80be0b5 -> 0x8072fac -> 0x806722c
If not, this could be used with gdb to get it: http://dovecot.org/list/dovecot/2006-November/017682.html