Looks like once the problem is triggered, the imapd hangs up after
unsuccessfully trying to get an flock() on
"/var/imap/user/x/user.seen". I was able to trigger the problem by
using Netscape Messenger and dragging multiple nested folders
containing lots of messages from "Local Mail" on the client to the
inbox on the server.
Here is strace output for the imapd process up to the point where it
hangs:
18:42:31.889524 open("/var/imap/user/d/deanl.seen", O_RDWR|O_CREAT,
0666) = 18 <0.000030>
18:42:31.889617 fstat64(18, {st_mode=S_IFREG|0600, st_size=3505, ...})
= 0 <0.000006>
18:42:31.889701 old_mmap(NULL, 3505, PROT_READ, MAP_SHARED, 18, 0) =
0x405b2000 <0.000010>
18:42:31.889791 time([993865351]) = 993865351 <0.000005>
18:42:31.889933 getpid() = 2714 <0.000005>
18:42:31.890055 rt_sigaction(SIGPIPE, {0x4027d988, [], 0x4000000},
{SIG_IGN}, 8) = 0 <0.000007>
18:42:31.890149 send(13, "<183>Jun 29 18:42:31 imapd[2714]"..., 88, 0)
= 88 <0.000206>
18:42:31.890421 rt_sigaction(SIGPIPE, {SIG_IGN}, NULL, 8) = 0
<0.000006>
18:42:31.890482 flock(18, LOCK_EX) = ? ERESTARTSYS (To be
restarted) <171.880242>
18:45:23.770822 --- SIGTERM (Terminated) ---
I'm hoping someone will recognize the problem from this information -
the only copy of imapd I have right now is stripped so I can't get a
useful backtrace. If you would like more info, just let me know & I'll
dig into it Monday. By the way, I should probably mention I'm using
kernel 2.4.3.
Thanks for your help!
-Jules
Lawrence Greenfield wrote:
>
> Use strace and gdb to determine what the hung imapd is waiting for.
>
> Larry
>
> Date: Wed, 27 Jun 2001 12:49:47 -0700
> From: Jules Agee <[EMAIL PROTECTED]>
>
> I've been using cyrus imapd 2.0.9-1 from a Red Hat rpm file on a RH
> 7.0 server. I am using LDAP authentication through PAM with
> openldap 1.2.11.
>
> I'm not seeing anything unusual in the logs when this happens, but
> every week or two, one of my 15 test users will report they can't
> access any of their folders. I ran ethereal on the client and it
> appears that imapd just quits responding AFTER the user
> authenticates, and their mail client (Netscape 4.75... I know, but
> I'm stuck with it) basically just sits and spins waiting for a
> response.
>
> That user, and only that user, then cannot access their mail until
> I shut down the master process and re-start it. Users who haven't
> triggered the problem can access their mail fine. I haven't noticed
> any patterns of any particular users triggering the problem more
> often than others.
>
> Any suggestions? Like I said before, the problem is very
> intermittent, but I am concerned that it will be a major problem
> when I drop a few hundred more users on the server. If there is
> anything I can do to get more information about the problem the
> next time it crops up, please let me know. Thanks in advance for
> your help!
>
> --
> Jules Agee
> System Administrator
> Pacific Coast Feather Co.
> [EMAIL PROTECTED] x284
