--- Nick Ustinov <[EMAIL PROTECTED]> wrote:
> well, it seems that this problem is only experienced
> by redhat users. i
> played with different kernels, same result. what was
> cyrus ver that worked
> fine for you? i tried 2.0.9 - it was just like
> 2.0.12.
Our server was running 1.5.21 first (I don't know
exactly since I wasn't here at that time), 1.6.24
without LMTP then, 1.6.24 run without a problems, the
load would go to over 20 sometimes, the system was
overall slower , but no deadlocks. Now the avarage
load is much lower, but the problems happen.
This is a redhat 6.2 now I think.
Helmut
> nick
>
> -----Original Message-----
> From: [EMAIL PROTECTED]
> [mailto:[EMAIL PROTECTED]]
> Sent: Wednesday, April 25, 2001 19:55
> To: [EMAIL PROTECTED]
> Cc: [EMAIL PROTECTED]
> Subject: forking problem cd..
>
>
> Hello fellow listmembers,
>
> We switched to a new version of the cyrus imapd
> (version 2.0.12) and i think
> we have encounted the 'locking problem' that is
> being discussed here.
>
> What we are seeing here is that it takes a long
> while for the greetings to
> show even after the connection has been established.
> A whole lot of tracing
> and debugging turned up that the process is waiting
> to get access to the
> mailboxes.db (and related files like __db* and
> log.*). Carefull reading and
> backtracking through the sourcecodes show that the
> problem probably is
> located in the db3 lib somewhere. Following is a
> piece of the strace:
>
> select(0, NULL, NULL, NULL, {1, 0}) = 0
> (Timeout)
> select(0, NULL, NULL, NULL, {1, 0}) = 0
> (Timeout)
> select(0, NULL, NULL, NULL, {1, 0}) = 0
> (Timeout)
> select(0, NULL, NULL, NULL, {1, 0}) = 0
> (Timeout)
> select(0, NULL, NULL, NULL, {1, 0}) = 0
> (Timeout)
> select(0, NULL, NULL, NULL, {1, 0}) = 0
> (Timeout)
> open("/var/imap/db/log.0000000002",
> O_RDWR|O_CREAT|O_LARGEFILE, 0644) = 5
> fcntl(5, F_SETFD, FD_CLOEXEC) = 0
> _llseek(5, 9426640, [9426640], SEEK_SET) = 0
> write(5,
>
"\225\326\217\0a\316\"E;\0\0\0\215\0\0\0~5\0\200\0\0\0\0"...,
> 59) =
> 59
> fsync(5) = 0
>
>
> After adding some debug code in the pop3d we found
> out that the problem has
> to be somewhere in the mboxlist_open function
> (mboxlist.c:1707) most likely
> in the DB->open call. I suspect that the select in
> the strace is called by
> the function __os_sleep in the sleepycat DB code.
>
> We tried several things with amounts of prespawed
> processen and limiting the
> amount of concurrent lmtpds etc. The best effective
> was disabling postfix
> completely. Everything ran smoothly, unfortunately
> it is not really an
> option. Possibly the problem is related to the lmtpd
> code somehow. By the
> way, the proposed 'singeinstancestore' solution did
> not change anything
> here.
>
> I hope this information is helpfull to anyone who is
> researching this
> problem. If anyone has made any progress in solving
> this problem i would
> greatly appreciate a posting on the list or to
> private email. I have to make
> a descision whether we keep trying to solve this
> problem or downgrade to the
> old mailserver again.
>
> For the record, our setup:
> hardware:
> Compaq DL380
> 512Mb mem
> Dual-PIII 933
> 6x18.2 Gb disks (smartarray 2x3disks in raid5)
> software:
> RedHat 6.2
> Linux 2.4.2(smp)
> Sleepycat DB 3.2.9
> Cyrus IMAPD 2.0.12
> Cyrus SASL 1.5.24
> PostFix 19991231-pl13
>
> Regards,
>
> Hugo
>
> --
>
--------------------------------------------------------------
> "That i'm paranoid doesn't mean they aren't out to
> get me!"
>
--------------------------------------------------------------
> Hugo Trippaers (HT2-6BONE) ISION Internet BV
> System Engineer (RHCE) http://www.ision.nl
>
> [EMAIL PROTECTED]
__________________________________________________
Do You Yahoo!?
Yahoo! Auctions - buy the things you want at great prices
http://auctions.yahoo.com/
