We've set up a Cyrus 2.0.7 server on a Sun Ultra 220R (dual cpu) and during stress
testing found a problem that seems to be related to Berkeley-DB (both 
versions 3.0.55 and 3.1.17).

The stress test consists of 4 clients on 4 hosts connecting as the cyrus admin
user  to 4 different user.INBOX mailboxes. All clients then
*upload* 50 Megs of messages while creating a mailbox hierarchy.

The core of the client is following perl script:

    $folder = new Mail::Folder('AUTODETECT', $filename) or return;

    $imap->create($mailbox);
    $imap->setacl($mailbox, 'cyrus', 'lrswipcda');
    $imap->setacl($mailbox, $user, 'lrswipcda');
    $imap->select($mailbox);

    $num  = $folder->first_message();
    while ($num != 0) {
        $message = $folder->get_message($num);

        printf "mailbox %s message %d\r", $mailbox, $num if $verbose;

        $message = $folder->get_message($num);
        $imap->append($mailbox, $message->as_string);

        $num = $folder->next_message($num);
    }
    print "\n" if $verbose;
    $imap->DELETEACL("MAILBOX", $mailbox, "cyrus");
    $folder->close;


This code snippet sits in a loop over all "classic" mailboxes for a user
on the client. The code is actually part of a migration tool.

The problem is: Cyrus hangs after a certain time (the period it will
function correctly is rather unpredictable)

The logfile shows the following

Dec 13 17:25:50 mail imapd[12745]: [ID 866726 local6.error] DBERROR db3: 3 lockers
Dec 13 17:25:50 mail imapd[12745]: [ID 866726 local6.error] DBERROR db3: Aborting 
locker 800000a0
Dec 13 17:25:50 mail imapd[12748]: [ID 520099 local6.debug] aborting txn 2147483808l
Dec 13 17:25:50 mail imapd[12748]: [ID 866726 local6.error] DBERROR db3: 2 lockers
Dec 13 17:25:50 mail imapd[12748]: [ID 654676 local6.debug] starting txn 2147483810l
Dec 13 17:25:50 mail imapd[12746]: [ID 866726 local6.error] DBERROR db3: 3 lockers
Dec 13 17:25:50 mail imapd[12746]: [ID 866726 local6.error] DBERROR db3: Aborting 
locker 800000a1
Dec 13 17:25:50 mail imapd[12746]: [ID 520099 local6.debug] aborting txn 2147483809l
Dec 13 17:25:50 mail imapd[12745]: [ID 976968 local6.debug] committing txn 2147483807l
Dec 13 17:25:50 mail imapd[12748]: [ID 866726 local6.error] DBERROR db3: 3 lockers
Dec 13 17:25:50 mail imapd[12747]: [ID 866726 local6.error] DBERROR db3: 3 lockers
Dec 13 17:25:50 mail imapd[12747]: [ID 654676 local6.debug] starting txn 2147483811l
Dec 13 17:25:50 mail imapd[12745]: [ID 654676 local6.debug] starting txn 2147483812l


Has anybody seen this problem before? Or have a pointer where to look?

Regards,
Gert

-- 
------------------------------------------------------------------------------
|     For faculty staff and students: Please do not send e-mail regarding    |
|     day-to-day computer and networking operations to my personal address   |
|                                                                            |
|                     Instead, use:  [EMAIL PROTECTED]                    |
------------------------------------------------------------------------------
Gert Poletiek       [EMAIL PROTECTED]       Phone:      +31 20 525 7503/7463
Computer and Network Group                                  Faculty of Science
University of Amsterdam, Kruislaan 403, NL-1098 SJ  Amsterdam, The Netherlands

Reply via email to