Probably the master process has run out of file descriptors. You need a couple for every imap process.

Try "plimit -n 4096 <master pid>" or similar or modify the defaults in /etc/system:-

set rlim_fd_cur=4096 set rlim_fd_max=4096

You might also want the imap processes to continue in this case, in which case set "shutdown_clients = no" in dovecot.conf.

Hope this helps, Chris

Dirk Essl wrote:

Hi,

I am currently migrating a big bunch of mail accounts to a dovecot server running 1.0rc2 on Solaris 10

suddenly all imap processes died and I see this in my dovecot.log:

dovecot: Sep 26 14:54:31 Error: pipe() failed: Too many open files dovecot: Sep 26 14:54:31 Fatal: dup2(stderr) failed: Bad file number dovecot: Sep 26 14:54:31 Error: child 29376 (imap) returned error 89 dovecot: Sep 26 15:15:48 Error: pipe() failed: Too many open files dovecot: Sep 26 15:15:48 Fatal: dup2(stderr) failed: Bad file number dovecot: Sep 26 15:15:49 Error: child 1563 (imap) returned error 89 dovecot: Sep 26 15:15:55 Error: pipe() failed: Too many open files dovecot: Sep 26 15:15:55 Fatal: dup2(stderr) failed: Bad file number dovecot: Sep 26 15:15:55 Error: child 1572 (imap) returned error 89 dovecot: Sep 26 15:15:58 Error: pipe() failed: Too many open files dovecot: Sep 26 15:15:58 Fatal: dup2(stderr) failed: Bad file number dovecot: Sep 26 15:15:58 Error: child 1577 (imap) returned error 89 dovecot: Sep 26 15:18:25 Error: pipe() failed: Too many open files dovecot: Sep 26 15:18:25 Fatal: dup2(stderr) failed: Bad file number dovecot: Sep 26 15:18:25 Error: child 1659 (imap) returned error 89 dovecot: Sep 26 15:18:27 Error: pipe() failed: Too many open files dovecot: Sep 26 15:18:27 Fatal: dup2(stderr) failed: Bad file number dovecot: Sep 26 15:18:27 Error: child 1662 (imap) returned error 89 dovecot: Sep 26 15:18:32 Error: socketpair() failed: Too many open files dovecot: Sep 26 15:18:32 Error: Temporary failure in creating login processes, slowing down for now dovecot: Sep 26 15:18:32 Error: pipe() failed: Too many open files

what could cause the error? is this from Solaris or from dovecot?

TIA dirk

-- --+---+---+---+---+---+---+---+---+---+---+---+---+---+---+---+---+---+- Christopher Wakelin, c.d.wakelin@reading.ac.uk IT Services Centre, The University of Reading, Tel: +44 (0)118 378 8439 Whiteknights, Reading, RG6 2AF, UK Fax: +44 (0)118 975 3094