[Dovecot] imap-login: Panic: file ioloop.c: line 35 (io_add): assertion failed:
My dovecot directors (v2.0.8) had 3 instances of these yesterday:
imap-login: Panic: file ioloop.c: line 35 (io_add): assertion failed: (fd >= 0)
master: Error: service(imap-login): child 17374 killed with signal 6 (core dumps disabled)
imap-login: Panic: file ioloop.c: line 35 (io_add): assertion failed: (fd >= 0)
master: Error: service(imap-login): child 1780 killed with signal 6 (core dumps disabled)
imap-login: Panic: file ioloop.c: line 35 (io_add): assertion failed: (fd >= 0)
master: Error: service(imap-login): child 10349 killed with signal 6 (core dumps disabled)dovecot "assertion failed" normally scares me quite a bit, but I don't think this can be very bad, since it's only on the directors and we're not doing lmtp here yet.. More of a FYI, and please let me know if I should enable something to get better debug data if/when it happens again.
-jf
On Fri, 2011-04-01 at 09:25 +0200, Jan-Frode Myklebust wrote:
My dovecot directors (v2.0.8) had 3 instances of these yesterday:
imap-login: Panic: file ioloop.c: line 35 (io_add): assertion failed: (fd >= 0) master: Error: service(imap-login): child 17374 killed with signal 6 (core dumps disabled)
Are these still happening? I don't see any obvious way for them to happen.. There were no other errors logged, right? Could you try with the attached patch? Hopefully then it logs where the io_add() is called with wrong fd.
On Mon, Apr 04, 2011 at 04:39:35PM +0300, Timo Sirainen wrote:
On Fri, 2011-04-01 at 09:25 +0200, Jan-Frode Myklebust wrote:
My dovecot directors (v2.0.8) had 3 instances of these yesterday:
imap-login: Panic: file ioloop.c: line 35 (io_add): assertion failed: (fd >= 0) master: Error: service(imap-login): child 17374 killed with signal 6 (core dumps disabled)
Are these still happening?
Yes, seems to be happening every second day or so.. Started without we really changing anything, so it´s a bit strange.
I don't see any obvious way for them to happen.. There were no other errors logged, right? Could you try with the attached patch? Hopefully then it logs where the io_add() is called with wrong fd.
Thanks!
I probably woun´t be able to test the patch before after Easter, but if they´re still happening then, I´ll apply it and report what we see.
-jf
participants (2)
-
Jan-Frode Myklebust
-
Timo Sirainen