Author: Vikaas BV Date: To: exim-users Subject: [Exim] System Lockout on running Exim
Hi,
I have been using exim (3.02/3.13) (on Slackware Linux - kernel 2.2.1) in a
multidomain config for about six months now. It was working fine till about
2 months ago. All of a sudden, the system totally hangs (with out being
able to reboot with the reboot command - option of only rebooting with an
unconditional reboot [ shutdown -nr ]). I suspect that the machine is
running out of RAM, but I dont see any log to that effect in the panic log.
What I observe is that a lot of exim processes (I have counted upto 40) get
spawned/forked and then all these processes go into zombie state. Further
to this, the server is unable to start any new process and since it is
remotely managed, the only option left is to ask the colocation service
provider to hard-boot the system. After several such cycles, I have stopped
running the exim server and without it, the server is very stable.
Following is the config of the system
Intel PII 400
128MB RAM
9.1 GB Ultra2 (Adaptec 7890) Adapter and HDD
Intel EtherExpress PRO 10/100 ethernet adapter
Apart from the mail service for 12 domains, the system is also running BIND
8.x and Postgres 6.5.3 public domain RDBMS.
Further, I am using address (with regexp) rewrites (about 8) and a transport
that invokes a Perl program which talks to Postgres DB.
I tried several config options like limiting the number of (exim) forked
processes, cutting down on some domains etc but none seem to work.
I changed the version of Exim from 3.02 to 3.13 and a similar phenomenon
occured again. A surprising aspect is that the server locks out at times in
just about 20 mins.
Please advise.
Vikaas BV
System Administrator
Oyeindia Internetworks Private Limited
No. 54, Richmond Road,
Bangalore - 560025
INDIA
