Fork failing, any advice?

From: Nikola Milutinovic (Nikola.Milutinovic@ev.co.yu)
Date: Thu Nov 04 2004 - 06:04:46 EST


Hi managers and experts.

I have a dosen Alpha servers acting as Oracle, DNS and E-Mail servers. I
do not have the option to migrate e-mail from the Alphas, so I'm kinda
stuck.

The e-mail system is working in conjunction with AntiVirus software:

- Sendmail 8.12.10 (with Milter API)
- AMaViS Daemon v0.1 (with Milter daemon)
- Sophie v3.02 (daemon using Sophos AV engine)
- Sophos AV for Tru64 UNIX (current - 2 months, hourly updates)

When this works, it works like a charm. We have had a viral infection
and our sub-admins are not verry diligent in cleaning the infected
machines. This means that I do have more viral sending attempt than
usual. On several machines I'm seeing this in the logs:

Nov 4 11:04:15 edzr sophie[1852]: ERROR : fork() failed!
[Operation would block]

Nov 4 11:04:16 edzr sm-mta[29871]: NOQUEUE: SYSERR(root): daemon:
cannot fork: Operation would block

Nov 4 11:04:16 edzr amavisd[19861]: shutdown server - cannot fork

As a result, since sendmail is configured to temporarily fail receiving
mail in case of AV failure, all mail stops. When I get to the machine
and restart AMaViSD and Sophie, it all works again.

What is causing the failed fork() operation?

Machines have more than 70% swap free and max-proc-user is 256, more
than enough for either Sendmail, AMaViS or Sophie to fork. Besides,
Sophie is running as "root" (I know, I know, security...)

Nix.



This archive was generated by hypermail 2.1.7 : Sat Apr 12 2008 - 10:50:10 EDT