Re: [cobalt-users] sendmail crash

[Ed Booher Jr <ebooher@xxxxxxxxxxx>]

Kevin,

How new is this RaQ3i?  My company just installed two new RaQ3i's
and they have been in production roughly two months.  We had a
tough time of it at first where sendmail died on a regular
basis.  Regular enough I wrote a shell script to restart it.  As
we happened to find out, the default configuration sets the time
out of a piece of spooled mail to 60 minutes.  It's possible that
each of those failed 137Kb files were sitting there for an hour
each waiting for their completion.  Check your sendmail
configuration files and make sure that the time outs for spooling
are set to 10 minutes, or less.

Ed Booher
Network Engineer
One Call Internet
http://www.onecall.net/

"If Linux is an 18-wheeler semi, capable of pulling multi-ton
loads cross-country, BeOS is a slick Porsche 911 Turbo."
     - Franco Vitaliano / OPEN Magazine -


[Disclaimer - Any and all views, opinions, outlooks,
philosophies, words of wisdom, words of brash stupidity, and
principles outlined in this post are the belief of the Reverend
Eddie W. Booher, Jr. and are not necessarily synonymous with the
views of his employer or religion.]

Kevin D wrote:
> 
> Hello,
> 
> This is the second time tragedy has struck sendmail on my Raq3i. A user
> tried to send a 137kb attachment, which is well within the file size limit
> set on my raq for attachments. The email was sent by Outlook Express 98,
> under windows 98. It looks like every time Outlook tried to connect to
> sendmail, the connection timed out, so the email stayed in the users outbox,
> and thus the message was resent every 10 minutes or so. Well, the processes
> that sendmail spawned to receive these messages never terminated. Thus,
> there were about 15 sendmail child processes open, and sendmail was shut
> down under "extreme load."
> 
> Well, I telnet'd in and cleaned up all the dead processes, and then
> restarted sendmail. Of course, within 1/2 hour or so, outlook had pumped
> enough instances of this message to almost crash sendmail again... so I had
> to call the user up and have him delete the offending email from outlook.
> 
> Does anyone have any idea why a single attachment might make sendmail freak
> out like this? I went to sendmail.org and found the following known bug:
> 
> * accept() problem on Linux.
> 
>   The accept() in sendmail daemon loop can return ETIMEDOUT.  An
>   error is reported to syslog:
> 
>   Jun  9 17:14:12 hostname sendmail[207]: NOQUEUE: SYSERR(root):
>    getrequests: accept: Connection timed out
> 
>   "Connection timed out" is not documented as a valid return from
>   accept(2) and this was believed to be a bug in the Linux kernel.
>   Later information from the Linux kernel group states that Linux
>   2.0 kernels follow RFC1122 while sendmail follows the original BSD
>   (now POSIX 1003.1g draft) specification.  The 2.1.X and later kernels
>   will follow the POSIX draft.
> 
> However, the sample log entry above did not appear in my messages or
> maillog. I also read about a problem sendmail as with the null (0x00)
> characer (presumably because it uses C strings).
> 
> Has anyone else experienced similar issues? Are there any
> patches/workarounds for this, or do I have to be constantly vigilant for
> these evil little attachments?
> 
> Thanks,
> Kevin
> 
> _______________________________________________
> cobalt-users mailing list
> cobalt-users@xxxxxxxxxxxxxxx
> To Subscribe or Unsubscribe, please go to:
> http://list.cobalt.com/mailman/listinfo/cobalt-users