Sendmail and Unexpected EOM

[Scott Silva]

on 11/1/2007 2:18 PM Denis Beauchemin spake the following:
> Scott Silva a écrit :
>> I'm getting a lot of collect: premature EOM: unexpected close
>> errors from certain servers, and a google gives me much conflicting info.
>>
>> Running CentOS 4 and current stable MailScanner.
>>
>>  I have tried decreasing the MTU on the external interface, and only 
>> succeeded in stopping all traffic. Tried disabling window scaling, and 
>> that also had no effect. Could the default be to set no-fragment on 
>> packets from CentOS 4?
>>
> Scott,
> 
> I modified these TCP settings in RHEL5 to correct this situation:
> (from /etc/rc.local):
> /sbin/sysctl -w net.ipv4.tcp_wmem="4096 16384 131072"
> /sbin/sysctl -w net.ipv4.tcp_rmem="4096 87380 174760"
> 
> For some unknown reason it wouldn't work in /etc/sysctl.conf...
> 
> Denis
> 
Did you put it in /etc/sysctl.conf without the quotes?
It seems to hold with a sysctl -p. At least it doesn't barf.

-- 
MailScanner is like deodorant...
You hope everybody uses it, and
you notice quickly if they don't!!!!