Hello.
We have a couple of Exim 3.36 #8 gateways for inbound and outbound mail.
They also have Trend InterScan (3.8-Build_1080) and SpamAssassin. All this
is running on Sun Solaris.
Lately, toward the middle of the day when the load gets high, remote servers
receive response 554 from at least one of the Exim servers causing them to
bounce mail.
We noticed that SpamAssassin grabbed ~31MB of RAM for each spamd process. We
added more RAM, but that didn't help.
We can also see that when the server is loaded, idle CPU is 0.0% for a long
time (yet when we add all the CPU utilization numbers shown in top, they
don't add up to 100%).
Anyway, why is Exim sending a permanent failure response 554 causing an NDR
instead of a temporary failure 4xx? Is there a way to change it?
Also is it Exim doing this or could it be Trend or SpamAssassin sending the
response?
For now I think we will lower the max number of SMTP connections to prevent
the Exim servers from getting overloaded.
Thanks
Andrey Fyodorov
