http://issues.apache.org/SpamAssassin/show_bug.cgi?id=4594





------- Additional Comments From [EMAIL PROTECTED]  2006-07-12 06:40 -------
(In reply to comment #38)
> Please attach a copy of your spamd log, complete with time stamps, around the
> time it happens.

Sorry, I didn't read the bug thoroughly enough. --round-robin fixed the issue. 

By the way, we get significantly better performance with --round-robin too. With
roughly the same load on the spamd servers, we can feed in a lot more from the
MX before they start refusing connections due to high load.

> Also, can you do anything to reproduce this?

It were pretty much exactly the same symptoms as described above. also the log
entry was the same.

> Any observations about when it happens?  Does it only happen under heavy 
> load? 
> How heavy?

Well, having the servers run at a load of about 4-6 you can pretty much wait for
it to happen. On 15 servers you wouldn't wait much longer than 15 mins for one
to crash. Chances are, within 4 hours 50% of them are gone.

With --round-robin it didn't happen once :-)

Thanks and greetings...

Stephan



------- You are receiving this mail because: -------
You are the assignee for the bug, or are watching the assignee.

Reply via email to