On Wed, 5 Dec 2001, Joseph Kezar wrote:
> It seems as though my high load problem is consistent with the
generation of exim -bp -q1h threads.
Do you really mean -bp? Or is that a typo for -bd?
Under usuall circumstances, I
believe, there is only one occurance when I run 'ps ax | grep exim' of
'exim -bp -q1h'. When my top command tells me that 1 or more threads
have reached 99% of cpu > usage 'ps ax | grep exim' yealds more than 1
process with that exact name.
OK, it must be -bd. That starts the daemon. When an incoming call
arrives, the daemon forks a copy of itself to accept the incoming
message. It will show as the same process command. That's why I
suggested you run "exiwhat" to find out what the processes are actually
doing.
When I run /etc/rc.d/init.d/exim stop
all is well. And is well for an undetermined amount of time. Sometimes
upwards of 1/2 hour(at which point I feel my problem has solved itself).
And then it reverts to the same high > load. With one extra spawned
process the load goes from .4 to 1.5. If I let it continue to run it
will soon spawn another process and the load doubles. And I beleive
this goes on until it says no more smtp connections available. > I'm not
possitive about that last sentence though.
What value have you set in smtp_accept_max?
--
Philip Hazel University of Cambridge Computing Service,
ph10@??? Cambridge, England. Phone: +44 1223 334714.
