> Nov 2 2009 11:40am from davew @uncnsrd
>Subject: Re: Citadel commit log: revision 7989
>Â
>> Mon Nov 02 2009 11:54:17 AM EST from IGnatius T Foobar @ Uncensored
>>Subject: Re: Citadel commit log: revision 7989
>>
>>
>>My test server now spits out "Server strangled due to machine load
average
>>too high" once per second.
>>
>>
>>
>>
>>
>>
>
>
>
>Hmm, so the load average (as per uptime) divided by the number of
processors
>is 1.00 or higher. Is that right for your test machine?
>
>I was wondering about that.
>
>The idea is to prevent citserver from overloading the machine and
bringing it
>to its knees.
>
>What is the output from uptime and how many processors/cores does it
have?
>
>/* Only start new threads if we are not going to overload the machine */
> if (CtdlThreadGetLoadAvg() < ((double)1.00)) {
>
>This line in threads.c (arround 1363) controls that, probably that
figure
>should be higher or perhaps the method of getting the load average is
bad on
>that box. I used 1.0 because my machine becomes almost unuseable after
1.0.
>
>Â
>
This should really be a configurable parameter for production
environments. It is absolutely possible for a busy mail server to reach
>=1.0
- Stu