Kevin Druet <[EMAIL PROTECTED]> writes:
> the load average on that server is quite high right now as we have a
> couple of donated servers for the September 11th cause which are
> sustaining a large number of hits, we are also running an ircd on that
> server.
>
>
> but even still, load averages below 50 are typically ignored, rarely
> do we see any problems with those levels,, possibly its a hardware
> thing ??
Could be. What kernel are you using ?
I'm running a dual PIII-500 with 1Gb RAM, 88 virtual servers (1166 processes)
and the load average is "0.01 0.14 0.19".
There is no easy answer. Try profiling the system.
>
> At 09:52 PM 9/16/2001 +0100, you wrote:
> >we are using 1.4.9 - pre gen skel, the server is doing nothing work wise
> >(about 7gb transfer a month - just 30 IP's) we have mysql running on each VS
> >(about 450 processes in total) but its on red hat (work station arch on this
> >one) and no way would ours would run with a 47 load average - it would just
> >go into a death spiral as all the extra files start to open to keep up
> >The problem seems to be that each VS runs about 20 processes at a stand
> >still and if you calc each one at 2mb and .9% cpu - well it adds up to more
> >than is/could be available if you put say double the VS's on - so we are
> >getting swap used (and iGB ram installed) and load runs wild even if you do
> >the slightest thing on the server, runs out of file descriptors and seg
> >faults etc
If you are suggesting a 0.9% cpu use by all processes, then they are
all running ! Why are they all in a run state ? The whole idea of a
working shared system is that most of the processes are in a sleep state.
------------------------- The freeVSD Support List --------------------------
Subscribe: mailto:[EMAIL PROTECTED]?body=subscribe%20freevsd-support
Unsubscribe: mailto:[EMAIL PROTECTED]?body=unsubscribe%20freevsd-support
Archives: http://freevsd.org/support/mail-archives/freevsd-support
-----------------------------------------------------------------------------
