>This is a rather strange crash. What kind of request rate and bandwidth is
>this box serving? You might just be overloading it (though Squid should be
>more forthcoming about why it is keeling over). Usually the problem of
>overload becomes apparent some time before a crash, though, and errors also
>usually result (could not allocate memory is a common one for an overloaded
>Squid--even if plenty of memory is available, I guess fragmentation of the
>available memory leads to an inability for the OS to free up enough
>contiguous memory to answer Squids allocation).
>
>I haven't seen any random sig 6 crashes in my boxes, but we've only
>recently upgraded everyone to the Squid 2.4 series from
>2.2STABLE5+hno...maybe I just haven't begun to hear about them yet.
Hello,
The problem is, the load is never all that high when it does this. I have
had this box going at 110+ reqs/sec, feeding 550K/sec without a problem, and
it will die at 50 reqs/sec or less. It does not seem to have to do with
load. Only once have I ever had a malloc error in the cache.log or syslog.
I agree that I am pushing the box quite a bit, however I have semi-decent
hardware to do it too. But there does not seem to be a critical point where
squid fails, and it will mostly do it early in the day before it hits high
loads.
Ron
_________________________________________________________________
Get your FREE download of MSN Explorer at http://explorer.msn.com/intl.asp.
Received on Tue Mar 19 2002 - 15:46:50 MST
This archive was generated by hypermail pre-2.1.9 : Tue Dec 09 2003 - 17:07:00 MST