On Saturday 06 December 2008 19:52:10 Rick Chisholm wrote:
> there might be a common denominator here - this is happening on a Dell
> SC1435, dual-core Opteron 2212... maybe it's an AMD thing. That may
> change my plans for another server.
I seriously doubt it's an AMD thing.
Since it only happened rarely I figured it was probably an unfortunate
coincidence of 2 events.
Assuming random timing, the shorter the events, the less likely they are to
overlap and of course the faster the CPU and RAM, the shorter the events are
likely to be.
Plus; I have 32 external acls. That's only 8 per core and each core is faster
so the startup ought to be much faster.
If anyone actually knows about this stuff feel free to interrupt me anytime
you like...
Hmmm, just noticed a strange thing, none of the external acls are on CPU0.
>
> ian j hart wrote:
> > On Tuesday 02 December 2008 14:40:19 Rick Chisholm wrote:
> >
> > Sorry I can't offer a solution but I can confirm this as an issue.
> >
> > I used to get this on our old work hardware;
> > CPU AMD athlon 2100+ (single core at 1733Mhz IIRC) PC3200 DDR.
> >
> > Since upgrading (Q6600 quad core at 2.4Ghz, DDR2 667Mhz?) I'm not getting
> > this any more. [Rubs lucky rubber chicken to avoid the kiss of death].
> >
> > This suggests it may be a subtle timing issue which will be tricky to
> > find.
> >
> > I got into the habit of running top after the reconfig. If one of the
> > squid owned processes hits 100% CPU, kill with extreme prejudice.
> > Actually I still do this.
> >
> > IIRC it was the old squid process which did this, but I could easily be
> > wrong. If you can confirm that it may help one of the developers to find
> > the problem.
> >
> > HTH
-- ian j hartReceived on Sat Dec 06 2008 - 20:48:41 MST
This archive was generated by hypermail 2.2.0 : Sun Dec 07 2008 - 12:00:02 MST