Re: Squid hangs weekly on SUN Ultra

Andreas Strotmann wrote:
>
> On Mar 30, 9:59, Dancer wrote:
> > Subject: Re: Squid hangs weekly on SUN Ultra
> > > Ermm, no. If the process cannot be killed, then your operating system
> has blown
> > > up. Squid _may_ have contributed to that, but ultimately, it ain't
> > > squid's fault, IMO.
> > >
> > > Your machine obviously has deeper problems, and any program that
> applies
> > > that level of network/disk/cpu workload is going to break it. Again,
> > > IMO.
> > >
> > > D
>
> My thinking exactly. But how do you prove that it's simply something that
> is bound to happen with the kind of load we want the machine to handle?
> In other words, how do you tell your management that it needs to invest
> to meet that kind of demand? After all, this machine should really be
> able to handle 2Mbit/sec, shouldn't it?
>
> Sun Microsystems appears to be baffled, too. They can't explain what's
> happening, and they blame it on Squid --- an SOP (someone else's
> problem)...

Hmph. A user-process causing the operating-system to hang. It's a
solaris bug, and no mistake.

My last job, we invested heavily in Sun hardware. Sun hardware is good.
But we got very sick of the bugs in the solaris 2.5 IP drivers. Very
VERY sick. Ultimately we started replacing the OS with whatever we could
run on them (FreeBSD at the time), and mothballing the ultra's and
netra's until linux or BSD caught up with them.

Maybe 2.6 fixes this. I dunno, but I recall weary times going over
bug-lists that seemed longer than the feature list and wondering "why".
Nevertheless, the hardware is good. All it needs is a 'rocking' OS to
put on it.

D
Received on Mon Mar 30 1998 - 04:50:17 MST