Well, I'm not in the central or eastern U.S., I'm on the west coast and
what we've been having here is anything but a heat-wave.
Further; these machines were completely stable (but much slower) on
2.0.35, and I've got the exact same hardware running SunOS 4.1.4 with no
stability problem. In addition the SunOS box has more heat producing hardware
in it as it has a FDDI card in addition to what is in the other boxes.
The machines are only warm to the touch externally and if I pop the case,
the Ross module heat sinks while hot, are not so hot that I can't keep my
finger on them.
I've got a dozen other machines running on the same power and they aren't
having problems. This problem appears only to manifest itself on multi-CPU
boxes. It was bad with 2.2.0 and has gradually gotten better but it's still
a problem with 2.2.10.
I've heard some other people say the Alan Cox patches fix things for them.
I tried to compile with ac12, but at the loading stage get several unresolved
symbols. I've tried 2.2.11pre3, no improvement and it actually seems to be
somewhat worse in terms of effeciency, CPU utilization goes up relative to
2.2.10 and there are some weird load spikes that don't happen with 2.2.10.
On single CPU boxes it seems to be fine.
It is NOT random OOPs, it is only the spin_lock_irqsave() messages that I
posted earlier for me. I know people with PC SMP machines are having a
different problem, this is Sparc hardware here.
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu
Please read the FAQ at http://www.tux.org/lkml/