2003-06-05 04:32:27

by Dan Maas

[permalink] [raw]
Subject: Alpha hang after 24hrs (2.4.21-rc6)

I recently upgraded my Alpha EV56 PC164 system from kernel 2.2.25 to
2.4.21-rc6. With the newer kernel the system hangs reliably after
roughly 24 hours of uptime.

Normally I would suspect heat or RAM failure, but the system seems to
work fine with the older kernel.

The symptom is a total lock-up/freeze, nothing is printed to the
console or syslog. I have not tried a serial console yet but that is
my next step.

The system is a stock 1-CPU 500MHz Alpha, 128MB RAM, nothing special,
just two 3c59x ethernet cards and an AIC788x SCSI controller. It
functions as a web server and NAT gateway. Most of the time it is just
idle. The only unusual thing in the 2.4 kernel configuration is that
I'm using the new QoS packet filtering options (for wondershaper).

This machine has served me reliably for years, and I'd hate to see it
go down for good ;). I really do need the new QoS stuff in 2.4 though.

Any ideas? Please CC to me.

Thanks,
Dan


2003-06-05 04:44:16

by Willy Tarreau

[permalink] [raw]
Subject: Re: Alpha hang after 24hrs (2.4.21-rc6)

On Thu, Jun 05, 2003 at 12:45:57AM -0400, Dan Maas wrote:

> The symptom is a total lock-up/freeze, nothing is printed to the
> console or syslog. I have not tried a serial console yet but that is
> my next step.

I've encountered this with older kernels when I tried a GigE NIC.

> The system is a stock 1-CPU 500MHz Alpha, 128MB RAM, nothing special,
> just two 3c59x ethernet cards and an AIC788x SCSI controller.
^^^^^^^
you should retry with the newer driver from :
http://people.freebsd.org/~gibbs/linux/SRC

Mine has been running 2.4.21-rc3 + this driver for 8 days now without problem,
and there's not much difference between -rc3 and -rc6.

Regards,
Willy

2003-06-06 05:55:09

by Dan Maas

[permalink] [raw]
Subject: Re: Alpha hang after 24hrs (2.4.21-rc6)

> > The system is a stock 1-CPU 500MHz Alpha, 128MB RAM, nothing
> > special, just two 3c59x ethernet cards and an AIC788x SCSI
> > controller.

> Mine has been running 2.4.21-rc3 + this driver for 8 days now
> without problem, and there's not much difference between -rc3 and
> -rc6.

Sorry, it didn't help. 2.4.21-rc6 with Gibbs' aic7xxx driver hung
after about 18 hrs, then again after ~1 hr.

I just loaded up 2.4.20 with Gibbs' driver. We'll see how this goes.

Dan

2003-06-11 03:44:52

by Dan Maas

[permalink] [raw]
Subject: Re: Alpha hang after 24hrs (2.4.21-rc6)

* Dan Maas ([email protected]) wrote:
> I just loaded up 2.4.20 with Gibbs' driver. We'll see how this goes.

Still no luck. I'm betting on hardware failure now. Maybe heat or bad
RAM. (never got an oops record, sorry).

Dan