* Brian <bmcewen@comcast.net> [2004-08-31 12:31:48]:
> 
> On Tuesday, August 31, 2004, at 02:58 AM, Matt Dainty wrote:
> 
> >On Tue, 2004-08-31 at 04:30, Brian wrote:
> >>On Monday, August 30, 2004, at 06:29 PM, Matt Dainty wrote:
> >>
> >>>currently have figures. I use the Qube entirely as NFS storage and I
> >>>can
> >>>get NetBSD to hang consistently trying to get files of approximately
> >>>500MB-1GB in size on and off the NFS mounts, it'll take about 4GB max
> >>>before locking hard, anything after that is a bonus :-/
> >>
> >>Which NetBSD?
> >
> >Fairly sure it's 1.6.2, at the very least it'll be 1.6.1, I haven't
> >resorted to trying -current on it yet.
> 
> How do you deal with the impact on syslog?  A cron job to clear it ever 
> hour or so?

Most of the time, you get the message followed by "Last message repeated
1748 times", so it doesn't fill up quite as quickly as it might. To be
honest, I think I removed the message from the kernel and recompiled.

At the moment though, I think the SCSI adapter is sat in the one Qube
with Linux running on it where it works without spamming syslog.

Although perhaps we have our wires crossed slightly? I was referring to
1.6.2/1.6.1 as the version of NetBSD that I can lock up with NFS
traffic.

I was wondering if the kernel message "ahc0: Data Parity Error Detected
during address or write data phase" has anything to do with the
PCI_COMMAND_PARITY_ENABLE code snippets I saw crop up in the recent
thread "Initialization of PCI"? I can't really claim to understand much
of what was said, but I just noticed 'parity' crop up in both... *shrug*

Matt
-- 
"Accept anything. Then explain it your way."