On Tue, 18 Apr 2000, David Holl wrote:
> These are multiprocessor machines? Have you tried running on only one
> CPU? (vaguely recalling messages many months ago about multiprocessor
> troubles)
No, haven't tried that. But I just might, even though we need the extra power
from a second CPU...
See, funny(?) thing is, some of the servers (we have sth like 15 of 'em), work
perfectly alright, never freeze, while others freeze up occasionally (once a
week or something like that).
This is what makes this so annoying. It's rather tedious sitting around
waiting for a box to freeze up when it does so once every 10 day or so.
But, anyway, I'll check all cables etc, just to make sure. Disks might differ
slightly - not sure about that, although all are 9 GB IBM Ultrastar disks.
BTW, all disks are internal (three disks in each machine), and the SCSI cables
are those U2W cables that come with the P2B-DS motherboard, same in all
machines.
Thank you all for all input. I'll hook up some extra logging, check all
the cabling and get back to you if I manage to make out what's causing this.
Ciao!
/m
>
> On Tue, 18 Apr 2000, Mikael Eriksson wrote:
>
> -On Tue, 18 Apr 2000, Chris Mauritz wrote:
> -
> -> Are you SURE it's not a cabling issue? I've had 2940U2w cards act strangely
> -> both under Linux and NT when there were problems with cables or terminators.
> -> I've gotten into the habit of using SCA drive cages and keeping the LVD
> -> cable lengths to a minimum (just between the cage and the controller).
> -> Also, make sure you have active (not passive) termination.
> ->
> -> Cheers,
> ->
> -> Chris
> -
> -Yes, I'm quite sure, since we've had this problem on at least 5 or 6 machines
> -(all with identical hw-configs).
> -
> -What I've furthermore seen, is that those servers without striped disks never
> -have frozen up.
> -
> -On the other hand, not all striped servers have crashed (yet).
> -
> -
> -/m
> -
> -
> ->
> -> ----- Original Message -----
> -> From: <[EMAIL PROTECTED]>
> -> To: <[EMAIL PROTECTED]>; <[EMAIL PROTECTED]>
> -> Sent: Tuesday, April 18, 2000 12:31 PM
> -> Subject: adaptec 2940u2w hangups
> ->
> ->
> -> > Hi All!
> -> >
> -> > Lately, we have been experiencing some serious problems with our Linux
> -> servers
> -> > using RAID0 on Adaptec 2940U2W. The machines, which are under quite some
> -> load,
> -> > suddenly dies and must be cold-restarted. When they get back online again,
> -> > there's is no sign of anything going awry in any logfile. The just plunge
> -> into
> -> > deep-freeze, zero-Kelvin mode. *argh*
> -> >
> -> > Currently, the machines are running Linux 2.2.14 with latest raid-patches
> -> > (Mingo's raid-2.2.14-B1-patch), but we've seen the problem under 2.2.13 as
> -> > well.
> -> >
> -> > As I said, there's nothing in the log files that would indicate what's
> -> wrong.
> -> > Installing the software watchdog kernel module/watchdogd didn't help
> -> either.
> -> >
> -> > The situation is getting somewhat embarrassing, as we've been pushing
> -> pretty
> -> > hard towards Linux. We're considering moving all servers to non-RAID
> -> > configurations, but we'd really prefer RAID0.
> -> >
> -> > I've also noticed a few other postings about problems/hangups with
> -> 2940/AIC79xx
> -> > on Linux RAID, so it seems we're not alone with this problem.
> -> >
> -> > Does anyone have any kind of information as to the status of this. Is the
> -> > bug(s) identified? Is there a solution (other than stop using RAID)?
> -> >
> -> >
> -> > Hardware setup: RH Linux 6.1/2.2.14/raid-2.2.14-B1 on dual PIII
> -> motherboards
> -> > (ASUS P2B-DS) and U2W SCSI IBM disks, 512+ MB RAM.
> -> >
> -> >
> -> > /m
> -> >
> -> >
> -> >
> -> >
> -> >
> ->
> ->
> -
> -
> -/m
> -
> -
> -
>
>
/m
