stress tests from floppy...

[VJ]

Hello,

Well, the SCSI problems I'm experiencing seem to have gotten worse.

http://forums.murc.ws/showthread.php?s=&threadid=43041

Occasional system freezes, harddrives no longer being found, even in a very basic config (2 harddisk on Channel A).


Tech support is very helpfull so far, but while waiting for their answer I'm trying to locate the problem myself. Are there any tests (cpu-tests, memorytests, ...) I could use to stress my system ? I'd like to make sure it is not a fault processor or a faulty mainboard causing the SCSI mayhem; so the test should preferably be ran from floppy.

I just downloaded memtest86 (version 3), will give it a go later today...


Jörg

[The PIT]

Mmm I thought you were replacing one of the drives????

[VJ]

It is complicated, here goes...

I got errors in the event log (id 9, 11 and 15); occasional crashes. After some testing/swapping, I noticed one harddisk had a false contact (loose connector), causing it to lose power now and then. So this drive has been sent back for replacement... (the connector issue was confirmed in a second computer)

Meanwhile, I bought a new drive (had shortage of storage space anyway), a Seagate Cheetah 10K.6. The system now was configured:
channel A: id 0 Seagate
channel A: id 1 IBM
channel B: id 4 Plextor 4220

However, I got the same errors. Supermicro had some ideas, and at first it seems I didn't have problems with the Plextor removed (they started suspecting the IBM firmware - but I have the correct version and more recently the second U320 channel). Yesterday evening, this changed: with only the Seagate and IBM connected, my system frooze (just like that), and required a reboot. This happened twice, and in one if the occasions I noticed the SCSI-bios did not list the Seagate drive. Non of the freezes yielded particular entries in the log file.

I'm beginning to suspect the SCSI-controller (which is onboard) or the mainboard. I wanted to rule out a bad CPU/memory. Memtest ran a couple of times without errors (but I think it doesn't really use the second CPU).


Jörg

[The PIT]

Sounds like dodgy cabling after you've being pulling all the wires around for a bit when you been checking the drivers previously.
Had a similar problem with scsi cd's and that was the connector on one of the cd drives.

[VJ]

Well, I have tried with the following config:
Channel A, ID 0: Seagate
Channel B, ID 1: IBM
(which implies that the IBM is on another cable)

This yields similar errormessages, some concerning the IBM, others concerning the controller (not sure which channel).

I only have one U320 rated cable; the other one is just a regular 68-pin SCSI cable (40 MB/s or 80 MB/s, not sure). Either way, in this scenario, the IBM was connected with that spare cable...

It is really driving me nuts...


Jörg

[The PIT]

See my edited reply phone kept ringing sorry. Sounds like another duff socket to me.

[VJ]

Originally posted by The PIT
Had a similar problem with scsi cd's and that was the connector on one of the cd drives.

Yes, my Quantum harddisk is sent back for the same reason... Currently however, fiddling with the wires doesn't change anything. Just all of a sudden I get the errors in the log file (and occasionally a freeze).

Just curious: would losing the C-drive cause XP to freeze without any message ?

My next attempt will be with only the Seagate on the onboard channel A; after that, I'll try the disks on another SCSI-controller. I have a 2940UW somewhere; while it isn't U320, it can be used as a host for the disks in SE mode.


Jörg

[The PIT]

Yes losing a drive will cause XP to freeze. Did on my system.

[VJ]

Thanks, our sysadmin wasn't sure about that (he's a Linux-adept... )

Oh, I have logged the voltaged of the PSU, and they seem ok...


Jörg

[The PIT]

Looks like a case of disconnecting the drives one by one until you get a stable running system and then re-introduce then until you find the culprit.

[VJ]

Well, I had the problem with
Quantum + IBM + Plextor
Quantum + IBM
Seagate + IBM + Plextor
Seagate + Plextor
Seagate + IBM



Jörg

[The PIT]

What happens with single drive. Looking more like the controller though. Buy a cheapo one just to see what happens unless you've got a spare.

[VJ]

A cheapo U320 controller...?

Well, I have a 2940UW - this will force the drives in SE mode (they support this) - which I can use. Only problem is that it is in my second rig (60+ km away)...


Jörg

[The PIT]

A cheapo general scsi probably would do. Thats should tell you if the controller acting up or not. Going back to the powersupply I had problems with mine at the Rad 9800 pro. On load one of the drives would start clicking but recording the voltages showed no dip. Replaced the power supply the problem went away. So you never know.

[VJ]

Well, SuperMicro has told me to send in the mainboard for testing; and they accepted my offer of going there with my full rig, so they can truly test whether it is related to disks/mainboard/cabling, ...


So, I know where to go to...


Jörg