LSI MegaRAID on Tyan S2469 MB

LSI MegaRAID on Tyan S2469 MB

[Steve Sutphen]

I am having a difficult problem with an LSI MegaRAID 320-2 that is installed
on a Tyan S2469 mother board with AMD Athlon's.  The disks that I am using
are Seagate 147GB drives. 
The problem is that the MegaRAID slows to a crawl, and then slows so much
that it times out and cannot complete the queued commands (for ever it
seems).
I have verified that the controller is working.  I have not been able to 
get it to fail when I move the controller, disk array, and boot disk 
to a different system with a Tyan S2466 motherboard (and dual Athlon's).

Some things that I have tried include: replacing the S2469 motherboard
with another copy of the same.  I have upgraded the BIOS code on the 
motherboard to the latest version V1.07.  I have upgraded the firmware 
on the LSI MegaRAID to the latest version on their web site.  I have 
tried a 2.4.18, a 2.4.23 and a 2.4.34 Linux kernel.  I have tried an
earlier version (I cannot check right now to see which one--what ever
came standard in the 2.4.23 kernel) and have upgraded to the 2.0.9 
version of the megaraid2 driver (I see that a 2.0.10 is now out, I
will try it tomorrow).  I have tried the system with just one CPU.
I have switched the CPU's with the ones on the S2466 based system mentioned
above.  I have replaced all the memory.  I not only have tested the disk
arrays by installing them in the S2466 based system, I have also hooked
them up to the on-board SCSI on the S2469 motherboard, and have
run bonnie++ testing with three of the drives in parallel (i.e. three
bonnie++ jobs in the background).  

My current theory is that there is some sort of incompatibility between
that motherboard (or perhaps the AMD chipset) and the LSI Logic MegaRAID
controller.  

Does anyone have a similar system that is working?  Has anyone encountered
similar problems?  What should I do next?

Thank you.
	steve@cs.ualberta.ca.

PS I have on occasion had the system running for long enough that I could 
complete a test.   When I did I found the performance was not up to 
what I expected.  Sequential writes to a 3-disk RAID-5 array were
14MB/s, re-writes were 7.5MB/s and reads were 48MB/s.  Are these 
reasonable?
	ss.

Re: LSI MegaRAID on Tyan S2469 MB

[Fabien Salvi]

Steve Sutphen a écrit:
> I am having a difficult problem with an LSI MegaRAID 320-2 that is installed
> on a Tyan S2469 mother board with AMD Athlon's.  The disks that I am using
> are Seagate 147GB drives. 
> The problem is that the MegaRAID slows to a crawl, and then slows so much
> that it times out and cannot complete the queued commands (for ever it
> seems).
> I have verified that the controller is working.  I have not been able to 
> get it to fail when I move the controller, disk array, and boot disk 
> to a different system with a Tyan S2466 motherboard (and dual Athlon's).
> 
> [...]

> My current theory is that there is some sort of incompatibility between
> that motherboard (or perhaps the AMD chipset) and the LSI Logic MegaRAID
> controller.  

You're right :)
See below...

> Does anyone have a similar system that is working?  Has anyone encountered
> similar problems?  What should I do next?

Yes, I have exactly the same problems with the same Motherboard and I've 
got the solution !

In my system, I use a ICP-Vortex GDT8514 RAID controller in a Chenbro 
chassis and get segmentation fault, core dumps, kernel panics, etc...

I've contacted Chenbro, ICP-Vortex and Tyan about the problem.
Chenbro seems not in cause.

ICP-Vortex say me to set J92 jumper (force the primary PCI bus to 
operate at 33 Mhz) and it's the solution.

There seem to be a big problem with AMD chipset and 66Mhz bus.

If you look at the Tyan documentation, they say :

<<
Due to the tight PCI 66Mhz timing of the AMD-760 MPX chipset, it may be 
necessary to set the bus to 33Mhz when using Multiple PCI cards.
 >>

ICP-Vortex said :

<<
we've found that the board seems to run stable (at least during our tests)
if J92 is set, if J92 is open similar effects came up.
We've tested the MSI K7D Master which didn't show these effetcs.
 >>

And Tyan :

<<
The problem is still directly with the ICP/Vortex device.  You should be
able to run this card in 66MHz mode with the non-SCSI version of the board.
The AMD chipset is very picky and tricky to program for and it would seem
that ICP/Vortex is still having issues with this.

Now these cards have been tested and found to work correctly without the
problems you are seeing with your particular brand:

AMI Elite 1600
DPT SmartRAID PM1564U3
Mylex AcceleRAID 250
Adaptec 3210s
 >>

But, there is no way to make it work correctly at 66 Mhz.

Good luck !

PS: I'm interested to know if it will correct your problem.



-- 
Fabien SALVI

-
To unsubscribe from this list: send the line "unsubscribe linux-scsi" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html