RE: megaraid: ABORTING problem

RE: megaraid: ABORTING problem

[Mukker, Atul]

 > In trying to do some performance analysis on my LSI Login 
> SATA-150-6, I
> 
> megaraid: ABORTING-188cf36 cmd=2a <c=0 t=5 l=0>
> megaraid: ABORTING-188cf36[52], fw owner.
>
You seem to be running a older version of the driver. Can you upgrade to the
latest 2.10.3 or 2.20.x.x and re-test. Also, what is the firmware version on
this controller.

-Atul Mukker

RE: megaraid: ABORTING problem

[Paul Wagland]

On Tue, 2004-04-13 at 17:09, Mukker, Atul wrote:
>  > In trying to do some performance analysis on my LSI Login 
> > SATA-150-6, I
> > 
> > megaraid: ABORTING-188cf36 cmd=2a <c=0 t=5 l=0>
> > megaraid: ABORTING-188cf36[52], fw owner.
> >
> You seem to be running a older version of the driver. Can you upgrade to the
> latest 2.10.3 or 2.20.x.x and re-test. Also, what is the firmware version on
> this controller.
> 
> -Atul Mukker

Ah. Guess I had better get to work adding in sysfs support to the new
unified driver then ;-)

This was the 2.6.4 kernel driver, and from /proc/hba0/config I get:

Version =712T:G116, DRAM = 64Mb

I was running these tests as part of a baseline to see what, if any,
performance differences there would be with the new unified driver.
Should this be fixed in that new driver?

Thanks in advance,
Paul

RE: megaraid: ABORTING problem

[Bagalkote, Sreenivas]

>Please teach me how to build
>the latest unified driver(2.20.B1) by 2.6.5.

1. mkdir -p drivers/scsi/unified/beta
2. Extract the driver files into the directory
3. From that directory, run 
	make -C /usr/src/linux SUBDIRS=$PWD modules

>Is there any update patch from 2.00.3 to 2.20.B1?

2.20.B1 is a new effort. It doesn't have any incremental
patch from an older driver.

Sreenivas

RE: megaraid: ABORTING problem

[Tomita, Haruo]

> The new drivers have the new error handling, so your server 
> should still be
> operative if commands start to abort.

Please teach me how to build
the latest unified driver(2.20.B1) by 2.6.5.
Is there any update patch from 2.00.3 to 2.20.B1?
--
Haruo

RE: megaraid: ABORTING problem

[Mukker, Atul]

> I was running these tests as part of a baseline to see what, if any,
> performance differences there would be with the new unified driver.
> Should this be fixed in that new driver?
The new drivers have the new error handling, so your server should still be
operative if commands start to abort.

megaraid: ABORTING problem

[Paul Wagland]

Hi all,

In trying to do some performance analysis on my LSI Login SATA-150-6, I
set it off running bonnie against multiple filesystems against multiple
stripe sizes and RAID levels. Well, all was fine for about three days (I
was running a lot of tests ;-) and then I get back to my machine today
to discover that it was locked up, and the following was on the console:

megaraid: ABORTING-188cf36 cmd=2a <c=0 t=5 l=0>
megaraid: ABORTING-188cf36[52], fw owner.
megaraid: ABORTING-188cf37 cmd=2a <c=0 t=5 l=0>
megaraid: ABORTING-188cf37[51], fw owner.
megaraid: ABORTING-188cf38 cmd=2a <c=0 t=5 l=0>
megaraid: ABORTING-188cf38[37], fw owner.
megaraid: ABORTING-188cf39 cmd=2a <c=0 t=5 l=0>
megaraid: ABORTING-188cf39[3e], fw owner.
megaraid: ABORTING-188cf3a cmd=2a <c=0 t=5 l=0>
megaraid: ABORTING-188cf3a[3], fw owner.
megaraid: ABORTING-188cf3b cmd=2a <c=0 t=5 l=0>
megaraid: ABORTING-188cf3b[65], fw owner.
megaraid: ABORTING-188cf3c cmd=2a <c=0 t=5 l=0>
megaraid: ABORTING-188cf3c[c], fw owner.
megaraid: ABORTING-188cf3d cmd=2a <c=0 t=5 l=0>
megaraid: ABORTING-188cf3d[1e], fw owner.
megaraid: ABORTING-188cf3e cmd=2a <c=0 t=5 l=0>
megaraid: ABORTING-188cf3e[57], fw owner.
megaraid: ABORTING-188cf3f cmd=2a <c=0 t=5 l=0>
megaraid: ABORTING-188cf3f[b], fw owner.
megaraid: ABORTING-188cf40 cmd=2a <c=0 t=5 l=0>
megaraid: ABORTING-188cf40[5d], fw owner.
megaraid: ABORTING-188cf41 cmd=2a <c=0 t=5 l=0>
megaraid: ABORTING-188cf41[5e], fw owner.


Does anyone know what is the cause of this failure?

I can see in the code that this comes out of the
megaraid_abort_and_reset() function, and this is in turn called by
megaraid_abort() and megaraid_reset(). These in turn appear to be called
from the SCSI layer? But why? And why did the machine hang?

Thanks in advance for any help,
Cheers,
Paul