Compaq Proliant 8500 issue with Integrated SMARTArrayRAIDController (ida)

[Andrew Heyn]

-----Original Message-----
From: owner-freebsd-questions at freebsd.org
[mailto:owner-freebsd-questions at freebsd.org]On Behalf Of Ted
Mittelstaedt
Sent: Monday, April 11, 2005 10:08 AM
To: Andrew Heyn; freebsd-questions at freebsd.org
Subject: RE: Compaq Proliant 8500 issue with Integrated
SMARTArrayRAIDController (ida)


owner-freebsd-questions at freebsd.org wrote:
> On Saturday 09 April 2005 05:35, Ted Mittelstaedt wrote:
>> owner-freebsd-questions at freebsd.org wrote:
>>> Hi,
>>>
>>> I have a Compaq Proliant 8500 with the integrated SMART Array RAID
>>> controller.  I recall seeing "Symbios" and "ARM" on a chip on the
>>> center of the PCI module must be the
>>> RAID controller.  I
>>> used to have extreme problems even getting the system to boot
>>> up until I
>>> used the SmartStart
>>> CD and disabled the "Array Accelerator" for my one and only RAID1+0
>>> Container.  (Before doing this) I would get numerous
>>> ida0: soft write error and if the system did manage to boot
>>> up, a process
>>> might read the disk, and
>>> forever be stuck in some kernel routine between userland and
>>> the disk that
>>> gets a block or whatever.
>>> Now, I only get an occasional "ida0: soft read/write error" which
>>> occasionally causes a 15 or so second delay.  The "Array
>>> Accelerator" for the Integrated SMART array controller is 8MB of
>>> read-only cache. Other SMART Array models like the 4200 have
>>> battery backed up cache that can be user-separated between write
>>> and read cache.
>>>
>>> I'm wondering if anybody has ever seen the problem mentioned above.
>>>
>>> I would hate to have to replace the whole
>>> PCI module because of some bad controller ram since that darn thing
>>> is integrated, and would make useless the
>>> internal bays if another raid card was added.  As a note, the
>>> contacts between the hard drive and the drive module
>>> have been cleaned out multiple times for all the drives in the
>>> array.  The connection between the drive module and
>>> the back of the computer is sturdy and clean.  There are only
>>> TWO cables in
>>> this entire system that I know of, and one
>>> is for the IDE CDROM, and one is for the floppy.  So, cabling
>>> cannot be a problem.  I also have two working PSUs that
>>> each have a 120V line going into it, so I doubt it's a lack of
>>> power.  Even though 220V is recommended for both of them,
>>> it works fine with even just one 120V line.
>>>
>>> I asked the HP/Compaq forum and they weren't able to give me
>>> much more of an
>>> answer than "check the cabling" and "blow
>>> off the dust" which I found extremely irritating because the
>>> data is carried
>>> on copper wires that resemble the pins found
>>> on an IDE hard drive or floppy, not your standard "cabling."  I
>>> might ultimately be wrong...but I doubt it.
>>
>> You should ask Windows questions in a Windows forum.  Oh, you aren't
>> running Windows on this system?  Must be FreeBSD 2.2 then, right?
>>
>> Ted
>>
>
>
> Right now, the machine runs FreeBSD 5.3-RELEASE.  I've also
> tried a specially
> modified  6.0-CURRENT kernel upon suggestion of Matthew N.
> Dodd, and FreeBSD
> 4.11, but the nature of the problem never changed and no more useful
> information was able to be found.  The same thing also
> happened running linux
> 2.4.27 and some version of Linux 2.6.  I had given up hope
> that it was a
> software issue and was trying to see if any of the number of
> people on this
> list have ever had a machine that did this, or had details
> about somebody
> else's machine that did the same thing also.  Also, the ROM on
> the controller
> and the primary system BIOS have been updated to the latest available
> versions.  I have also updated the firmwares on the disk drives.
>
>

I can tell you right off the EISA versions of this controller don't
work at all.

Seems to me I recall some discussion a couple years back that there
were problems with certain versions of this controller.  Check in
the mailing list archives, but more importantly check the google
news archives, as I thought I saw the thread on Usenet.

Ted

------------

Ted,

Thanks for the response.

ida0 at pci0:14:0: class=0x010400 card=0x40400e11 chip=0x00101000 rev=0x01
hdr=0x00
    vendor   = 'LSI Logic (Was: Symbios Logic, NCR)'
    device   = 'LSI53C1510 I2O-Ready PCI RAID Ultra2 SCSI Controller
(Intelligent mode)'
    class    = mass storage
    subclass = RAID

I believe that output from pciconf -v -l is enough to show that it's not
EISA...  Does anything
else about that model ring any bells for you?

The only reference to something related to my problem is people reporting
ida0: soft error or
ida0: soft read/write error, which are related to having a failed drive...
I noticed myself that
if the raid container wasn't 100%, those errors would come out by the
thousands...  But nobody reports
temporarily halts, or having to disable the integrated smart array's read
cache to boot up.

Thanks,
Andrew