Promise PDC20269 (Ultra133 TX2) + Software RAID

Promise PDC20269 (Ultra133 TX2) + Software RAID

[Ryan Verner]

Howdy,

I did an upgrade on a system the other day; we went from 2 * 8G drives 
in software RAID1, running off the motherboard's IDE chipset, to 2 * 
80G drives in software RAID1, running off a Promise Ultra133 TX2 card.

I upgraded the kernel at the same time to 2.4.24 w/ grsec patches.  The 
drives are detected fine:

PDC20269: IDE controller at PCI slot 00:0d.0
PCI: Found IRQ 10 for device 00:0d.0
PDC20269: chipset revision 2
PDC20269: not 100% native mode: will probe irqs later
PDC20269: ROM enabled at 0xe5000000
     ide2: BM-DMA at 0xe400-0xe407, BIOS settings: hde:pio, hdf:pio
     ide3: BM-DMA at 0xe408-0xe40f, BIOS settings: hdg:pio, hdh:pio
hde: WDC WD800JB-00ETA0, ATA DISK drive
blk: queue c01a2db8, I/O limit 4095Mb (mask 0xffffffff)
hdg: WDC WD800JB-00ETA0, ATA DISK drive
blk: queue c01a3224, I/O limit 4095Mb (mask 0xffffffff)
ide2 at 0xd400-0xd407,0xd802 on irq 10
ide3 at 0xdc00-0xdc07,0xe002 on irq 10
hde: attached ide-disk driver.
hde: host protected area => 1
hde: 156301488 sectors (80026 MB) w/8192KiB Cache, CHS=9729/255/63, 
UDMA(100)
hdg: attached ide-disk driver.
hdg: host protected area => 1
hdg: 156301488 sectors (80026 MB) w/8192KiB Cache, CHS=9729/255/63, 
UDMA(100)

However, we get these sorts of errors often:

hdg: dma_timer_expiry: dma status == 0x22
hdg: error waiting for DMA
hdg: dma timeout retry: status=0x58 { DriveReady SeekComplete 
DataRequest }
hdg: status timeout: status=0xd0 { Busy }
PDC202XX: Secondary channel reset.
hdg: drive not ready for command
ide3: reset: success

And the machine is randomly locking up, and of course, on reboot, the 
raid array is rebuilt.  Ouch.  Any clues as to why?  I'm sure the hard 
drive hasn't failed as it's brand new; I suspect a chipset 
compatibility problem or something.

R

--

Signature space for rent.

Re: Promise PDC20269 (Ultra133 TX2) + Software RAID

[Ryan Verner]

On 03/02/2004, at 2:08 PM, Ryan Verner wrote:
> And the machine is randomly locking up, and of course, on reboot, the 
> raid array is rebuilt.  Ouch.  Any clues as to why?  I'm sure the hard 
> drive hasn't failed as it's brand new; I suspect a chipset 
> compatibility problem or something.

Definitely seems to be this.  Swapped the drives back over to the 
onboard-IDE chipset, which is much slower (raid rebuilds at only 
7MB/sec instead of 25), but certainly none of these problems.

Known issue?

R

--

Signature space for rent.

Re: Promise PDC20269 (Ultra133 TX2) + Software RAID

[Pasi Kärkkäinen]

On Tue, Feb 03, 2004 at 04:42:15PM +1030, Ryan Verner wrote:
> On 03/02/2004, at 2:08 PM, Ryan Verner wrote:
> >And the machine is randomly locking up, and of course, on reboot, the 
> >raid array is rebuilt.  Ouch.  Any clues as to why?  I'm sure the hard 
> >drive hasn't failed as it's brand new; I suspect a chipset 
> >compatibility problem or something.
> 
> Definitely seems to be this.  Swapped the drives back over to the 
> onboard-IDE chipset, which is much slower (raid rebuilds at only 
> 7MB/sec instead of 25), but certainly none of these problems.
> 
> Known issue?
> 

I'm running Ultra-133 TX2 with software RAID successfully without any
problems. This is on (mainly) Intel P3 platforms.

-- Pasi Kärkkäinen
       
                                   ^
                                .     .
                                 Linux
                              /    -    \
                             Choice.of.the
                           .Next.Generation.

Re: Promise PDC20269 (Ultra133 TX2) + Software RAID

[Gunnlaugur Thor Briem]

Ryan Verner wrote:

> On 03/02/2004, at 2:08 PM, Ryan Verner wrote:
>
>> And the machine is randomly locking up, and of course, on reboot, the 
>> raid array is rebuilt.  Ouch.  Any clues as to why?  I'm sure the 
>> hard drive hasn't failed as it's brand new; I suspect a chipset 
>> compatibility problem or something.
>
>
> Definitely seems to be this.  Swapped the drives back over to the 
> onboard-IDE chipset, which is much slower (raid rebuilds at only 
> 7MB/sec instead of 25), but certainly none of these problems.
>
> Known issue?


Yep: http://bugzilla.kernel.org/show_bug.cgi?id=1888

I'm getting this with the Promise 20376. Presumably not a hardware 
problem: I can't reproduce it under Windows XP with Promise's supplied 
Windows driver, and Jeff Garzik calls it a known bug -- "should have a 
fix soon".

Jeff, is it "known" in the sense that you know roughly what goes wrong, 
but just haven't gotten around to fixing it? I'd like to tinker and try 
to fix it, but am "unfamiliar around those parts".

    - Gulli

Re: Promise PDC20269 (Ultra133 TX2) + Software RAID

[Pedro Larroy]

On Tue, Feb 03, 2004 at 02:08:31PM +1030, Ryan Verner wrote:

> And the machine is randomly locking up, and of course, on reboot, the 
> raid array is rebuilt.  Ouch.  Any clues as to why?  I'm sure the hard 
> drive hasn't failed as it's brand new; I suspect a chipset 
> compatibility problem or something.
> 
> R
> 

I have similar issues with 20269. I have two cards on one box doing sw raid5
on 6 ide drives. It only runs stably with 2.4.19
It has been many months since I assembled that box, and I've tried kernels
from 2.4.20-ac, 2.5.x, 2.6.2  and all hang after some time running.

I remember that pdcs also hanged a dual processor box.

Regards. Pedro.

Re: Promise PDC20269 (Ultra133 TX2) + Software RAID

[Andreas Steinmetz]

Pedro Larroy wrote:
> On Tue, Feb 03, 2004 at 02:08:31PM +1030, Ryan Verner wrote:
> 
> 
>>And the machine is randomly locking up, and of course, on reboot, the 
>>raid array is rebuilt.  Ouch.  Any clues as to why?  I'm sure the hard 
>>drive hasn't failed as it's brand new; I suspect a chipset 
>>compatibility problem or something.
>>
>>R
>>
> 
> 
> I have similar issues with 20269. I have two cards on one box doing sw raid5
> on 6 ide drives. It only runs stably with 2.4.19
> It has been many months since I assembled that box, and I've tried kernels
> from 2.4.20-ac, 2.5.x, 2.6.2  and all hang after some time running.
> 
> I remember that pdcs also hanged a dual processor box.
> 

In my case (see my mail to lkml today) I do suspect concurrent disk 
access and IO-APIC to be the culprit. If you're using an IO-APIC try 
booting with either "noapic" or "hdx=serialize" where hdx is one of the 
disks of your controller card.
-- 
Andreas Steinmetz

Re: Promise PDC20269 (Ultra133 TX2) + Software RAID

[Bartlomiej Zolnierkiewicz]

On Wednesday 04 of February 2004 22:22, Andreas Steinmetz wrote:
> Pedro Larroy wrote:
> > On Tue, Feb 03, 2004 at 02:08:31PM +1030, Ryan Verner wrote:
> >>And the machine is randomly locking up, and of course, on reboot, the
> >>raid array is rebuilt.  Ouch.  Any clues as to why?  I'm sure the hard
> >>drive hasn't failed as it's brand new; I suspect a chipset
> >>compatibility problem or something.
> >>
> >>R
> >
> > I have similar issues with 20269. I have two cards on one box doing sw
> > raid5 on 6 ide drives. It only runs stably with 2.4.19
> > It has been many months since I assembled that box, and I've tried
> > kernels from 2.4.20-ac, 2.5.x, 2.6.2  and all hang after some time
> > running.
> >
> > I remember that pdcs also hanged a dual processor box.
>
> In my case (see my mail to lkml today) I do suspect concurrent disk
> access and IO-APIC to be the culprit. If you're using an IO-APIC try
> booting with either "noapic" or "hdx=serialize" where hdx is one of the
> disks of your controller card.

Try "noapic" first, "hdx=serialize" will kill you RAID performance.

---bart

Re: Promise PDC20269 (Ultra133 TX2) + Software RAID

[Pedro Larroy]

On Wednesday 04 February 2004 23:06, Bartlomiej Zolnierkiewicz wrote:
> On Wednesday 04 of February 2004 22:22, Andreas Steinmetz wrote:
> > Pedro Larroy wrote:
> > > On Tue, Feb 03, 2004 at 02:08:31PM +1030, Ryan Verner wrote:
> > >>And the machine is randomly locking up, and of course, on reboot, the
> > >>raid array is rebuilt.  Ouch.  Any clues as to why?  I'm sure the hard
> > >>drive hasn't failed as it's brand new; I suspect a chipset
> > >>compatibility problem or something.
> > >>
> > >>R
> > >
> > > I have similar issues with 20269. I have two cards on one box doing sw
> > > raid5 on 6 ide drives. It only runs stably with 2.4.19
> > > It has been many months since I assembled that box, and I've tried
> > > kernels from 2.4.20-ac, 2.5.x, 2.6.2  and all hang after some time
> > > running.
> > >
> > > I remember that pdcs also hanged a dual processor box.
> >
> > In my case (see my mail to lkml today) I do suspect concurrent disk
> > access and IO-APIC to be the culprit. If you're using an IO-APIC try
> > booting with either "noapic" or "hdx=serialize" where hdx is one of the
> > disks of your controller card.
>
> Try "noapic" first, "hdx=serialize" will kill you RAID performance.
>
> ---bart

That box doesn't have APIC. just XT-PIC   :(

-- 
  Pedro Larroy Tovar  |  piotr%member.fsf.org 

Software patents are a threat to innovation in Europe please check: 
	http://www.eurolinux.org/