Sun Fire X2100 SATA problem [was - sun x2100 gmirror problem]

[Miroslav Lachman]

andrej at antiszoc.hu wrote:

> Hi,
> 
> We're using gmirror on our sun fire x2100 and FreeBSD 6.1-p10. Some days
> ago I found this in the logs:
> 
> Apr  1 02:12:05 x2100 kernel: ad6: WARNING - WRITE_DMA48 UDMA ICRC error
> (retrying request) LBA=612960533
> Apr  1 02:12:05 x2100 kernel: ad6: FAILURE - WRITE_DMA48
> status=51<READY,DSC,ERROR> error=10<NID_NOT_FOUND> LBA=612960533
> Apr  1 02:12:05 x2100 kernel: GEOM_MIRROR: Request failed (error=5).
> ad6[WRITE(offset=313835792896, length=4096)]
> Apr  1 02:12:05 x2100 kernel: GEOM_MIRROR: Device gm0: provider ad6
> disconnected.
> 
> Normally it looks like a disk error, but I think our half year old disks
> (WD RE2) shouldn't fail after this short time. Of course they have moving
> parts so they MAY fail. :( Yesterday I tried to reinit the sata channel
> and insert the disk back into the mirror. I got this:
> 
> Apr  3 23:00:32 x2100 kernel: GEOM_MIRROR: Device gm0: provider ad6 detected.
> Apr  3 23:00:32 x2100 kernel: GEOM_MIRROR: Device gm0: rebuilding provider
> ad6.
> Apr  3 23:00:36 x2100 kernel: ad6: WARNING - WRITE_DMA UDMA ICRC error
> (retrying request) LBA=245760
> Apr  3 23:00:38 x2100 kernel: ad6: WARNING - WRITE_DMA UDMA ICRC error
> (retrying request) LBA=392576
> Apr  3 23:00:38 x2100 kernel: ad6: WARNING - WRITE_DMA UDMA ICRC error
> (retrying request) LBA=392960
> Apr  3 23:00:53 x2100 kernel: ad6: FAILURE - device detached
> 
> After this, the disk disappeared from the sata channel completely.
> 
> The wierd is that we used the onboard nvidia-raid and the very same error
> occured, but there was no report in the kernel the machine just don't
> asked for operating system. Later I found out that the disk was forgotten
> ~2 weeks before that reboot (data was ~2 week old on it). Otherwise that
> "forgotten/failed" disk was also half year old and was fine without a
> problem.
> 
> Is there anybody who experienced something similar with SUN X2100 or any
> other servers running FreeBSD 6 and sata?
> 
> Regards,
> Andras

Hi,

I can confirm your problem. I have same problem on one X2100 but not on 
the others. Currenty I have 4 X2100 machines, but only one with this 
strange problem. The problem is not caused by HDD it self, I tried to 
replace it with brand new and same error appears after few days. May be 
there are some problems with cables / connectors or something on mainboard.
I am well known by problems with SATA(n) disk drives problems / 
disappearing on this list and local (czech) mailing list. I had similar 
problems on ASUS boards with Intel chipsets... so in my point of view - 
there is something bad with SATA in general. I never had problem like 
this with old good ATA drives.

I have not solution for this problem. Disk is OK after reboot for a few 
dasy or weeks... if there is somebody which can help with investigating 
this kind of problem, I'll be happy to cooperate.

output of dmesg, smartctl, gmirror etc.:
http://www.quip.cz/1/freebsd/sata-hdd-problems/2007-03-07_errors_ad6.txt

Miroslav Lachman