ATA READ command timeout (and worse)

[.VWV.]

Andrew Chan wrote:

> I have been running 2 Seagate ATA66 disks (one on each channel) on a
> HighPoint HPT370 ATA100 controller (FreeBSD 4.7 and then 4.8) for
> more than a year and never had any problems.
>
> I recently built a new server with 2 Maxtor ATA133 80GB disks on a
> Promise Fasttrak 100/TX2 and have been having random timeout problems
> since. Sometime the system would be okay for a few days with no
> problems, even after doing a few "make world" runs. Sometime timeout
> problems showed up only minutes after a fresh install (I had been
> re-installing to try to debug the problem). When timeout occured, the
> ata driver took out the disk from the mirror array, and then either I
> am stupid or something is wierd with the Promise firmware, there just
> doesn't seem to be a good way to re-sync'ing array without major file
> system corruption and the system was toasted.

!Never touch arrays after they have been configured for an installed system!


> Assumming the disks, cables and power supply are all okay, is timeout
> problems on Promise RAID cards still an existing and known issue
> experienced by others?

HighPoint chipset ATA-RAID controllers work fine when they are onboard - as
example, on an Abit AD7-RAID mainboard -. When they are connected to PCI bus
like on Promise boards, it's often a circus. I don't know why, but be happy:
on Linux is much worse.
We all have to choose among the complications and the costs of the U160 or
U320 SCSI at the 'shiny' side, and the apparent simplicity with a lot of
lacks and problems of the ATA UDMA at the 'dark' side. In my opinion, forget
SATA at all, but mine is only an intuition of a
'tecno-perfectionist-sensitive'...

> Andrew

VITTORI