Re[2]: Whay is broken ATARAID that ignored?

Hello Harald,

Monday, July 12, 2004, 7:09:03 PM, you wrote:

> But only if you replace the drive or  null-out the first sectors. If the drive
> comes back after a reboot (the same drive) then you end up having two raid
> arrays.
> That's not what I call a working raid support!

For all I know that's actually a bug in the controller firmware (there
are a few weirdnesses in HPT controllers, actually, mine sometimes
won't boot after detecting a supposedly degraded array which is
usually not even broken).

>> It IS supported (in stable, anyhow, can't comment about recent
>> CURRENT).

> No, if you create a raid array with the controllers BIOS FreeBSD doesn't care
> about.

With Highpoint and Promise it definitely does. I just installed STABLE on
two different machines. Now Silicon Image is another matter but I
wouldn't bet 1 cent on that crap.

> Also of you Shutdown the machine, power on the machine, modify files,
> power down, reconnect the drive and power up again FreeBSD (and the 
> controller)

Controller's bug. How would FreeBSD know about that? Now of course the
controller should know better but those "software" RAID controllers are
about as dumb as it gets.

> don't recognice the broken mirror and FreeBSD ist panicing!
> Regardless wether it's the vendors fault or the ataraid fault, it
> doesn't work so this should be mentioned!

I know. But rather this than FreeBSD going on to write to a RAID1 array
that's actually out of sync. Nobody in their sane mind will yank out
drives, modify just one and then expect it to still work.

As to documenting it, that wouldn't hurt.


>> Well I do agree the rebuild procedure should be described better.
>> If you want a troublefree (S)ATA RAID solution, nothing beats the
>> 3Ware cards. But cheap they aren't.
> I know and I have one running here,

One of mine lately keeps throwing out one apparently working drive from
my RAID5 array. But for RAID1, they are tops (and the background
rebuild actually works like one would expect it to).

> but that's not a reason to list
> controllers as working when they are not.
> Not to mention the completely broken ataraid in -current.

As for CURRENT, I've stopped caring for it (I lack the time). It's
broken in so many places and won't be fit for production a long time.



Best regards,
 Gabriel