DragonFly BSD
DragonFly users List (threaded) for 2005-09
[Date Prev][Date Next]  [Thread Prev][Thread Next]  [Date Index][Thread Index]

Re: More on vinum woes

From: Dave Hayes <dave@xxxxxxxxxxx>
Date: Tue, 13 Sep 2005 07:04:16 -0700

Matthew Dillon <dillon@xxxxxxxxxxxxxxxxxxxx> writes:
> just look at all the bug reports FreeBSD gets related to ATA
> chipsets, and we aren't any better since our ATA code is essentially
> the same.

I've been having trouble with ATA and FreeBSD for years, so I can
confirm this with my own experience. 

> DMA issues and actual data corruption can occur quite easily.  The
> fact that data corruption can occur so easily with these ATA
> chipsets scares me.

If this is true, this might explain a few random crashes on a number
of boxes. Things like "panic ffs_valloc: dup alloc" and "panic
ufs_dirbad: bad dir" on machines that dont do anything tricky.  I
honestly had no idea that it was easy to corrupt data on ATA drives.

As an aside, Is there any hard data about this kind of corruption?

> I say let a 3ware card talk to the SATA drives and let us
> communicate with the 3ware card using sane, well understood and well
> supported protocols, rather then us trying to keep up with the
> billions of different ATA chipsets most of which have hardware that
> is so buggy their vendors have to come out with a new chipset rev
> every few weeks. I have to say, the 3ware cards have made a believer
> out of me.

Since you are personally using this solution, I'm thinking I could do
a lot worse. I need reliability more than anything (which is why I've
gone DragonFly). The way vinum is behaving, maybe I should try this
strategy and see how it works? 

Might I ask the exact model number of the 3ware card(s) you use for
RAID 5? If I do things this way I've got to buy two at the same time,
and I'd like to be accurate. ;)
Dave Hayes - Consultant - Altadena CA, USA - dave@xxxxxxxxxxx 
>>> The opinions expressed above are entirely my own <<<

It is only knowledge that will destroy bias.

[Date Prev][Date Next]  [Thread Prev][Thread Next]  [Date Index][Thread Index]