Jeremy Hansen wrote:
>
> Hmm...ok well let me explain what I'm doing.
>
> Basically we have the file server, which is currently running
> 4 9 gig drives in a raid 0/1 configuration on the Adaptec U2W.
> The web servers which basically mount the raid volume via nfs
> surround the machine.
>
> If I can, I'd like to eliminate the need to use nfs from the
> web servers and just share the device across the machines.
>
> What it sounds like you're saying is, it is possible as far
> as the hardware is concerned, but the code isn't ready.
>
> Read only in this case wouldn't really be an options, but
> wouldn't locking take care of this somehow?
Nope. What happens when machine A reads from the Raid array, machine B then
writes to the portion that was just read by machine A, then another process on
machine A wants to read that same spot again. Machine A reads the information
directly out of cache instead of checking the disks and voila, you just got
stale information. There is no way around this currently in Linux. This
would require either a userspace sync daemon that worked amongst all machines
on the bus, or a kernel level solution.
> Raid Inc. wants $10,000 for their "solution". I was trying
> to use what we already have.
--
Doug Ledford <[EMAIL PROTECTED]>
Opinions expressed are my own, but
they should be everybody's.
-
To unsubscribe from this list: send the line "unsubscribe linux-scsi" in
the body of a message to [EMAIL PROTECTED]
