On Mon, 16 Nov 1998, Dan Bethe wrote:
> > The reason why I'd like to do this in the first place, is that the box
> > involved needs serious uptime/HA, and I don't want to go to all the
> > trouble of RAIDing the system and data disks, just so a physical disk
> > failure hosting swap space can crash the box.
>
> How about just making two separate swap devices (i.e. partitions)
> on each physical disk, and letting the kernel stripe it for you? It just
> automatically does that across multiple swap devices.
> Good idea, keep up the good work :)
This increases performance of the swap space, but that's not my goal.
I want the machine to stay up if a disk fails. If there's swap space in
use on that failed drive, we're going down.
Only solution at this point appears to be to not use swap at all, or
perhaps to swap to files on raided filesystems (haven't tested, and don't
have an available test box atm).
I'm now crossposting to server-linux as well.
Guys, I really need a way for drive failures to not down the host.
Hardware involved is a Compaq Proliant 6500R, with hot-swap SCSI tray,
etc. Linux's RAID support allows me to cover the case of mounted
filesystems, including the / fs, but I can't seem to put swap onto a RAID1
device, so I need some other way of making sure a dead drive won't take
live swapspace with it and hang the box.
Ideas please?
-------------------------------------------------------------------------
Louis Mandelstam # man 8 consulting -o)
email: [EMAIL PROTECTED] System administration contracting /\\
Mobile: +27-83-289-0310 'whois LM497' for current contact info _\_v
---------------------[ Penguin Cloning Specialist ]----------------------
