[Beowulf] motherboards for diskless nodes

John Hearns john.hearns at streamline-computing.com
Sat Feb 26 03:04:47 PST 2005


On Fri, 2005-02-25 at 17:18 -0500, Mark Hahn wrote:
> you means s/node/disk/ right?  sure, but doing raid1 on a "diskless"
> node is not insane.  though frankly, if your disk failure rate is 
> that high, I'd probably do something like intermittently store
> checkpoints off-node.

We recently put in mirrored system disks on a cluster.
The nodes in question are beefy SunFire V40z's, and the cluster is
intended to run long (ie. week long) jobs and the concern was raised
about long jobs failing due to a disk failure.




More information about the Beowulf mailing list