updating the Linux kernel
dunna001 at bama.ua.edu
Fri Jun 9 19:45:50 PDT 2000
Now, I might completly miss something here, but shouldn't all *distibuted*
parallel programs assume that a node may not return. After all, what do you
assume about hardware failures? So, while it may not be a *good* way to do it,
In a properlly paralized application, shouldn't you be able to take down any
random node other than the job allocation node, AT ANY TIME, and have that job
reallocated (Yeah, you lose the local work, but those tasks should be
checkpointed frequently). I just don't think that you should EVER be able to
lose more than 5-10 minutes worth of work on a given node, and if you can, you
should re-examine your program design. So just kill the boxes, and update
them, one at a time.
"Elegant, Documented, On Time; Choose 2"
Email: dunna001 at bama.ua.edu
More information about the Beowulf