[Swift-user] Why so few nodes ?

[Ben Clifford]

On Tue, 23 Sep 2008, Xueyuan Zhou wrote:

> also noticed there are some free nodes there. I am wondering why I can only
> have about 8 running job (on 4 nodes), and about 20~30 in Q, while dozons of
> nodes are free.

If you have jobs sitting in the queue on teraport then that is something 
to do with the way that the queueing system is working on teragrid.

I put graphs here:

http://www.ci.uchicago.edu/~benc/tmp/report-test1M-20080922-2203-gamhpzp1/

It looks like your getting around 8 jobs running at once, which matches up 
with your observation in the above quote.

Mihael says this:

> May it be because you're using the "fast" queue, which may have a 
> limitation on the maximum number of nodes you can get at one time?

though the queue policy page here:

http://www.ci.uchicago.edu/wiki/bin/view/Teraport/QueuePolicies

does not mention anything about it.

Though the wiki also contractions this from Mihael:

> How do you reach that conclusion? As far as I can tell, by default, one
> job maps to one node (regardless of the number of cores).

by saying:

> Single-processor jobs submitted by a user are paired up on a single node 
> to maximize cluster efficiency. 

I haven't tried that myself.

--