Pod Cluster

Status
Yes

Documentation for the Pod cluster is to the left (under the 'pod usage' menu), with some basic info below.

Key things to know about Pod (versus Knot)

  • It uses the Slurm queueing system, so you'll have to slightly rewrite your PBS job files!
  • Each regular node has 40 cores, so you might want to adjust how many cores you ask for (i.e. most people used 12 on knot, but with more powerful cores, you might not even want an entire 40 core node)

At the moment, you login with 'ssh pod-login1.cnsi.ucsb.edu'. Don't yet use the name 'pod.cnsi.ucsb.edu' as that may put you randomly on one of two login nodes, and we only have login1 configured at the moment.

You can copy files that you need directly from knot, or central at /csc/knot or /csc/central

x2go (or putty, of course) is a good way to login. Or Putty.

Pod uses a different queueing system 'slurm' so you'll have to adjust how you run jobs, but it's only minor changes. We will move knot and braid to slurm during the fall.

Related Links

  • Using X2GO for GUI Login