Hardware cluster requirements

Users/Site Administrators
#1

Hi,

We are exploring use Trinity with HPC and PBS in our lab. This is our hardware:

ServerA: QuadCore CPU and 8Gb RAM
ServerB: 16Cores and 128Gb
ServerC: 32Cores and 128Gb

Can we use ServerA as master and ServerB & ServerC as nodes? In that way the bigger servers can work in parallel and master server just manage the scheduler.

#2

Your plan is spot on,

ServerA: QuadCore CPU and 8Gb RAM – PBS Pro Server/Sched
ServerB: 16Cores and 128Gb - PBS MoM ( compute node 1 )
ServerC: 32Cores and 128Gb - PBS MoM ( compute node 2)

Note: To avoid confusion of the word “server” in this context
PBS Server/Scheduler - distributes or schedules the jobs based on policy/limits etc
PBS Mom (Compute Nodes) – does the computational work

#3

Thanks for the info and the clarification.

Is there any official documentation for OpenPBS? I found a website but is an interpretation of the concepts.

#4

Documentation can be found at this link: Altair PBS Works | Support and Documentation

Please note this is PBS Pro OSS and not OpenPBS, some information below: