[gpfsug-discuss] Portability interface

[Simon Thompson]

We've always taken it to mean ..

RHEL != CentOS
7.1 != 7.2 (though mostly down to the kernel).
ppc64le != x86_64

But never differentiated by microarchitecture. That doesn't mean to say we are correct in these assumptions __

Simon

On 22/09/2020, 10:17, "gpfsug-discuss-bounces at spectrumscale.org on behalf of Jonathan Buzzard" <gpfsug-discuss-bounces at spectrumscale.org on behalf of jonathan.buzzard at strath.ac.uk> wrote:


    I have a question about using RPM's for the portability interface on 
    different CPU's.

    According to /usr/lpp/mmfs/src/README

        The generated RPM can ONLY be deployed to the machine with
        identical architecture, distribution level, Linux kernel version
        and GPFS version.

    So does this mean that if I have a heterogeneous cluster with some 
    machines on  Skylake and some on Sandy Bridge but all running on say 
    RHEL 7.8 and all using GPFS 5.0.5 I have to have different RPM's for the 
    two CPU's?

    Or when it says "identical architecture" does it mean x86-64, ppc etc. 
    and not variations with the x86-64, ppc class? Assuming some minimum 
    level is met.

    Obviously the actual Linux kernel being stock RedHat would be the same 
    on every machine regardless of whether it's Skylake or Sandy Bridge, or 
    even for that matter an AMD processor.

    Consequently it seems strange that I would need different portability 
    interfaces. Would it help to generate the portability layer RPM's on a 
    Sandy Bridge machine and work no the presumption anything that runs on 
    Sandy Bridge will run on Skylake?


    JAB.

    -- 
    Jonathan A. Buzzard                         Tel: +44141-5483420
    HPC System Administrator, ARCHIE-WeSt.
    University of Strathclyde, John Anderson Building, Glasgow. G4 0NG
    _______________________________________________
    gpfsug-discuss mailing list
    gpfsug-discuss at spectrumscale.org
    http://gpfsug.org/mailman/listinfo/gpfsug-discuss