Topas not running - Aix

This is a discussion on Topas not running - Aix ; I get the following message when I try running topas: root>topas SpmiCreateStatSet can't create StatSet root> What can be causing this?...

+ Reply to Thread
Results 1 to 3 of 3

Thread: Topas not running

  1. Topas not running

    I get the following message when I try running topas:
    root>topas

    SpmiCreateStatSet can't create StatSet
    root>




    What can be causing this?


  2. Re: Topas not running

    Not sure at all, but have a look on /dev/mem and /dev/kmem and the given
    rights - I think topas can't read this two devices

    wilsjamie@gmail.com schrieb:
    > I get the following message when I try running topas:
    > root>topas
    >
    > SpmiCreateStatSet can't create StatSet
    > root>
    >
    >
    >
    >
    > What can be causing this?
    >


  3. Re: Topas not running

    wilsjamie@gmail.com writes:

    > I get the following message when I try running topas:
    > root>topas
    >
    > SpmiCreateStatSet can't create StatSet
    > root>
    >


    One of the recording daemons that uses the Spmi API has "gotten its
    knickers in a knot". IIRC, it has something to do with corruption of
    the shared memory underlying the API.

    This could be one of the daemons supplied in the base OS, or one of
    the ones in Performance AIDE product.

    xmtrend, topas ( if you've enabled xmtopas in inetd ), xmwlm recording
    in /etc/perf/daily or /etc/perf/wlm depending on start up parameters,
    xmservd, and any other SPMI data consumer or provider are all
    candidates.

    You'll need to terminate all the SPMI consumer and producer programs
    running on the machine to free the shared memory segment.

    There were some APARs shipped for the Performance AIDE product several
    years ago ( all finally bundled into 5.3 SP06, IIRC) that solved this
    for the AIDE supplied daemons ... at least, I haven't seen it occur in
    a couple of years now.

    The xmwlm base OS supplied daemon appears to have been forked from an
    earlier version of the xmtrend program. I've seen some "stuttering" in
    the /etc/perf/daily/xmwlm.YYMMDD files that makes me wonder whether
    they forgot to include all the fixes.

    The AIX 5.3 SP05 xmwlm appears to have "issues", but I haven't seen the
    problem on my SP06 machines. ( I won't claim to have performed an
    exhaustive investigation just yet. )

    One manifestation of "the issues" is when you look at the ASCII output
    of topasout -c xmwlm.YYMMDD, and see lines where there is no metric
    name, thusly ...

    Time=2007/10/25 01:05:10, LPAR/idle=94 ( this line is OK )
    Time=2007/10/25 01:05:10, =22 ( OOPS )

    There may be some correlation with the error occuring on "busy"
    machines, with many hundreds of statistics being collected. I've seen
    the above error with some frequency ( on SP05 ) on my DB2 machines,
    which have upwards of 800 hdisks .

    The may also be some correlation with the state of WLM on the LPAR in
    question.

    I just noticed this a day or so ago myself, and haven't had a chance
    to fully "run it to ground" yet.

    --
    #include /* I don't speak for IBM ... */
    /* Heck, I don't even speak for myself */
    /* Don't believe me ? Ask my wife :-) */
    Richard D. Latham lathamr@us.ibm.com

+ Reply to Thread