rc.powerfail / machstat issues

From: F. Even (aixlists_at_ELITISTS.ORG)
Date: 05/17/05

  • Next message: Holger.VanKoll_at_SWISSCOM.COM: "hacmp with oracle9.x and dispatchers" 
    Date:         Mon, 16 May 2005 18:02:25 -0500
To: aix-l@Princeton.EDU

    OK, I've got a box that had an rc.powerfail warning added to root's
    crontab due to the secondary power supply losing power (due to building
    maintenance knocking out one of it's power feeds). The server is a
    pSeries 630 Model 6C4. Now...I've cleared the cron entries, etc. What I
    would like to know though is how do I clear whatever registers that are
    indicating there still is a power problem? If I execute "rc.powerfail
    -h," it shoots back this:

    [quote]
    rc.powerfail -h

    Broadcast message from root@ (pts/96) at 13:40:35 ...

    rc.powerfail: init has received a SIGPWR signal.
            The system is now operating with a non-critical power problem.
            Execute rc.powerfail -h as the root user for more information.
    ^G^G^G^G

    Broadcast message from root@ (pts/96) at 13:40:57 ...

    rc.powerfail:
     This command is used to handle power problems with the system.
     There are several different states that the system can be in when
     the signal SIGPWR is received by init. The action taken will be
     determined by the value of the power status. The following table
     shows the values of the power status and action taken.
     Power
     Status Indication
     ------ ----------------------------------------------------------------
       0 System is running normally, there is no action taken.
       1 A non-critical cooling problem exists.
       2 A non-critical power problem exists.
       3 System facing a critical condition. Will start shutdown in 10
    minutes.
       4 System facing a severe condition. Will be halted in the next 20
    seconds.
     255 ERROR with the machstat command, system shutdown starts
    immediately. ^G^G^G^G
    [/quote]

    ...and re-adds the wall entry to root's crontab. NOW...if I check the
    documented flags to machstat:

    > machstat -p

    > echo $?
    0

    Everything seems like it should be fine (unless someone else can tell me a
    way to check the power supply status on this platform..."uesensor -l" is
    not supported).

    Now...looking through the rc.powerfail and rc.powerfail_chrp, I see
    references to an undocumented use of machstat, "machstat -f," which seems
    to be pulling a stored value of some sort:

    > /usr/sbin/machstat -f
    2 0

    ...now the question is...how do I get that to read as it does on a "well"
    system:

    > machstat -f
    0 0

    ....how do I reset that value? The box is remote, and it cannot be
    rebooted just for this, but the local hardware support has checked it out
    and can't see any issues with the power supply. It lost power over the
    weekend, that seems to have been what caused this...but it is good
    now...how do I get the system to recognize that?

    P.S. There are currently no errors in errpt for this issue...I think some
    entries might have been cleared out by co-workers before I looked at
    it...so there are no entries relating to this power issue currently in
    errpt.

    Thanks for any input anyone might have,
    Frank

  • Next message: Holger.VanKoll_at_SWISSCOM.COM: "hacmp with oracle9.x and dispatchers"

    Relevant Pages

    • Re: rc.powerfail / machstat issues
      ... Can you go through diags and log a repair action against the hardware ... crontab due to the secondary power supply losing power (due to building ... A non-critical power problem exists. ... 255 ERROR with the machstat command, ...
      (AIX-L)
    • Re: IBM monitor wont start up - relay keeps tripping
      ... When I read this I visualized someone having a display with a power ... nothing to do with the power problem. ... One had an open 1meg resistor in the screen circuit and the ... I did find a FET that has a D-S short. ...
      (sci.electronics.repair)
    • Re: display window shrunk
      ... this is a problem that must be corrected with the monitor ... in and of itself is odd since one would assume the docking station is ... plugged into a power source and when the laptop is removed, ... This would seem to indicate there's a power problem ...
      (microsoft.public.windowsxp.accessibility)
    • Re: HP Pavilion Laptop Owners with Bad Power Jacks....
      ... I have a ZD7000 that's having the power problem now. ... Can anyone here verify ... the battery or the powerjack. ...
      (comp.sys.hp.hardware)
    • Re: DRO toast?
      ... It's not a power problem, it gets juice all the way to the ... Anyone know of anyone who can diagnose and repair a mid 80s Teledyne-Gurley ... Most DROs use TTL level encoders that are electrically ...
      (rec.crafts.metalworking)