AIX

AIX

Connect with fellow AIX users and experts to gain knowledge, share insights, and solve problems.


#Power
 View Only
  • 1.  Why so many wait processes?

    Posted Fri June 09, 2006 01:04 PM

    Originally posted by: SystemAdmin


    I have a p5 550 with 4 CPUs, 4GB of memory running AIX5.2 08 02 .
    This machine doesn't have CuOD or an HMC. It is not a Q-version
    machine.
    Does anyone know why there is so many wait procs running?
    Why would there be more than 4 one for each CPU?
    here is a ps -elk

    F S UID PID PPID C PRI NI ADDR SZ WCHAN TTY TIME CMD
    303 A 0 0 0 120 16 -- 44011 20 - 0:03 swapper
    200003 A 0 1 0 0 60 20 540b5 832 - 0:00 init
    303 A 0 516 0 120 255 -- 48012 16 - 129:35 wait
    303 A 0 774 0 120 255 -- 4c013 16 - 90:02 wait
    303 A 0 1032 0 120 255 -- 50014 16 - 107:15 wait
    303 A 0 1290 0 120 255 -- 54015 16 - 116:34 wait
    303 A 0 1548 0 0 255 -- 58016 16 - 0:00 wait
    303 A 0 1806 0 0 255 -- 64019 16 - 0:00 wait
    303 A 0 2064 0 0 255 -- 6801a 16 - 0:00 wait
    303 A 0 2322 0 0 255 -- 6c01b 16 - 0:00 wait
    303 A 0 2580 0 0 255 -- 7001c 16 - 0:00 wait
    303 A 0 2838 0 0 255 -- 7401d 16 - 0:00 wait
    303 A 0 3096 0 0 255 -- 7801e 16 - 0:00 wait
    303 A 0 3354 0 0 255 -- 7c01f 16 - 0:00 wait
    303 A 0 3612 0 0 255 -- 20 16 - 0:00 wait
    303 A 0 3870 0 0 255 -- 4021 16 - 0:00 wait
    303 A 0 4128 0 0 255 -- 8022 16 - 0:00 wait
    303 A 0 4386 0 0 255 -- c023 16 - 0:00 wait
    303 A 0 4644 0 0 255 -- 10024 16 - 0:00 wait
    303 A 0 4902 0 0 255 -- 14025 16 - 0:00 wait
    303 A 0 5160 0 0 255 -- 18026 16 - 0:00 wait
    303 A 0 5418 0 0 255 -- 1c027 16 - 0:00 wait
    303 A 0 5676 0 0 255 -- 20028 16 - 0:00 wait
    303 A 0 5934 0 0 255 -- 24029 16 - 0:00 wait
    303 A 0 6192 0 0 255 -- 2802a 16 - 0:00 wait
    303 A 0 6450 0 0 255 -- 2c02b 16 - 0:00 wait
    303 A 0 6708 0 0 255 -- 3002c 16 - 0:00 wait
    303 A 0 6966 0 0 255 -- 3402d 16 - 0:00 wait
    303 A 0 7224 0 0 255 -- 3802e 16 - 0:00 wait
    303 A 0 7482 0 0 255 -- 3c02f 16 - 0:00 wait
    303 A 0 7740 0 0 255 -- 40030 16 - 0:00 wait
    303 A 0 7998 0 0 255 -- 44031 16 - 0:00 wait
    303 A 0 8256 0 0 255 -- 48032 16 - 0:00 wait
    303 A 0 8514 0 0 255 -- 4c033 16 - 0:00 wait
    303 A 0 8772 0 0 255 -- 50034 16 - 0:00 wait
    303 A 0 9030 0 0 255 -- 54035 16 - 0:00 wait
    303 A 0 9288 0 0 255 -- 58036 16 - 0:00 wait
    303 A 0 9546 0 0 255 -- 5c037 16 - 0:00 wait
    303 A 0 9546 0 0 255 -- 5c037 16 - 0:00 wait
    303 A 0 9804 0 0 255 -- 60038 16 - 0:00 wait
    303 A 0 10062 0 0 255 -- 64039 16 - 0:00 wait
    303 A 0 10320 0 0 255 -- 6803a 16 - 0:00 wait
    303 A 0 10578 0 0 255 -- 6c03b 16 - 0:00 wait
    303 A 0 10836 0 0 255 -- 7003c 16 - 0:00 wait
    303 A 0 11094 0 0 255 -- 7403d 16 - 0:00 wait
    303 A 0 11352 0 0 255 -- 7803e 16 - 0:00 wait
    303 A 0 11610 0 0 255 -- 7c03f 16 - 0:00 wait
    303 A 0 11868 0 0 255 -- 40 16 - 0:00 wait
    303 A 0 12126 0 0 255 -- 4041 16 - 0:00 wait
    303 A 0 12384 0 0 255 -- 8042 16 - 0:00 wait
    303 A 0 12642 0 0 255 -- c043 16 - 0:00 wait
    303 A 0 12900 0 0 255 -- 10044 16 - 0:00 wait
    303 A 0 13158 0 0 255 -- 14045 16 - 0:00 wait
    303 A 0 13416 0 0 255 -- 18046 16 - 0:00 wait
    303 A 0 13674 0 0 255 -- 1c047 16 - 0:00 wait
    303 A 0 13932 0 0 255 -- 20048 16 - 0:00 wait
    303 A 0 14190 0 0 255 -- 24049 16 - 0:00 wait
    303 A 0 14448 0 0 255 -- 2804a 16 - 0:00 wait
    303 A 0 14706 0 0 255 -- 2c04b 16 - 0:00 wait
    303 A 0 14964 0 0 255 -- 3004c 16 - 0:00 wait
    303 A 0 15222 0 0 255 -- 3404d 16 - 0:00 wait
    303 A 0 15480 0 0 255 -- 3804e 16 - 0:00 wait
    303 A 0 15738 0 0 255 -- 3c04f 16 - 0:00 wait
    303 A 0 15996 0 0 255 -- 40050 16 - 0:00 wait
    303 A 0 16254 0 0 255 -- 44051 16 - 0:00 wait
    303 A 0 16512 0 0 255 -- 48052 16 - 0:00 wait
    303 A 0 16770 0 0 255 -- 4c053 16 - 0:00 wait
    vmstat
    System Configuration: lcpu=4 mem=3792MB
    kthr memory page faults cpu
    -----------
    ------------

    r b avm fre re pi po fr sr cy in sy cs us sy id wa
    1 1 262206 3081 0 0 0 258 597 0 790 182455 663 4 15 67 14
    1 0 262022 3245 0 1 0 0 0 0 669 277303 324 5 21 73 1

    bindprocessor -q
    The available processors are: 0 1 2 3

    prtconf
    System Model: IBM,9113-550
    Machine Serial Number: 65BDA2E
    Processor Type: PowerPC_POWER5
    Number Of Processors: 4
    Processor Clock Speed: 1654 MHz
    CPU Type: 64-bit
    Kernel Type: 32-bit
    LPAR Info: 1 65-BDA2E
    Memory Size: 3792 MB
    Good Memory Size: 3792 MB
    Firmware Version: IBM,SF235_185
    Console Login: enable
    Auto Restart: true
    Full Core: false
    #AIX-Forum


  • 2.  Some suggestions

    Posted Thu June 22, 2006 06:04 AM

    Originally posted by: nagger


    I can't imagine why you are running AIX 5.2 - this throws away 30 to 40% of the machine speed as I don't think you can switch on SMT!

    No idea what a "Q-machine" or do you mean the p550 Q as in the 8 way version of the p550?

    Anyway, you should have four wait processes and I see our tha are actually running on the machine at the top of the list.

    The firmware is recent.
    The kernel is 32 bit - ugh!!

    So what are the other waits:
    You could, of course, be running lots of processes called wait although they do appear to be the same size as the in-use waits.
    AIX might have decided you could be running on a machine that could have Dynamic LPAR support and started 64 waits one for each of the possible CPUs
    This appears to be an LPAR "LPAR Info: 1 65-BDA2E" even though you say it has no HMC - is this an IVM machine. - I am thinking that perhaps it thinking you have 64 virtual processors.
    Or it could be a bug.

    I would call AIX support to investigate, go for the 64 bit kernel, upgrade to AIX 5.3 and install IVM so you have LPARs - but each to there own right!
    #AIX-Forum


  • 3.  Re: Some suggestions

    Posted Fri June 23, 2006 01:04 PM

    Originally posted by: SystemAdmin


    Here is the responce from IBM:

    The fact that you have 64 wait with "ps" processes is a normal
    behaviour.

    Indeed from ps output there's only 4 "wait" that are "running" and
    accumulating times.
    This is because you have 4 cpus. Hence 1 runnable "wait" process / cpu

    Now about having 64 wait processes: even if that system is a
    fullpartition, in fact it's an lpar, but only 1 lpar. Hence as any lpar
    there's an lpar profile (which you cannot see/edit because
    it's a fullpartition and without HMC).

    And this lpar profile is set with max cpu = 64 :
    original_ncpus... 00000004 max_ncpus........ 00000040 <-- 64 in decimal

    ps output shows a number of "wait" processes that represents the max
    number of cpu defined in the lpar profile.

    PS: In addition this can't lead to any problem as "wait" process means
    no cpu activity.
    #AIX-Forum