Topic
3 replies Latest Post - ‏2012-04-26T05:34:00Z by puvichakravarthy
molekuul
molekuul
4 Posts
ACCEPTED ANSWER

Pinned topic Strange CPU behavior

‏2012-04-24T12:24:04Z |
At our AIX 6 LPAR’s we see a strange CPU behavior, however the LPAR is not using a lot of CPU, the system is approx. 99% idle, the LPAR is using 400% CPU entitlement.
Hereby I show a part of nmon, it shows an entitlement of 0.5, approx. 100% idle, but an Used of 1.965
Furthermore in the bottom I see these “i”

┌─topas_nmon──p=Partitions───────Host=XXXXXXXX───────Refresh=2 secs───14:21.45────┐
│ CPU-Utilisation-Small-View ───────────EntitledCPU= 0.50 UsedCPU= 1.965────────│
│Logical CPUs 0----------25-----------50----------75----------100 │
│CPU User% Sys% Wait% Idle%| | | | | │
│ 0 0.5 1.0 0.0 98.5| > | │
│ 1 0.0 0.0 0.0 100.0| > | │
│ 2 0.0 3.5 0.0 96.5|s > | │
│ 3 0.0 4.8 0.0 95.2|ss > | │
│EntitleCapacity/VirtualCPU -----------|------------|-----------|------------
│EC+ 0.4 0.8 0.1 98.6|iiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiii| │
│ VP 0.4 0.8 0.1 96.9|iiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiii-| ii│
│EC= 393.1% VP= 98.3% --No Cap---|------------|-----------100% VP=2 CPU
│─────────────────────────────────────────────────────────────────────────────────│

lparstat shows the same numbers:

  1. lparstat 1

System configuration: type=Shared mode=Uncapped smt=On lcpu=4 mem=4096MB psize=7 ent=0.50

%user %sys %wait %idle physc %entc lbusy vcsw phint

-----
------
-----
-----
15.5 18.8 0.5 65.2 2.00 399.8 35.0 400 392
8.4 10.8 0.3 80.5 2.00 399.4 19.8 400 406
0.5 1.1 0.0 98.4 1.99 398.2 2.0 404 336
0.7 1.1 0.1 98.1 2.00 399.6 2.2 402 372

Can someone tell me what’s going on ?
Updated on 2012-04-26T05:34:00Z at 2012-04-26T05:34:00Z by puvichakravarthy
  • puvichakravarthy
    puvichakravarthy
    55 Posts
    ACCEPTED ANSWER

    Re: Strange CPU behavior

    ‏2012-04-25T04:59:12Z  in response to molekuul
    By any chance is any of the tunable modified in schedo ("schedo -FL" and see what values are not default)? I suspect if snooze_delay would have been set at -1. Could you verify?
    • molekuul
      molekuul
      4 Posts
      ACCEPTED ANSWER

      Re: Strange CPU behavior

      ‏2012-04-25T06:23:08Z  in response to puvichakravarthy
      Hy The snooze_delay is not set at -1, it is 0, hereby I send you the schedo -FL

      The OS we are running is:
      1. oslevel -s
      6100-05-06-1119
      1. schedo -FL
      NAME CUR DEF BOOT MIN MAX UNIT TYPE
      DEPENDENCIES

      affinity_lim 7 7 7 0 100 dispatches D

      big_tick_size 1 1 1 1 100 10 ms D

      ded_cpu_donate_thresh 80 80 80 0 100 % busy D

      fixed_pri_global 0 0 0 0 1 boolean D

      force_grq 0 0 0 0 1 boolean D

      maxspin 16K 16K 16K 1 4G-1 spins D

      pacefork 10 10 10 10 2G-1 clock ticks D

      proc_disk_stats 1 1 1 0 1 boolean D

      sched_D 16 16 16 0 32 D

      sched_R 16 16 16 0 32 D

      tb_balance_S0 2 2 2 0 2 ticks D

      tb_balance_S1 2 2 2 0 2 ticks D

      tb_threshold 100 100 100 10 1000 ticks D

      timeslice 1 1 1 0 2G-1 clock ticks D

      vpm_fold_policy 1 1 1 0 15 D

      vpm_xvcpus 0 0 0 -1 2G-1 processors D

      ##Restricted tunables

      %usDelta 100 100 100 0 100 D

      allowMCMmigrate 0 0 0 0 1 boolean D

      allow_vmx 1 1 1 0 1 boolean B

      clk_transition 12 12 12 6 10000 D

      fast_locks 0 0 boolean d

      hotlocks_enable 0 0 0 0 1 boolean D

      idle_migration_barrier 4 4 4 0 100 sixteenth D

      intr_stealing 0 0 0 0 1 boolean D

      jitter_control 0 0 0 0 1 boolean D

      krlock_confer2self 1 1 1 0 1 boolean D

      krlock_conferb4alloc 1 1 1 0 1 boolean D

      krlock_enable 1 1 1 0 1 boolean D

      krlock_spinb4alloc 1 1 1 1 2G-1 D

      krlock_spinb4confer 1K 1K 1K 0 2G-1 D

      n_idle_loop_vlopri 100 100 100 0 976K D

      search_globalrq_mload 256 256 256 0 4095M D

      search_smtrunq_mload 256 256 256 0 4095M D

      setnewrq_sidle_mload 384 384 384 0 4095M D

      shed_primrunq_mload 64 64 64 0 4095M D

      sidle_S1runq_mload 64 64 64 0 4095M D
      sidle_S2runq_mload

      sidle_S2runq_mload 134 134 134 0 4095M D
      sidle_S1runq_mload
      sidle_S3runq_mload

      sidle_S3runq_mload 134 134 134 0 4095M D
      sidle_S2runq_mload
      sidle_S4runq_mload

      sidle_S4runq_mload 4095M 4095M 4095M 0 4095M D
      sidle_S3runq_mload

      slock_spinb4confer 1K 1K 1K 0 2G-1 D

      smt_option_flags 0 0 0 0 1 numeric D

      smt_snooze_delay 0 0 0 -1 97656K microsecs D

      smt_tertiary_snooze_delay 0 0 0 -1 97656K microsecs D

      smtrunq_load_diff 2 2 2 1 4095M D

      tertiary_barrier_load 128 128 128 0 4095M D

      tick_sync 0 0 0 0 1 boolean D

      v_exempt_secs 2 2 2 0 2G-1 seconds D

      v_min_process 2 2 2 0 2G-1 processes D

      v_repage_hi 0 0 0 0 2G-1 D

      v_repage_proc 4 4 4 0 2G-1 D

      v_sec_wait 1 1 1 0 2G-1 seconds D

      vpm_min_sleep 500 500 500 20 999 milliseconds D


      n/a means parameter not supported by the current platform or kernel

      Parameter types:
      S = Static: cannot be changed
      D = Dynamic: can be freely changed
      B = Bosboot: can only be changed using bosboot and reboot
      R = Reboot: can only be changed during reboot
      C = Connect: changes are only effective for future socket connections
      M = Mount: changes are only effective for future mountings
      I = Incremental: can only be incremented
      d = deprecated: deprecated and cannot be changed

      Value conventions:
      K = Kilo: 2^10 G = Giga: 2^30 P = Peta: 2^50
      M = Mega: 2^20 T = Tera: 2^40 E = Exa: 2^60
      • puvichakravarthy
        puvichakravarthy
        55 Posts
        ACCEPTED ANSWER

        Re: Strange CPU behavior

        ‏2012-04-26T05:34:00Z  in response to molekuul
        Do you have a IBM support ticket opened already? Else I guess you should probably do that.