numbers for CPU load less meaningful than runque numbers?
list Timothy L. Myers
Anyone else running AIX 5.3ml3 seeing load numbers coming form uptime, to be less useful than the mpstat runque numbers?
mpstat reports -
cpu min maj mpc int cs ics rq mig lpa sysc us sy wa id pc 0 275 1 0 160 7069 2986 0 13277 100.0 12950 17.9 9.6 3.9 68.6 0.51 1 16 0 0 165 7986 3601 1 11905 100.0 9791 17.6 8.7 2.5 71.3 0.49 2 179 1 0 163 8064 4211 0 11460 100.0 11374 16.3 9.0 2.7 72.0 0.49 3 132 2 0 319 9523 4963 0 14572 100.0 9470 18.1 9.5 3.0 69.4 0.51 ALL 602 4 0 807 32642 15761 1 51214 100.0 43585 17.5 9.2 3.0 70.3 2.00 --------------------------------------------------------------------------------------------------------
0 251 4 0 160 6926 2864 0 12986 100.0 11030 17.6 9.0 3.0 70.4 0.49 1 45 1 0 159 8211 3864 0 11921 100.0 13722 19.0 9.0 4.1 67.9 0.50 2 166 3 0 155 8425 4490 0 11967 100.0 10509 16.3 8.3 2.6 72.9 0.49 3 100 0 0 310 9537 4970 0 14084 100.0 10018 18.5 8.8 3.6 69.2 0.51 ALL 562 8 0 784 33099 16188 0 50958 100.0 45279 17.9 8.7 3.3 70.1 1.99
uptime reports -
09:06AM up 2 days, 14:51, 56 users, load average: 41.50, 34.30, 29.37
I am not seeing any of the slowness I would expect from this kind of load so I have spent the last couple of days with snapshots trying to figure it out just what is going wrong.
I don't know but I suspect that the problems with the utmp and wtmp files are causing finger issues
The performance and processes are much more in line with mpstat run queue numbers than with uptime run able process numbers.
Timothy L Myers
Senior Developer
Autom 5226 South 31st Place
Phoenix AZ 85040
list Jeff Newman
How does it compare to say, sar -P ALL 1 in terms of %idle? -Jeff
▸
On 1/24/06, Myers, Timothy L. <user-96e7c936b511@xymon.invalid> wrote:Anyone else running AIX 5.3ml3 seeing load numbers coming form uptime, to be less useful than the mpstat runque numbers? *mpstat reports -*** cpu min maj mpc int cs ics rq mig lpa sysc us sy wa id pc 0 275 1 0 160 7069 2986 0 13277 100.0 12950 17.9 9.6 3.9 68.6 0.51 1 16 0 0 165 7986 3601 1 11905 100.0 9791 17.6 8.7 2.5 71.3 0.49 2 179 1 0 163 8064 4211 0 11460 100.0 11374 16.3 9.0 2.7 72.0 0.49 3 132 2 0 319 9523 4963 0 14572 100.0 9470 18.1 9.5 3.0 69.4 0.51 ALL 602 4 0 807 32642 15761 1 51214 100.0 43585 17.5 9.2 3.0 70.3 2.00 0 251 4 0 160 6926 2864 0 12986 100.0 11030 17.6 9.0 3.0 70.4 0.49 1 45 1 0 159 8211 3864 0 11921 100.0 13722 19.0 9.0 4.1 67.9 0.50 2 166 3 0 155 8425 4490 0 11967 100.0 10509 16.3 8.3 2.6 72.9 0.49 3 100 0 0 310 9537 4970 0 14084 100.0 10018 18.5 8.8 3.6 69.2 0.51 ALL 562 8 0 784 33099 16188 0 50958 100.0 45279 17.9 8.7 3.3 70.1 1.99 *uptime reports -* 09:06AM up 2 days, 14:51, 56 users, load average: 41.50, 34.30, 29.37 I am not seeing any of the slowness I would expect from this kind of load so I have spent the last couple of days with snapshots trying to figure it out just what is going wrong. I don't know but I suspect that the problems with the utmp and wtmp files are causing finger issues The performance and processes are much more in line with mpstat run queue numbers than with uptime run able process numbers. *Timothy L Myers* *Senior Developer* *Autom * 5226 South 31st Place Phoenix AZ 85040
list Timothy L. Myers
The idle reports the same on sar as mpstat, I also noticed that vmstat run queue matches uptime numbers. I haven't seen numbers this hi on 4.3.3 systems with out them being slow to the point of pain. I am getting loads of 40 - 60 in production with still showing idle time.
▸
-----Original Message-----
From: Jeff Newman [mailto:user-e96740e73ca8@xymon.invalid]
Sent: Tuesday, January 24, 2006 11:34 AM
To: user-ae9b8668bcde@xymon.invalid
Subject: Re: [hobbit] numbers for CPU load less meaningful than runque numbers?
How does it compare to say, sar -P ALL 1 in terms of %idle?
-Jeff
On 1/24/06, Myers, Timothy L. < user-96e7c936b511@xymon.invalid> wrote:
Anyone else running AIX 5.3ml3 seeing load numbers coming form uptime, to be less useful than the mpstat runque numbers?
mpstat reports - cpu min maj mpc int cs ics rq mig lpa sysc us sy wa id pc 0 275 1 0 160 7069 2986 0 13277 100.0 12950 17.9 9.6 3.9 68.6 0.51 1 16 0 0 165 7986 3601 1 11905 100.0 9791 17.6 8.7 2.5 71.3 0.49 2 179 1 0 163 8064 4211 0 11460 100.0 11374 16.3 9.0 2.7 72.0 0.49 3 132 2 0 319 9523 4963 0 14572 100.0 9470 18.1 9.5 3.0 69.4 0.51 ALL 602 4 0 807 32642 15761 1 51214 100.0 43585 17.5 9.2 3.0 70.3 2.00 -------------------------------------------------------------------------------------------------------- 0 251 4 0 160 6926 2864 0 12986 100.0 11030 17.6 9.0 3.0 70.4 0.49 1 45 1 0 159 8211 3864 0 11921 100.0 13722 19.0 9.0 4.1 67.9 0.50 2 166 3 0 155 8425 4490 0 11967 100.0 10509 16.3 8.3 2.6 72.9 0.49 3 100 0 0 310 9537 4970 0 14084 100.0 10018 18.5 8.8 3.6 69.2 0.51 ALL 562 8 0 784 33099 16188 0 50958 100.0 45279 17.9 8.7 3.3 70.1 1.99
uptime reports -
09:06AM up 2 days, 14:51, 56 users, load average: 41.50, 34.30, 29.37
I am not seeing any of the slowness I would expect from this kind of load so I have spent the last couple of days with snapshots trying to figure it out just what is going wrong.
I don't know but I suspect that the problems with the utmp and wtmp files are causing finger issues
The performance and processes are much more in line with mpstat run queue numbers than with uptime run able process numbers.
Timothy L Myers Senior Developer
Autom 5226 South 31st Place Phoenix AZ 85040