In nmon, if I use the "c" option, i see the 32 logical cpus and usage is at about 50%.

If I look at nmon, and the "p" option I see 8 CPU and it's at 100.4% of CPUs.

Currently we're troubleshooting performance, and I'm wondering if I'm hitting a wall with CPU or if CPU usage is fine.

Can anyone tell me the difference between "c" and "p"?
I thought that this lpar had 8 CPU dedicated to it, and from there there are 32 logical cpu's carved out of those 8. I may not understand this correctly.

There is no top process eating cpu. But comparing systems, my other systems are about at half of the usage that this system is at in "p" menu screen.

Hope that makes sense -

Any insight is appreciated. I was looking for a quick wiki or doc that had overview of lpar'd systems CPUs and logical CPUs and what menu is the best to determine if i actually have free cpu.

Although I am not very familiar with that program I know that these are all polling files inside your /proc/ directory. Usually /proc/stat is used to determine load information. /proc/loadavg will maintain your load history.


You can also use sar if it is installed as that has a good bit of detail and logs performance for however long you set it to


http://www.cyberciti.biz/tips/how-do...ilization.html

A short answer is that 'p' reports what the hardware is doing and 'c' reports what the scheduler thinks is going on.

I'm assuming you have Power 7 hardware, as that has 4-way SMT ('SMT threads/CPU' under the 'p' panel). So, 8 cores * 4 threads = 32 logical CPUs. The entries under the 'p' option "Shared-CPU-Logical-Partition" represent actual physical CPU hardware usage. Note that while the LPAR may have 8 virtual CPUs allocated, it might have a lower Entitled Capacity. If the partition is Capped, it can only use up to it's Entitled Capacity. If it's Uncapped, it can use as many CPUs as there are virtual CPUs and run over 100%.

If panel lists something like "Summary: Entitled= 8.00 Used 8.00 (100.0%) 100.0% of CPUs in System" then that's most likely a real CPU bottleneck.

You can also switch the CPU stats in 'c' to PURR mode with the '#' key. This reports stats from performance registers on the CPU rather then what AIX's scheduler sees.

For some information about micro-partitioning see section 2.3 in the RedBook "PowerVM Virtualization on IBM System p:
Introduction and Configuration"

Also, see: https://www.ibm.com/developerworks/c...ors365?lang=en

A nice video on Linux Performace Tuning

uptime - Tell how long the system has been running.

uptime gives a one line display of the following information. The current time, how long the system has been running, how many users are currently logged on, and the system load averages for the past 1, 5, and 15 minutes.

top provides an ongoing look at processor activity in real time. It displays a listing of the most CPU-intensive tasks on the system, and can provide an interactive interface for manipulating processes

The mpstat command writes to standard output activities for each available processor, processor 0 being the first one. Global average activities among all processors are also reported. The mpstat command can be used both on SMP and UP machines, but in the latter, only global average activities will be printed. If no activity has been selected, then the default report is the CPU utilization report.


iostat - The iostat command is used for monitoring system input/output device
loading by observing the time the devices are active in relation to
their average transfer rates.

vmstat reports information about processes, memory, paging, block IO, traps, and cpu activity. The first report produced gives averages since the last reboot

free - display information about free and used memory on the system


ping : ping uses the ICMP protocol's mandatory ECHO_REQUEST datagram to elicit an ICMP ECHO_RESPONSE from a host or gateway


Dstat allows you to view all of your system resources instantly, you can eg. compare disk usage in combination with interrupts from your IDE controller, or compare the network bandwidth numbers directly with the disk throughput (in the same interval)

wingnut reads well, but I want to try to improve it a bit.

c is "CPU by processor" - as you state, that should be logical processor to prevent confusion
p is "LPAR stats" (if an LPAR)

in more normal *NIX terms, c is presenting the stats based on the familiar concepts of user%, sys%, wait% and io% (vmstat like, but per processor. Note sar -P ALL, and mpstat NN (where NN is an interval) also gives per logical CPU stats).

The 'p' option is - guessing here - what I call "physical" usage. On other AIX commands, such as vmstat, mpstat and lparstat provide two columns, one labeled ec (entitled capacity as you mentioned) and pc (what I call physical/processor consumption). Hence, I suspect that the 'p' option is reporting the equivalent of the 'pc' column from the lparstat command.

@boyd98 - I hope after looking at the output of vmstat, mpstat and lparstat you will have a better understanding of what nmon is providing.

Good Luck!

Well... this is AIX, not linux, right?

Can you cut and paste an iteration of a topas page when the system is cpu-bound (not the first one, though)?

It does sound cpu-bound, the question is (obviously), why? is there paging/swapping? is the lrud overtaxed? Are file systems overly full? Runaway processes? Once we discover what's the proximate cause, then you can look at the root cause. the processes showing in the lower left of topas are the first place I'd look, to see what's running. I'd also want to see the values in user, sys, and wait. maybe show us a sar 3 5?

dan