Forum Discussion
Cirrustmstat versus CLI load verus Dashboard CPU utilization?
We have an old LTM 1500 running 10.2.4 HF5.
I'm concerned that the Celeron single CPU system is running above 1.0 load on the command line. tmstat reads about 26% Big IP dashboard reads 55%
This is with about 6000 connections, 50 Mbps. here is output from "top" on the CLI: 5460 root RT 0 1523m 30m 21m S 33.7 1.5 329391:36 tmm 3176 root 15 0 32800 10m 5416 R 7.3 0.5 13345:40 bigd 3196 root 15 0 87504 52m 8304 S 4.6 2.6 5312:19 mcpd 3175 root 15 0 17964 4032 2340 S 2.6 0.2 6187:58 statsd 5488 tomcat 18 0 295m 169m 2812 S 1.0 8.4 225:11.77 java
Which of these CPU utilization metrics should I go by? We are expecting additional traffic and I am concerned that the 1500 will not be able to handle it.
3 Replies
- JRahm
Admin
the linux performance commands really can't be trusted because the TMM reserves much of the system resources at boot time. You definitely want to view BIG-IP performance metrics from TMM commands and stats.
hoolioCirrostratus
tmstat and the GUI's CPU utilization stats should show the same CPU usage for TMM.
I think top should also be accurate for TMM by 10.2.4. There was a bug where linux utilities couldn't see TMM's actual utilization but that was for v9 and older v10 versions.
Are you sure you were checking both at the same time?
Aaron
pjcampbell_7243It might have been that the CPU spiked when I opened the dashboard. top and tmstat show similar but not the same CPU usage for tmm.
So, should I not be concerned that the Linux cpuload is often over 1.0 with a single processor system as long as the tmstat or top cpu is reasonable?
