Performance monitoring

From Stuff!
Jump to: navigation, search

CPU

mpstat, topas -P, w, lparstat, ps, iostat -tT 1, tprofy, curty

Memory

vmstat, svmon, ps -o fields , topas, ipcs -m

Network I/O

[ent|tok|fddi|atm]stat, netstat, netpmony, topas -E

Disk I/O

iostat, fcstat, lvmstat, filemony, fileplace, topas -D

Application

truss, probevue, tprofy, svmon -P pid , ps -o fields -p pid

topas - The ~ character toggles to nmon-mode in topas


Check for disk stat history collection

lsattr -HEl sys0 -a iostat

Enable historical disk statistic collection

chdev -l sys0 -a iostat=true

Comprehensive detail monitoring

In order to produce a deep and very detailed monitoring, a background nmon process can be run which will collect the data included in the all of the commands above. However the data received this way has to go through an nmon parser to get the data look like something readable. the command syntac for nmon is:

nmon -fT -c <number> -s <number of seconds between samples> -m <file_directory>

What this will do, is put nmon into background and collect data <count> number of times each separated by <numeber of seconds between samples>.

Example:

nmon -fT -c 3600 -s 10 -m /tmp

Will create a file *.nmon in /tmp and store there 3600 samples of system data each separated by 10 second interval.