System resource monitoring

[Andy Dye]

I'm fairly new to HP-UX and need some input on monitoring resource usage.

I'm running 10.20 and our year end processing is running into a bottle neck somewhere.

I'm looking at "top" and have read descriptions of what I'm looking at, but am not sure what's good or bad. (Other than obvious stuff. Like load and idle.) I also have some instructions for using vmstat and iostat. (All info I have is from "HP Certified" by Rafeeq Ur Rehman)

Which statistics should I be focusing on? How would you recommend I monitor these?

Thank you.

[John Carr_2]

Hi

see if you have glance plus monitor on your system command gpm gives a visual graphical display of what is happening. If you do not have it you can download an evaluation 60 day copy

http://www.software.hp.com/

you can also search the forum using the search box at the top of this page as this has been covered many many times

:-) John.

[Marco Santerre]

Alogn with using Glance, which is a very good product, I would suggest you also look at its documentation (some of which you can find at http://docs.hp.com, search for GlancePlus)

[MANOJ SRIVASTAVA]

Andy

You can also get a trial version of sarcheck from www.sarcheck.com which sets up a data collector in the cron using sar and you can run to get formatted output report which is really discriptive and useful.


Manoj Srivastava

[Stuart Abramson_2 · Accepted solution]

1. glance. Download trial as described above.

2. MeasureWare/PerfView (two products). You can download trials as described above, but this is complicated.

3. Set up automatic sar data collection and review sar outputs - all of them. go through man sar.

Here is how you set up sar data collection. It takes some disk space:

1. To set it up:

Start sar:
cd /root/doc
crontab -l > crontab.mmdd
vi crontab.mmdd
add:
0 * * * * /usr/lib/sa/sa1 600 6
crontab < cron.mmdd
crontab -l
mkdir /var/adm/sa

2. To use it:

a. The sar collector puts data in:

/var/adm/sa/

in files named: saNN

sa01
sa02

where NN is the day of the month on which the data was collected.


abramsd@adedd09p:/var/adm/sa$ ll
total 294718
-rw-r--r-- 1 root sys 5810240 Oct 1 23:50 sa01
-rw-r--r-- 1 root sys 6243840 Oct 2 23:50 sa02
...
-rw-r--r-- 1 root sys 3409920 Sep 14 00:50 sa13
-rw-r--r-- 1 root sys 780480 Oct 14 23:50 sa14

b. To find out table space usage on ANY system right now:

sar -v 5 5

c. To get a 24 hour history of today:

sar -v

d. To get a 24 hour history of October 8:

sar -v -f /var/adm/sa/sa08

3. Maintenance:

a. The files in /var/adm/sa write over after 30 days:

October 1 was written on sa01
November 1 will overwrite sa01

b. You might want to put something in place to wipe out 2 week old files:

find /var/adm/sa -mtime +15 -exec rm {} \;

[Steven E. Protter]

This script will probably help.

Attached.

SEP

[PVR]

Read the man pages of sar, iostat, vmstat etc..This will help in monitoring the system.

try to get glanceplus. This will give graphical representation and very easy to interpret

[Tim Sanko]

I am lazy and have cron issue this script hourly.

00 * * * * /root/SCRIPTS/monitor.sh > /dev/null 2>&1

Month=`/usr/bin/date +%m`
Day=`/usr/bin/date +%d`
Year=`/usr/bin/date +%y`
Hour=`/usr/bin/date +%H`
Minute=`/usr/bin/date +%M`
LOGFILE=/root/LOG/SAR/$Year$Month$Day-$Hour$Minute.sar

/usr/sbin/sar -A -o $LOGFILE 60 29 > /dev/nul &

This allows hourly breakdown or summation of daily logs.

I personnally check the output file daily.

Tim