Application Perf Mgmt (BAC / BSM) Practitioners Forum
cancel

Sitescope 11.11 Monitoring Solaris 10 Server Physical Memory Monitoring

jtwcarboy
New Member.

Sitescope 11.11 Monitoring Solaris 10 Server Physical Memory Monitoring

Has anyone came up with a solution or custom COMMAND= Line for Solaris10 Physical Memory Monitoring?

I have written a custom script which works ok but i was hoping hp would have fixed the default monitor.

You can easily tell its monitoring swap by comparing top output ot the sitescope monitor.

Any help is appreciated.  It appears that the built in tool monitors swap not physical memory

4 REPLIES
Sergey Popov
Outstanding Contributor.

Re: Sitescope 11.11 Monitoring Solaris 10 Server Physical Memory Monitoring

Hi.

11.11's memory monitor should monitor both swap and real RAM usage (and it does so in our lab environment).
Could you, please, be more specific about the monitored system?

Also, let me tear down the command that is currently used (yes, it was me who wrote that mess).
First it gets total amount of RAM via prtconf:
prtconf | grep "Mem" | awk '{print $3*1024;}'

 

Then current free RAM is taken from sar:
sar -r 1 1 | grep -v "freemem" | grep -v "SunOS" | grep -v "^$" | awk '{print $2;}' 

 

Ok, and I don't remember why did I need the awk part for as it essentially reprints these results. For some older variant, I think, when total RAM and percentage were transferred and I had to divide one thing on another.

Anyway, take a look on these, please, maybe it will help us understand what exactly isn't working.

Thank you.
Sergey.

_______________
This thread is now 20% cooler
Highlighted
mgg4
New Member.

Re: Sitescope 11.11 Monitoring Solaris 10 Server Physical Memory Monitoring

I'm having what I think is a related issue.  I'm using Sitescope 11.11 to Monitor a Solaris 10 server.  The physical memory is showing as 100% used, but running the sar -r command referenced in this thread, I see I have free memory.  In addition, the value for pages/sec is totally out of reasonable range (over 5 billion pages/sec).  I'm also seeing "Swap in use" values greater than zero, but the swap -l command on the system shows the swap file is 0% used.

 

Is this just the way SS views Solaris 10, or is there something I can do to make these number make sense?

 

--Mark

joncarmichael
Frequent Contributor.

Re: Sitescope 11.11 Monitoring Solaris 10 Server Physical Memory Monitoring

I had the same issue in which SUN physical memory would always show 100%.  The problem is with the cmd in the templates.os.  Replace the entry and restart SiS and it should work.

 

ORIGINAL:

 

#

id=physicalMemory

command=echo `prtconf | grep "Mem" | awk '{print $3*1024;}'`" "`sar -r 1 1 | grep -v "freemem" | grep -v "SunOS" | grep -v "^$" | awk '{print $2;}'`" "`getconf PAGE_SIZE` | awk '{print $1,$2*$3/1024;}'

total=1

free=2

swapUnit=1024

 

 

NEW:

#

id=physicalMemory

command=echo `/usr/sbin/prtconf | grep "Mem" | awk '{print $3*1024;}'`" "`sar -r 1 1 | tail -1 | awk '{print $2;}'`" "`getconf PAGE_SIZE` | awk '{print $1,$2*$3/1024;}'

total=1

free=2

swapUnit=1024

 

The problem was the prtconf cmd is not found unless run w/ path /usr/sbin/prtconf so was messing up the calculation (resulting in 100%).  I also had to shorten the statement as Sitescope could not process the length for some reason.  Good luck

 

 

Dennis Handly
Acclaimed Contributor.

Re: Sitescope 11.11 Monitoring Solaris 10 Server Physical Memory Monitoring

>I also had to shorten the statement

 

A lot of your grep awk pipelines can be shortened:

command=echo `/usr/sbin/prtconf | awk '/Mem/ {print $3*1024;}'`" "`sar -r 1 1 | tail -1 | awk '{print $2;}'`" "`getconf PAGE_SIZE` | awk '{print $1,$2*$3/1024;}'

 

grep -v -e freemem -e SunOS -e "^$"