Solaris CPU Utilization

JessSewpersadh · May 25, 2022, 12:25pm

Good day,

We upgraded to Check_Mk 2.0p22 a month ago from 1.6.0p25. Services were monitored using agent and snmp. Once the upgrade was done I noticed that memory utilization was being displayed incorrectly as the data from snmp was incorrect and clashing with the agent output so we disabled snmp when we upgraded to the new agents. However, I have noticed that the CPU Utilization has disappeared from the checks. I discussed this was monitored via snmp rule - hr_cpu. I have enabled the rule for Stat_grab to use the agent and baked a new agent and deployed it to a server to test, however i notice that stat_grab has not been picked up by the agent and therefore the agent is still not displaying the CPU Utilization. I do not see an actual rule for CPU Utilization. Please advise if i need to deploy the stat_grab.txt plug-in to each server manually to pick up the CPU Utilization.

Thank you
Jess

davidwayne · May 26, 2022, 3:27pm

SNMP + Agent = is not a good idea. I faced a lot of problems with this and at the end of the day it also doesn’t make sense to use SNMP when agent is already available as the metrics from agent are always better than SNMP and you have the freedom to do whatever you want to do with the agent using local checks + agent plugins.

Do you even have statgrab installed on your Solaris boxes?
If yes, then please donwload the agent output and look for a section that starts with :
<<statgrab_something>>>

You can also look at the script and try to run the commands individually in the script to see if you receive any output from this command:

github.com

Checkmk/checkmk/blob/master/agents/plugins/hpux_statgrab#L22


      
          CMK_VERSION="2.3.0b1"
          
          # this is for users who compiled statgrab on hp-ux.
          # note you'll need a 0.18+ version, from their github page at
          # https://github.com/i-scream/libstatgrab
          # flags used for compiling - disable documentation, examples and set*id
          
          if command -v statgrab >/dev/null; then
              if statgrab const. cpu. general. mem. page. proc. swap. user. >"/tmp/statgrab.$$" 2>/dev/null; then
                  for s in proc cpu page; do
                      echo "<<<statgrab_$s>>>"
                      grep "^$s\." "/tmp/statgrab.$$" | cut -d. -f2-99 | sed 's/ *= */ /'
                  done
          
                  echo '<<<statgrab_mem>>>'
                  grep -E "^(swap|mem)\." "/tmp/statgrab.$$" | sed 's/ *= */ /'
          
                  echo '<<<uptime>>>'
                  grep -E "^general\.uptime" "/tmp/statgrab.$$" | sed 's/.* //'
          
              fi

JessSewpersadh · May 26, 2022, 3:43pm

Hello Wayne,
I checked the agent output and it does not seem like statgrab was copied to the server as a plug-in when we installed the agent package. Does this mean I need to add the file manually to each Solaris host in order for to work?

chauhan_sudhir · May 26, 2022, 4:57pm

The main question would be : is statgrab utility even installed/compiled on your Solaris host? If yes, try to run the commands manually inside the bash script that @davidwayne mentioned and see if that even gives you any output or not. Or just run the bash script manually on the command line.

JessSewpersadh · May 27, 2022, 7:11am

Hello Sudhir,
I will arrange with the administrator to run the commands or the script and see what output we get.

Regards
Jess

JessSewpersadh · May 27, 2022, 7:59am

Hi Sudhir / Wayne,

We copied the Stat_grab script to a host and ran it and got the below output:

root@db2srsqalz1:/tmp# ksh -x ./hpux_statgrab

:
CMK_VERSION=2.0.0p22
which statgrab
1> /dev/null
root@db2srsqalz1:/tmp#

JessSewpersadh · May 27, 2022, 8:04am

I am going to place a call as we are missing crucial information with the monitoring…

Regards
Jess

davidwayne · May 27, 2022, 8:59am

This looks like the bash script doesn’t return anything.

JessSewpersadh · May 27, 2022, 9:02am

JessSewpersadh:

We upgraded to Check_Mk 2.0p22 a month ago from 1.6.0p25. Services were monitored using agent and snmp. Once the upgrade was done I noticed that memory utilization was being displayed incorrectly as the data from snmp was incorrect and clashing with the agent output so we disabled snmp when we upgraded to the new agents. However, I have noticed that the CPU Utilization has disappeared from the checks. I discussed this was monitored via snmp rule - hr_cpu. I have enabled the rule for Stat_grab to use the agent and baked a new agent and deployed it to a server to test, however i notice that stat_grab has not been picked up by the agent and therefore the agent is still not displaying the CPU Utilization. I do not see an actual rule for CPU Utilization. Please advise if i need to deploy the stat_grab.txt plug-in to each server manually to pick up the CPU Utilization.

No it did not…we ran the script as it and not the commands manually.

davidwayne · May 27, 2022, 9:40am

I would also suggest to try running the commands manually inside the script. If the script doesn’t return anything then the agent won’t be able to provide with this information.
Can you share the output of this command:

which statgrab

JessSewpersadh · May 27, 2022, 9:47am

This is what the commands came back with. Nothing else.

CMK_VERSION=2.0.0p22
which statgrab
1> /dev/null
root@db2srsqalz1:/tmp#

davidwayne · May 27, 2022, 9:48am

Did you ran this ? If not, then please do so:

root@db2srsqalz1:/tmp# which statgrab

JessSewpersadh · May 27, 2022, 9:51am

it was run and no output came back.

davidwayne · May 27, 2022, 9:52am

Then please install statgrab on your Solaris box if you want to use the script.

JessSewpersadh · May 27, 2022, 11:57am

Thanks. Arranging that with the Unix administrator to install now.

JessSewpersadh · May 27, 2022, 3:23pm

Worked with the Solaris team and installed the package. Tested it on a few hosts and can see CPU Utilization now.

Thanks Wayne and Sudhir.

appreciate your assistance.

regards
Jess

davidwayne · May 27, 2022, 5:28pm

Glad to hear that its fixed now. Please mark this as resolved.

system · May 27, 2023, 5:29pm

This topic was automatically closed 365 days after the last reply. New replies are no longer allowed. Contact an admin if you think this should be re-opened.