Unix Linux Two state monitoring

Question

Unix Linux Two state monitoring

Ravi Bhushan 126

Hi All,

I have created two monitor for scripts through Unix/Linux two state monitor. The output value set for the alert is 15000 and for the second is 11. If it goes above this it will generate an alert, but for the threshold defined 15000 it is throwing alert even if the output is 50 and similar is going with the second monitor. It is generating alert for the value less than 11 also. Below is the screen shot

![59602-image.png ]3

Please help me on this.

Regards,
Ravi Bhushan

3 answers

Your answer

Answer 1

CyrAz 5,181

Chances are you are facing a very common issue when creating workflows in the console : the fields are typed as "Strings", not "Integers". Therefore greater/lower comparision are not working as expected.
Read here : https://argonsys.com/microsoft-cloud/library/using-integers-and-other-non-string-data-types-in-rules-and-monitors/

Ravi Bhushan 126 Reputation points

2021-01-25T15:24:54.817+00:00

Hi @CyrilAzoulay,

I have changed the value from strings to Integers but now it is not throwing any alert. Even for much higher value the status is showing healthy but it should have raised a critical alert.

Regards,
Ravi
CyrAz 5,181 Reputation points

2021-01-25T17:00:26.693+00:00

Then maybe there is a different issue in the code. Can you post the full XML code ?
Ravi Bhushan 126 Reputation points

2021-01-25T18:37:35.04+00:00

60381-web.xml

I have attached the xml file.
I am getting issue with monitor error log monitoring and load monitoring. These are newly created.

Regards,
Ravi

Answer 2

CyrAz 5,181

I had a quick look at the XML and it seems to be OK, however I also just noticed in the screenshot in your first post that StdOut contains at least two lines : a number and then a "last login" date.
So that's not an integer, and if you have a look at the Operations Manager event log in your SCOM Management Servers you'll probably find errors about the workflow failing to convert data to integer type or something like that.

From there you have two options :

Modify the .sh scripts so they will output an integer ONLY.
Modify the SCOM workflow to include a powershell parsing of stdout. In that case I would recommend using the Unix authoring library :
https://social.technet.microsoft.com/wiki/contents/articles/18084.scom-2012-unixlinux-monitoring-with-scripts.aspx#Handling_complex_shell_command_script_output_with_PowerShell_scripts

Ravi Bhushan 126 Reputation points

2021-01-26T15:36:12.86+00:00

The output was from the task which we created for the testing purpose. The original value for this is an integer value. Attached is the output of the executed script.
CyrAz 5,181 Reputation points

2021-01-26T15:53:12.97+00:00

When the agent runs a command, it runs it using the following syntax :

/bin/sh -c "command"

That may produce a different output than the one you're getting when running bash script.sh, so could you try running precisely the following command :

/bin/sh -c "/opt/MastekAutomation/KongErrorCount.sh"

Answer 3

CyrAz 5,181

Also please try running it using the scom agent account, not yours; and even better try running it as a network job using ssh as this will reflect more closely the agent actual behavior :

ssh scomuser@localhost '/bin/sh -c "/opt/MastekAutomation/KongErrorCount.sh"'

Ravi Bhushan 61 Reputation points

2021-02-19T15:14:27.237+00:00

We tried with the same but still getting the same issue. In SCOM console it is giving
the output is a string "54/nLast login: (...):
CyrAz 5,181 Reputation points

2021-02-19T20:48:24.397+00:00

I'm not following, you said previously you were sure the output was an integer. If it's a string, you obviously can't make a numerical comparison
Ravi Bhushan 61 Reputation points

2021-02-20T20:18:56.933+00:00

The Output on unix machine is giving integer value. But on SCOM console it is giving the output as a string "54/nLast login.
CyrAz 5,181 Reputation points

2021-02-20T22:32:21.237+00:00

Could you show us the content of this .sh script?
Ravi Bhushan 61 Reputation points

2021-02-22T09:10:38.257+00:00

!/bin/bash

StartDate=$(date --date="-15 min" "+%Y/%m/%d %H:%M:%S")
EndDate=$(date "+%Y/%m/%d %H:%M:%S")
path="/usr/local/kong/logs/error.log"

"grep" Command will display all lines having [error] in error log which is then

piped to cut command

"cut" command will just cut this output as per the characters mentioned, here it

is cut based on '['

Output of cut then piped to 'awk' which will be delimed by ":" and $1,$2 and $3

will be fields having values of "Date,Hours","Minutes" and "Seconds".

This output will be then piped to awk command in which we have assigned values o

f start and end datetime.

These values are then compared with the input $0 (Whole Input) which is then pip

ed to "wc"

StartDate is 30 minute prior to the Current datetime (EndDate)

"wc" will count number of lines

This script will output the number of errors in last 15 minutes
Ravi Bhushan 61 Reputation points

2021-02-22T09:12:17.013+00:00

[*********************]# more /opt/MastekAutomation/KongErrorCount.sh |wc -l
18
Ravi Bhushan 61 Reputation points

2021-02-26T11:31:22.513+00:00

One more thing we observed while testing is the scripts working correctly if the range is set to [0..9] and not with higher integers.
Ravi Bhushan 61 Reputation points

2021-03-02T10:06:47.207+00:00

@CyrilAzoulay ...Please help me on this.
CyrAz 5,181 Reputation points

2021-03-02T11:11:38.667+00:00

I'm sorry but I've already told you all I knew about this.
Moreover, your last posts are really not that helpful : the .sh script is unreadable and I have no idea what you're talking about with "the range is set to [0..9]".

Maybe you could try to be clearer in your explanations...
Ravi Bhushan 61 Reputation points

2021-03-02T12:04:58.317+00:00

Hi Cyril,

If we are checking it with the value in between 1-9 then it is working properly where as if we set the value greater than 10 it starts to generate alert even if the output value is less than 10.

Regards,
Ravi
CyrAz 5,181 Reputation points

2021-03-02T13:14:18.773+00:00

So we're always heading back to the same thing ; the script output is not an integer and/or the fields used for the comparison in your MP are not typed as integers.

Share via

Unix Linux Two state monitoring

3 answers

!/bin/bash

"grep" Command will display all lines having [error] in error log which is then

"cut" command will just cut this output as per the characters mentioned, here it

Output of cut then piped to 'awk' which will be delimed by ":" and $1,$2 and $3

This output will be then piped to awk command in which we have assigned values o

These values are then compared with the input $0 (Whole Input) which is then pip

StartDate is 30 minute prior to the Current datetime (EndDate)

"wc" will count number of lines

This script will output the number of errors in last 15 minutes

Your answer