check_xi_service_snmp_linux_load issue

[ahmad.zuhd]

It's now 3.0 however there are many times the result is 100 because the system is actually in idle state. This results in critical alarm being sent.

[scottwilkerson]

At what percent did you want to alarm on if not above 90%?

[ahmad.zuhd]

The problem not on the logic of sending the alarm since the output is 100. But it's on the plugin output which indicate 100 in case of ideal.

We have two potions, either to fix the problem in the plugin; and send output of almost zero in case of ideal or deal with the result as not being alarm.

I have tried to use the plugin -w 80-89 -c 90-99 however it seems that the plugin is not accepting this format.

If my question is not clear please let me know.

[abrist]

If the plugin adheres to the expected format of thresholds:
http://nagiosplug.sourceforge.net/devel ... HOLDFORMAT
You may be able to specify the thresholds as:

Code: Select all

-w 90: -c 80:

Which would warn when under 90 and critical when under 80.

[ahmad.zuhd]

Again .... I dont need to change the warming and critical ranges. They must be kept more than 80 for warming and more than 90 for critical. However, plug in is indicating the usage in wrong way; it send an output of 100 in case of ideal where it must be sending a value near to zero.

Can you review the case again instead of sending repetitive post replies?

[abrist]

Can you review the case again instead of sending repetitive post replies?

Please read the thresholds document. This perl script is reporting idle time instead of load for perfdata, does it not? If so, then 100 is ideal and you need to adjust your thresholds to warn when idle goes below 90 (-w 90:) indicating a 10% or more load and report critical when idle is below 80 (-c 80:) indicating a 20% or more load - the addition of a colon on the end inverts the threshold as per the thresholds document.

Alternatively, you can alter the perl script to report the actual load and not the idle time.

[ahmad.zuhd]

the plugin reports the usage time not the idle time.

Code: Select all

[root@nms libexec]# snmpget -v2c -c public <ip-address> "1.3.6.1.4.1.2021.11.11.0"; ./check_snmp_load_wizard.pl  -H <ip-address> -C public --v2c -w 80 -c 90 -T netsc -f
UCD-SNMP-MIB::ssCpuIdle.0 = INTEGER: 97
CPU used 3.0% (<80) : OK | cpu_prct_used=3%;80;90

After looking in the code, i have found that the plugin depends ion the OID UCD-SNMP-MIB::ssCpuIdle.0 to get the usage by 100-idle, and this is what they mean by the term 100-idle

From the script:

# for Net-snmp : oid returned idle time so load = 100-idle.
if ($o_check_type eq "netsc") {$load = 100 - $load; };

printf("CPU used %.1f%% (",$load);