Check latency issues

[Thomas Wittmann]

Hi
now i have a question too:
i have Synagios running on a Syno DS216j, have 3 hosts with 28 services in total.
Though Nagios seems to execute all checks in time, the Check latency check always says there are at least 5 checks running >600s late.
Load and CPU checks seem just fine.
Can anyone help me about this?


Tom

[Mark Clarkson]

Hi Tom,
The 'Check latency' check is not particularly smart. It assumes that every service check is set to run every 5 minutes. I assume you have 5 checks that run less often.

Here's the help from that plugin:

------------------- snip -----------------

Usage: check_statusdat_latency [options]

 -h   :  Display this help text.
 -np  :  Don't output performance data.
 -wp  :  Warning threshold for percentage of checks that are late.
 -cp  :  Critical threshold for percentage of checks that are late.
 -wl  :  Warning threshold for average lateness of late checks.
 -cl  :  Critical threshold for average lateness of late checks.
 -i   :  Check interval of most checks.
 -s   :  Location of nagios status.dat.

Running with no options is equivalent to:
    check_statusdat_latency -wp 5 -cp 50 -wl 60 -cl 300

------------------- snip -----------------

That plugin needs an update to do it dynamically, ignoring the -i option. This should fix it (it does for me!):

Find the lines in that plugin:

    stat=`sed -n '/servicestatus {/,/ *}/p' $statusdat | \
        awk -v interval=$interval -v offset=$offset '
            BEGIN { tt=0;t=0;ttn=0;min=1000;
            "/bin/date +%s" | getline b
            b=b-offset} /current_state=[123]/ { num_warncrit=num_warncrit+1; }
            /last_check=/ {
                tot=tot+1;
                a=substr( $0, index($0, "=")+1 );
                if(a<=1) { pending=pending+1; next };

and add the section:

            /check_interval=/ {
                interval=substr( $0, index($0, "=")+1 );
                interval=interval*60;
            };

so it looks like:

    stat=`sed -n '/servicestatus {/,/ *}/p' $statusdat | \
        awk -v interval=$interval -v offset=$offset '
            BEGIN { tt=0;t=0;ttn=0;min=1000;
            "/bin/date +%s" | getline b
            b=b-offset} /current_state=[123]/ { num_warncrit=num_warncrit+1; }
            /check_interval=/ {
                interval=substr( $0, index($0, "=")+1 );
                interval=interval*60;
            };
            /last_check=/ {
                tot=tot+1;
                a=substr( $0, index($0, "=")+1 );
                if(a<=1) { pending=pending+1; next };

The above fix assumes that "check_interval" always comes before "last_check" in the status.dat file. This is a bad assumption but in reality I think it is always true. If not then some extra checks (flags) would need to be added to the code to cover those cases.

Cheers!
Mark

[Thomas Wittmann]

Hi Mark
thank you. I thought this check is for checking if all checks (hahaha 3 times check in one sentence) are executed and computed with results on time.
At the moment i have no SSH access so this works for me now: check_statusdat_latency!-s /var/cache/nagios3/status.dat -wp 30 -cp 60 -wl 600 -cl 1000
Reading the plugin help would have been too easy :-D

Cheers from "Home of quattro" ;-)

Tom