external script wait time

[BusteR81]

hi

i had ran a script successfully from zabbix 1.4.2

when host is healthy it will return value >0
ELSE
-1

problem: when running script from SSH zabbix server and when system is healthy, the value (>0) will be returned instantly
BUT
if system is unhealthy, the value '-1' tends to return after 15-20 sec. Therefore zabbix_server.log shows timeout has occurs on reading the ITEM value from this script.

Question: how can i INCREASE the external scripts WAIT time so that it will NOT timeout while waiting for a (-ve) value in my case ?

cheer

[BusteR81]

did a search on forums

http://www.zabbix.com/forum/showthread.php?t=6058&highlight=timeout+script

Type must be set to ZABBIX Agent and ZABBIX Agent (active). Make sure that your script is fast enough, ZABBIX won't wait longer than 3 seconds by default.

HOW can i increase the 3 sec

[BusteR81]

answers?

no answers ? i did a turn around and tried using zabbix_trapper + zabbix_sender to obtain my Radius dB status data. but the SAD thing is trapper does not accpet negative (value) even in 1.4.2 and my alarm TRIGGERs at -(values)

[ptietjens]

Argh

Too bad this was never answered. I too, now am stuck in a conundrum where I need to check that routes are up in my WAN - there is no function to do so in Zabbix, and no script I write is going to be able to transverse long routes in under 3 seconds.



back to nagios, perhaps

[BusteR81]

well good news for you, yes zabbix server wait only 3 sec to get a answer (int, float, char etc) from the script but after 3sec, the script will still be run in the background. that is, eg. (the script must return a value in 3 sec and thereafter the script may do a trace route of 1-2 min and send it the the NOC)

for your case, u can return the value after " transverse long routes" by using zabbix_sender (trapper). look at zabbix_sender for details in the zabbix manual

cheers

Too bad this was never answered. I too, now am stuck in a conundrum where I need to check that routes are up in my WAN - there is no function to do so in Zabbix, and no script I write is going to be able to transverse long routes in under 3 seconds.



back to nagios, perhaps

[ptietjens]

Trapper

Yeah, that's the solution I arrived at, only I didn't bother with the external check, I just have those few (10) traceroutes done and sent to Trapper via CRON and it works fairly well.

Perhaps a note about the three second limitation should be somewhere in the docs, eh?

[lamont]

scanning through the sources, i suspect that the "Timeout=" variable in zabbix_server.conf applies to external checks as well as agent checks... not positive, however....

[lamont]

I just bumped my Timeout= up to 10 and in the config table in the database decreased refresh_unsupported down to 60. I also took my external script and wrapped the whole thing with an alarm timeout (and validated that the libraries that i'm using call select() for a timeout and don't use SIGARLM themselves).