External script goes unsupported, then supported, over and over

[chip.maurer]

I have an external script item configured in my server. It is a shell script that calls a python script, checks the result, then returns the value and exits. It's pretty simple:

root@zabbix-srv:/etc/zabbix/externalscripts# cat ecs.check.directory_tables.COKE
#!/bin/bash
trap '' SIGINT SIGTERM SIGQUIT SIGHUP SIGALRM
RESULT=`python /etc/zabbix/VS/sanitized-ecs-zabbix/scripts/check_directory_tables.COKE.py`
echo $RESULT
if [ "$RESULT" != "0" ]
then
exit 1
else
exit 0
fi

The script takes 10 seconds or so to execute, so I bumped the timeout to 30 seconds. The trigger goes supported/unsupported every 6 to 10 minutes, as seen in the server log:

2077:20160927:081624.074 item "vipr-host-10_247_179_200:ecs.check.directory_tables.COKE["{HOST.CONN}"]" became not supported: Timeout while executing a shell script.
2072:20160927:082605.490 item "vipr-host-10_247_179_200:ecs.check.directory_tables.COKE["{HOST.CONN}"]" became supported
2072:20160927:083226.360 item "vipr-host-10_247_179_200:ecs.check.directory_tables.COKE["{HOST.CONN}"]" became not supported: Timeout while executing a shell script.
2072:20160927:083606.883 item "vipr-host-10_247_179_200:ecs.check.directory_tables.COKE["{HOST.CONN}"]" became supported
2072:20160927:084623.424 item "vipr-host-10_247_179_200:ecs.check.directory_tables.COKE["{HOST.CONN}"]" became not supported: Timeout while executing a shell script.
2074:20160927:085604.755 item "vipr-host-10_247_179_200:ecs.check.directory_tables.COKE["{HOST.CONN}"]" became supported
2077:20160927:090045.365 item "vipr-host-10_247_179_200:ecs.check.directory_tables.COKE["{HOST.CONN}"]" became not supported: Timeout while executing a shell script.
2076:20160927:090606.100 item "vipr-host-10_247_179_200:ecs.check.directory_tables.COKE["{HOST.CONN}"]" became supported

So, what's going on? I checked the code, and the 'timeout' error is triggered from an EINTR on a named pipe read from the script run. The server is an Ubuntu 16.04 host, and Zabbix server is 3.2.

[asteroidyorkton]

I'd like to add to this about a item which gets timeout only on one server.

system.users.num
Not supported: Timeout in logs

For some reason zabbix_agent processes it for long time, happens like 20 times a day. I've no idea how to find its cause. I've updated the agent too. As soon as I get that not supported email, I test it with zabbix_get and it always gets me result. Only once or twice, I got timeout from zabbix_get.

Any suggestions are appreciated. Debug logs didn't help. OS logs looks normal. Not sure what to look into to find the cause.

[batchenr]

try to make in /etc/zabbix/zabbix_server.conf
Timeout=30

and restart

[chip.maurer]

Timeout is already set to 30 seconds

Forgot to mention that I had already set the timeout to max value:

root@zabbix-srv:/etc/zabbix# grep Timeout *.conf | grep -v '#'
zabbix_agentd.conf:Timeout=30
zabbix_server.conf:Timeout=30

[batchenr]

run your script with time command
time yourscript.sh to see how much time it takes

and when you run the script no errors ?
is it a script with sudo ?

[chip.maurer]

Generally takes 10-15 seconds

The script generally takes 10-15 seconds. I suppose it's possible that sometimes it takes more than 30, but it's really unlikely.

root@zabbix-srv:/etc/zabbix/externalscripts# time ./ecs.check.directory_tables.COKE
1536

real 0m11.241s
user 0m0.108s
sys 0m0.004s

[asteroidyorkton]

In my case, timeout is at its max 30 seconds. It shouldn't take more than 2 seconds for that command anyway. Of all the Ubuntu servers only this server fails (system.users.num runs who | wc -l command)

[batchenr]

maybe try to add debug level to /etc/zabbix/zabbix_server.conf and when it becomes un supported you will get more info.

if you think the script sometimes takes more then the time out - then here your answer

or sometimes the script gets an error - i dont know what the script does

[BlackCrystal]

i have this problem too. timeout=30 didnt work. is there anything else i can do?

[Atsushi]

Items that take longer than 30 seconds can not be acquired as items.

Please use cron etc. to output the result to the file and devise the mounting method such as referring to the file as the item. There is also a way to send it to the Zabbix server using the zabbix_sender command.