Discovery rule always fails with "Active check [...] is not supported"

[emcliqz]

I'm unable to get a custom discovery rule working. The agent always indicates it is not supported. The discovery config has these parameters:

key: sqs.discover
interval: 10  (low just for testing)
macro: {#QNAME}
type: Zabbix agent (Active)

In the zabbix_agentd.conf I have:

UserParameter=sqs.discover,/etc/zabbix/sysmonitor-ec2-metrics.sh sqs-discover

I run the command manually, as the zabbix user, and get what appears to be good output.

{"data": [{"{#QREGION}": "us-east-1", "{#QNAME}": "channelsmanager_stream_graphDB"}, {"{#QREGION}": "us-east-1", "{#QNAME}": "contentupdatesstream_inbox-1"}, {"{#QREGION}": "us-east-1", "{#QNAME}": "usersignals_inbox-1"}, {"{#QREGION}": "eu-west-1", "{#QNAME}": "em-test"}]}

I use several other custom active checks, but this is my first discovery rule. What am I doing wrong?

[emcliqz]

debug log

The following is the log when I enable zabbix debugging.

20524:20130715:082115.890 In zbx_popen() command:'/etc/zabbix/sysmonitor-ec2-metrics.sh sqs-discover'
 20524:20130715:082115.890 End of zbx_popen():4
 20543:20130715:082115.890 zbx_popen(): executing script
 20523:20130715:082116.816 In update_cpustats()
 20523:20130715:082116.816 End of update_cpustats()
 20523:20130715:082117.816 In update_cpustats()
 20523:20130715:082117.817 End of update_cpustats()
 20523:20130715:082118.817 In update_cpustats()
 20523:20130715:082118.817 End of update_cpustats()
 20524:20130715:082118.890 In zbx_waitpid()
 20524:20130715:082118.890 zbx_waitpid() killed by signal 15
 20524:20130715:082118.890 End of zbx_waitpid():20543
 20524:20130715:082118.890 For key [sqs.discovery] received value [ZBX_NOTSUPPORTED]
 20524:20130715:082118.890 In process_value() key:'sysmonitor-ec2:sqs.discovery' value:'ZBX_NOTSUPPORTED'

[emcliqz]

It appears the external program is killed because it takes too long. Is there anyway to increase the timeout?

[tchjts1]

It appears the external program is killed because it takes too long. Is there anyway to increase the timeout?

There is a "Timeout=" value on both the monitored host as well as Zabbix server. Try to go with the lowest value that works for you. I have my Zabbix server Timeout set to 10.

In Zabbix server it is in zabbix_server.conf
In Zabbix host it is in zabbix_agentd.conf

If you change either of those, the Zabbix processes have to be restarted.

[emcliqz]

Thank you. That's good to know. I ended up going with a cron solution where I regularly update some data in Redis then just have the Zabbix callback read from reddis. Since I'm reading from Amazon there's no guarantee on how long it might actually take.