Queue length over 1 day after upgrading item to active

[eger]

I upgraded a bunch of my item checks to the Zabbix agent (active) as I was reading the active check is more efficient and less strain on the server (my regular checks were getting up to between 5 and 10 minutes delayed).

After changing the majority of checks to active, the queue delay is just goign higher and higher for a lot of checks.

I have tried restarting mysqld and the zabbix_server and the queue remains the same. I am running 1.8.2.

Any ideas on how to resolve this or clear the queue and start fresh?

If active checks are better can anyone explain the point or when I would want to use the regular check instead?

[tchjts1]

Are you seeing fresh data on the items you changed to active?

The possibility that you have a mismatch of the Hostname= field of your zabbix_agentd.conf file on the host, and the "Name" of the host as you have it entered in the Zabbix front end, exists. But if that were the case, you would not be seeing any data coming in for any item that was an Active agent item.

If you are not seeing fresh data coming in, check to make sure that the above two names are exactly the same.

[eger]

Actually, the items that are in the queue are NOT getting fresh data (the last check was the time that I changed to active). I did not notice this as I assumed if they weren't getting fresh data there would be an alert (this is kind of scary)!

I have already previously checked to make sure the hostnames are all correct in the config files. I just chose one host to compare with and the Server name is correct (all lower case, not sure if case matters though).

Some of the items that are not returning new data are:

Average disk read queue length
CPU utilization 1 minute average
CPU utilization 5 minute average
CPU utilization 15 minute average
Swap space free
Swap space total
Memory free
Processes count
File write bytes per second
Number of threads
Drive space free (C)

These are what I see that are not returning just by quickly running through the queue. Looks like it is happening on all the hosts.

What I will try next is enabling the debug level (4) on host and agent and see if the logs have any helpful information.

If any ideas I am all ears.

[tchjts1]

The case only matters in that it matches exactly between the host config file and what you have in the front end.

Check your zabbix_agentd.log on the host for information on why there is a problem. I suspect you are going to see something like host [xxxxxxx] not found on server.

[eger]

In my agentd log I only get this:

 3280:20100412:134459 zabbix_agentd started. ZABBIX 1.6.6 (revision 7834).
  3576:20100412:134459 zabbix_agentd collector started
  1324:20100412:134459 zabbix_agentd listener started
  2904:20100412:134459 zabbix_agentd listener started
   496:20100412:134459 zabbix_agentd listener started
  3880:20100412:134459 zabbix_agentd listener started
  1608:20100412:134459 zabbix_agentd listener started
  3348:20100412:134459 zabbix_agentd active check started [watchdog.mydomain.com:10051]
  3348:20100412:134459 Can't open jason object
  3348:20100412:134659 Can't open jason object
  3348:20100412:134859 Can't open jason object

"Can't open jason object" just repeats. I AM able to telnet to watchdog.mydomain.com:10051 just fine. I can also telnet from Zabbix to the agent on port 10050 fine.

Any other ideas?

[tchjts1]

Any other ideas?

Yep - upgrade the agent to 1.6.7 or 1.6.8, then test again.

Jeff

[eger]

Yep - upgrade the agent to 1.6.7 or 1.6.8, then test again.

Jeff

Upgraded to 1.8.2 and it fixed it!

Ahhhh.... the joy of upgrading ~50 Windows machines

[tchjts1]

Ah. I assumed from your agent log since you were using 1.6.6 agent that you were also using 1.6.6 application code.

If you are using 1.8.2 application code, then by all means upgrade your clients with the 1.8.2 agent.

[sambhu.prakash]

Well this can also happen if the time is not synced between Zabbix server and Zabbix host. In my case, chronyd service was not running on the host server so the server was out of sync. Once the time is back to sync, everything went back normal. I got rid of a 100 items from the queue