Gaps in Graph, Logs state my key stops working for periods of time..

[HaveDill]

Odd issue. Randomly i'll have some machines stop sending Zabbix values of one of their performance counters - and then will start sending again... the log files show that for these gaps, it appears the key is unsupported for random breaks in time?

What's causing this? The polling interval is 30 seconds.

  2752:20150113:140014.426 active check "perf_counter[\Processor(_Total)\% Processor Time]" is not supported

[LenR]

Are the items agent or agent_active? If agent, check your timeout value, maybe the client can't respond before the timeout occurs. What do the logs show for that host? Network lost? Is it only perf_counter items, what about items provided by the zabbix agent? Is it all hosts or just one?

[HaveDill]

Are the items agent or agent_active? If agent, check your timeout value, maybe the client can't respond before the timeout occurs. What do the logs show for that host? Network lost? Is it only perf_counter items, what about items provided by the zabbix agent? Is it all hosts or just one?

It's agent active and it has happened on maybe 2-3 machines out of 300 or so i have setup so far.

I also have the agents timeout time set to 20, so hopefully that isn't the case? Seems to only be happening for certain values. I'd rather this stay as an active item though because i want these values pushed as soon as they are collected.

[tchjts1]

It's agent active and it has happened on maybe 2-3 machines out of 300 or so i have setup so far.

So it sounds like you have been adding in more monitored hosts?
Are you still running with the default values in your zabbix_server.conf ?

Would you mind having a look at this post https://www.zabbix.com/forum/showthread.php?t=47781 and then give us a screenshot of your 3 graphs like I show in that post, using a 1 day view?

[HaveDill]

So it sounds like you have been adding in more monitored hosts?
Are you still running with the default values in your zabbix_server.conf ?

Would you mind having a look at this post https://www.zabbix.com/forum/showthread.php?t=47781 and then give us a screenshot of your 3 graphs like I show in that post, using a 1 day view?

No i've modified the server.conf as my values/sec have gone up accordingly..
509 VPS
52,000 items
25000 triggers

http://imgur.com/a/f22ns#2

[tchjts1]

Graphs look OK, however, I would certainly increase the number of unreachable pollers you have.

You mention you have the agent timeout set to 20. Do you also have the Timeout= in zabbix_server.conf increased?

[HaveDill]

Graphs look OK, however, I would certainly increase the number of unreachable pollers you have.

You mention you have the agent timeout set to 20. Do you also have the Timeout= in zabbix_server.conf increased?

Timeout in the server.conf is hashed out, so the agent configs should be the one it is using

[tchjts1]

Timeout in the server.conf is hashed out, so the agent configs should be the one it is using

That's not how it works. A line that is preceded by the pound sign # is going to use the default setting. You would do well to create a new line under the original value, without the pound sign and make the value 15 or 20, then restart your Zabbix server process.