Graphs not working as expected...

[McCall]

Hi,

I am having a few problems with my graphs. They seem to stop working and then carry on, and sometimes just top working all together!

Here is an example of them stop working and then start again:



During the time that the graph is missing, the server remains contactable as can be seen in other graphs for that server.

The next graph might be a different issue. Here is an example of a graph stopping altogether:



Note - the server only started running the Zabbix agent on the 16th, so the missing graph at the start is correct.

I have noticed that the client seems to have stopped reporting CPU usage, but can monitor disk usage etc. I don't know why its stopped reporting CPU data. Again, the server is contactable during this time.



My data is stored in InnoDB MySQL tables, and I can't repair the tables as some other posts have suggested.

Can anyone give me any clues on how to proceed?

Thanks,

McCall

[McCall]

Hi - I have also noticed that my queue keeps jumping up to between 100 and 400 items in the "More than 5 minutes column".

I think all these problems are related.

[Jeffj]

I've been having the exact same problem since upgrading to 1.4.2 I have reduce the # of host I am monitoring and increased the interval on some checks to try and reduce the load on the server but still having the problem.

If anyone has had this problem and fixed it please let us know how.

Thanks

[Jeffj]

I was hoping upgrading to 1.4.4 would fix this but it just seems to make it worse.

Maybe I upgraded wrong. I did the following to upgrade from 1.4.2.

copied the 1.4.4 frontend directory over the existing one. Recompiled the zabbix_server and replace the existing zabbix_server.exe with the newly compiled one. Is there anything else I need to do???

Anyone know any reason this may be happening?

[Jeffj]

Disabling housekeeping seems to fix the problem.

[qix]

Try to tune your datebase parameters to get the best out of your system.
The queue must be as low as possible.

I noticed that some active items can slow down the queue when the agent doesn't respond or if the agents take to long to calculate custom items.

Adjusting the amount of pollers you have running might also work.

Hope this helps,