The queue displays items that are waiting for a refresh. The queue is just a logical representation of data. There is no IPC queue or any other queue mechanism in Zabbix.
Items monitored by proxies are also included in the queue - they will be counted as queued for the proxy history data update period.
Only items with scheduled refresh times are displayed in the queue. This means that the following item types are excluded from the queue:
Statistics shown by the queue is a good indicator of the performance of Zabbix server.
The queue is retrieved directly from Zabbix server using JSON protocol. The information is available only if Zabbix server is running.
To read the queue, go to Administration → Queue. Overview should be selected in the dropdown to the right.
The picture here is generally "green" so we may assume that the server is doing fine.
The queue shows one item waiting for 10 seconds and one for 5 minutes. Nice, it would be great to know what items these are.
To do just that, select Details in the dropdown in the upper right corner. Now you can see a list of those delayed items.
With these details provided it may be possible to find out why these items might be delayed.
With one or two delayed items there perhaps is no cause for alarm. They might get updated in a second. However, if you see a bunch of items getting delayed for too long, there might be a more serious problem.
Is the agent down?
Queue information from a child node is not up-to-date. The master node receives historical data with a certain delay (normally, up-to 10 seconds for inter-node data transfer), so the information is delayed.
The information from a child node also depends on:
A special internal item zabbix[queue,<from>,<to>] can be used to monitor the health of the queue in Zabbix. It will return the number of items delayed by the set amount of time. For more information see Internal items.