We have already seen that Zabbix is great at monitoring hosts but Zabbix is not limited to just collecting information from other hosts. The Zabbix internal items are items Zabbix can monitor to give us some insights on what's going on under the hood. Monitoring the internals is probably not your first task when you start your setup. Zabbix is configured from the start to work well but when your installation grows you will see the need to tweak/optimize some settings.
What do we need for this topic? You need to have your Zabbix server up and running. Internal items are internal checks, so we don't need an agent. Zabbix is perfectly capable to monitor its internals without the help of an agent. You only need to make sure that you have super administration rights.
- On your host or in a template, go to items and click Create item to add a new item.
- In the Name field add a name for the item.
- For Type, we select
Zabbix internal. - For the Key, we select
zabbix[process,<type>,<num>,<state>]. Here we change the parameters as follows:zabbix[process,poller,avg,busy]. - The Type of information that Zabbix will send us back is
Numeric (float). - For Units we can add the
%. - All other parameters can remain standard.
Our Zabbix internal key consists of some options that we can and cannot change. In the case of our item process options cannot be changed. Process means we will time a process; in our case the
poller process, so we replaced type with poller. We will monitor the average for all the poller processes that are in a busy state. To accomplish, this we had to replace <num> with <avg> to calculate the average data and <state> with <busy> to tell Zabbix to do this for all processes in a busy state.
Tip
Zabbix has provided templates for the Zabbix server and proxy to monitor internal items. These templates are called App Zabbix Server and App Zabbix Proxy. It's best to link these templates and monitor their data in the Latest data page from Zabbix.
Since Zabbix 2.4, internal checks are always processed by server or proxy regardless of the host maintenance status.
Internal checks are always calculated by Zabbix server or by Zabbix proxy if the host is monitored by a proxy.
Zabbix runs as a list of processes in the background. Each process that runs is responsible for some task. If you run the ps command as shown here, you will see multiple zabbix_server processes running. When you look closer, you will see that some of them are for pollers or trappers. Even with the identical process names, they all process different internal items:
When we want to monitor our Zabbix internal processes, it's good to know that they are split up and have different responsibilities.
- Alerter: This is the process responsible for sending messages.
- Configuration syncer: This is responsible for loading configuration data from the database into the cache.
- DB watchdog: This will check if our database is available and log this when it is not the case.
- Discoverer: This process will scan the network (autodiscovery).
- Escalator: This is process the escalations.
- History syncer: This process writes data collected into the database.
- HTTP poller: This is the process needed for Website Monitoring (scenarios).
- Housekeeper: This process that deletes old data from the database.
- ICMP pinger: This process is responsible for the ping of hosts.
- IPMI poller: This process will collect data via IPMI.
- Node watcher: The process is responsible for data exchange in a distributed monitoring (deprecated since 2.4).
- Self-monitoring: The process is responsible for the collection of internal data.
- Poller: The process responsible for the collection of data from Zabbix Agent (passive) and SNMP.
- Proxy poller: The process is responsible for collecting data from passive proxies.
- Timer: This process will execute the time-dependent triggers ( nodata () ).
- Trapper: This process will accept all incoming data of active agents, Zabbix sender and active proxies.
- Unreachable: This service will contact unreachable hosts to see if they might be available again.
Tip
We can increase these values in the zabbix_server.conf file. After we change them keep in mind that you have to restart the server. Also, more processes mean that the Zabbix server will require more resources; so don't go to crazy on them but also don't be too sparse, for example, a shortage in pollers can result in items becoming unavailable for some time.
- This is only one example and there are plenty more internal items that we can monitor with Zabbix. Take a look on the Zabbix website for a full list.
https://www.zabbix.com/documentation/2.4/manual/config/items/itemtypes/internal