Plugins provide an option to extend the monitoring capabilities of Zabbix. The plugins are written in Go programming language and supported for Zabbix agent 2 only. They provide an alternative to loadable modules (written in C), and other methods for extending Zabbix functionality, such as user parameters (agent metrics), external checks (agent-less monitoring), and system.run[]
Zabbix agent item .
The following features are specific to agent 2 and its plugins:
All metrics supported for Zabbix agent 2 are collected by plugins. The following plugins for Zabbix agent 2 are available out-of-the-box:
Plugin name | Description | Supported item keys | Comments |
---|---|---|---|
Agent | Metrics of the Zabbix agent being used. | agent.hostname, agent.ping, agent.version | Supported keys have the same parameters as Zabbix agent keys. |
Ceph | Ceph monitoring. | ceph.df.details, ceph.osd.stats, ceph.osd.discovery, ceph.osd.dump, ceph.ping, ceph.pool.discovery, ceph.status |
Supported keys can be used with Zabbix agent 2 only. |
CPU | System CPU monitoring (number of CPUs/CPU cores, discovered CPUs, utilization percentage). | system.cpu.discovery, system.cpu.num, system.cpu.util | Supported keys have the same parameters as Zabbix agent keys. |
Docker | Monitoring of Docker containers. | docker.container_info, docker.container_stats, docker.containers, docker.containers.discovery, docker.data_usage, docker.images, docker.images.discovery, docker.info, docker.ping |
App Docker monitoring template is available. Supported keys can be used with Zabbix agent 2 only. |
File | File metrics collection. | vfs.file.cksum, vfs.file.contents, vfs.file.exists, vfs.file.md5sum, vfs.file.regexp, vfs.file.regmatch, vfs.file.size, vfs.file.time |
Supported keys have the same parameters as Zabbix agent keys. |
Kernel | Kernel monitoring. | kernel.maxfiles, kernel.maxproc | Supported keys have the same parameters as Zabbix agent keys. |
Log | Log file monitoring. | log, log.count, logrt, logrt.count | Supported keys have the same parameters as Zabbix agent keys. |
Memcached | Memcached server monitoring. | memcached.ping, memchached.stats | App Memchached monitoring template is available. Supported keys can be used with Zabbix agent 2 only. |
Modbus | Reads Modbus data. | modbus.get | Supported keys have the same parameters as Zabbix agent keys. See also: - Plugin documentation - Plugin configuration parameters (Unix/Windows) |
Mongo DB | Monitoring of MongoDB servers and clusters (document-based, distributed database). | mongodb.collection.stats, mongodb.collections.discovery, mongodb.collections.usage, mongodb.connpool.stats, mongodb.db.stats, mongodb.db.discovery, mongodb.jumbo_chunks.count, mongodb.oplog.stats, mongodb.ping, mongodb.rs.config, mongodb.rs.status, mongodb.server.status, mongodb.sh.discovery |
Supported MongoDB versions: 3.6, 4.0, 4.2, 4.4. Requires a local read-only user in the admin database - follow the instructions for your configuration: STANDALONE - create a user for each single MongoDB node; REPLICASET - create a user on the primary node of the replica set; SHARDING - create a user for each shard in the cluster (on the primary node of the replica set). Then, create the same user on a mongos router. It will be automatically passed to configuration servers. use admindb.auth("admin", "<ADMIN_PASSWORD>")db.createUser({"user": "zabbix","pwd": "<PASSWORD>","roles": [{ role: "readAnyDatabase", db: "admin" },{ role: "clusterMonitor", db: "admin" },]}) Supported keys can be used with Zabbix agent 2 only.Supported since 5.2.6 |
MQTT | Receives published values of MQTT topics. | mqtt.get | Supported keys can be used with Zabbix agent 2 only. See also: - Plugin documentation - Plugin configuration parameters (Unix/Windows) |
MySQL | Monitoring of MySQL and its forks. | mysql.db.discovery, mysql.db.size, mysql.get_status_variables, mysql.ping, mysql.replication.discovery, mysql.replication.get_slave_status, mysql.version |
DB MySQL by Zabbix agent 2 monitoring template is available. Supported keys can be used with Zabbix agent 2 only. Default configuration: URI=tcp://localhost:3306 , username=root , ''password= ''. |
NetIf | Monitoring of network interfaces. | net.if.collisions, net.if.discovery, net.if.in, net.if.out, net.if.total | Supported keys have the same parameters as Zabbix agent keys. |
Oracle | Oracle Database monitoring. | oracle.diskgroups.stats, oracle.diskgroups.discovery, oracle.archive.info, oracle.archive.discovery, oracle.cdb.info, oracle.custom.query, oracle.datafiles.stats, oracle.db.discovery, oracle.fra.stats, oracle.instance.info, oracle.pdb.info, oracle.pdb.discovery, oracle.pga.stats, oracle.ping, oracle.proc.stats, oracle.redolog.info, oracle.sga.stats, oracle.sessions.stats, oracle.sys.metrics, oracle.sys.params, oracle.ts.stats, oracle.ts.discovery, oracle.user.info |
DB Oracle by Zabbix agent 2 monitoring template is available. Install the Oracle Instant Client before using the plugin. Supported keys can be used with Zabbix agent 2 only. Default configuration: URI=tcp://localhost:1521 , service=XE (service name).Functionality of the plugin can be extended with custom user-defined queries - see plugin documentation for configuration examples. |
PostgreSQL | Monitoring of PostgreSQL and its forks. | pgsql.ping, pgsql.db.discovery, pgsql.db.size, pgsql.db.age, pgsql.database.bloating_tables, pgsql.replication_lag.sec, pgsql.replication_lag.b, pgsql.replication.count, pgsql.replication.status, pgsql.replication.recovery_role, pgsql.cache.hit, pgsql.connections, pgsql.archive, pgsql.bgwriter, pgsql.dbstat.sum, pgsql.dbstat, pgsql.wal.stat, pgsql.locks, pgsql.pgsql.oldest.xid, pgsql.uptime |
DB PostgreSQL by Zabbix agent 2 monitoring template is available. Supported keys can be used with Zabbix agent 2 only. |
Proc | Process CPU utilization percentage. | proc.cpu.util | Supported key has the same parameters as Zabbix agent key. |
Redis | Redis server monitoring. | redis.config, redis.info, redis.ping, redis.slowlog.count | DB Redis monitoring template is available. Supported keys can be used with Zabbix agent 2 only. |
Smart | S.M.A.R.T. monitoring. | smart.attribute.discovery, smart.disk.discovery, smart.disk.get | Sudo/root access rights to smartctl are required for the user executing Zabbix agent 2. The minimum required smartctl version is 7.1. Supported keys can be used with Zabbix agent 2 only on Linux/Windows, both as a passive and active check. Supported since Zabbix 5.2.5. |
Swap | Swap space size in bytes/percentage. | system.swap.size | Supported key has the same parameters as Zabbix agent key. |
SystemRun | Runs specified command. | system.run | Supported key has the same parameters as Zabbix agent key. |
Systemd | Monitoring of systemd services. | systemd.unit.discovery, systemd.unit.get, systemd.unit.info | Supported keys can be used with Zabbix agent 2 only. |
TCP | TCP connection availability check. | net.tcp.port | Supported key has the same parameters as Zabbix agent key. |
UDP | Monitoring of the UDP services avaiability and performance. | net.udp.service, net.udp.service.perf | Supported keys have the same parameters as Zabbix agent keys. |
Uname | Retrieval of information about the system. | system.hostname, system.sw.arch, system.uname | Supported keys have the same parameters as Zabbix agent keys. |
Uptime | System uptime metrics collection. | system.uptime | Supported key has the same parameters as Zabbix agent key. |
VFSDev | VFS metrics collection. | vfs.dev.discovery, vfs.dev.read, vfs.dev.write | Supported keys have the same parameters as Zabbix agent keys. |
Web | Web page monitoring. | web.page.get, web.page.perf, web.page.regexp | Supported keys have the same parameters as Zabbix agent keys. |
ZabbixAsync | Asynchronous metrics collection. | net.tcp.listen, net.udp.listen, sensor, system.boottime, system.cpu.intr, system.cpu.load, system.cpu.switches, system.hw.cpu, system.hw.macaddr, system.localtime, system.sw.os, system.swap.in, system.swap.out, vfs.fs.discovery |
Supported keys have the same parameters as Zabbix agent keys. |
ZabbixStats | Zabbix server/proxy internal metrics or number of delayed items in a queue. | zabbix.stats | Supported keys have the same parameters as Zabbix agent keys. |
ZabbixSync | Synchronous metrics collection. | net.dns, net.dns.record, net.tcp.service, net.tcp.service.perf, proc.mem, proc.num, system.hw.chassis, system.hw.devices, system.sw.packages, system.users.num, vfs.dir.count, vfs.dir.size, vfs.fs.get, vfs.fs.inode, vfs.fs.size, vm.memory.size. |
Supported keys have the same parameters as Zabbix agent keys. |
Common configuration principles and best practices are described in this section.
For specific examples of plugin configuration see also:
All plugins are configured using Plugins.* parameter of the Zabbix agent 2 configuration file. Unlike other agent parameters, it is not a key/value type of parameter. It is a separate section where specific parameters of the plugin can be described. Each parameter should have the following structure:
Plugins.<PluginName>.<Parameter>=<Value>
Parameter names should adhere to the following requirements:
Named sessions represent an additional level of plugin parameters and can be used to define separate sets of authentication parameters for each of the instances being monitored. Each named session parameter should have the following structure:
Plugins.<PluginName>.<SessionName>.<Parameter>=<Value>
A session name can be used as a connString item key parameter instead of specifying a URI, username, and password separately. In item keys, the first parameter can be either a connString or a Uri. If the first key parameter matches a session name specified in the configuration file, the check will be executed using named session parameters. If the first key parameter doesn't match any session name, it will be treated as a Uri.
Note, that:
The list of available named session parameters depends on the plugin, see Zabbix agent 2 (UNIX) / Zabbix agent 2 (Windows) for details.
Example: Monitoring of two instances “MySQL1” and “MySQL2” can be configured in the following way:
Plugins.Mysql.Sessions.MySQL1.Uri=tcp://127.0.0.1:3306
Plugins.Mysql.Sessions.MySQL1.User=<UsernameForMySQL1>
Plugins.Mysql.Sessions.MySQL1.Password=<PasswordForMySQL1>
Plugins.Mysql.Sessions.MySQL2.Uri=tcp://127.0.0.1:3307
Plugins.Mysql.Sessions.MySQL2.User=<UsernameForMySQL2>
Plugins.Mysql.Sessions.MySQL2.Password=<PasswordForMySQL2>
Now, these names may be used as connStrings in keys instead of URIs:
If a parameter required for authentication is not provided in an item key or in the named session parameters, the plugin will use a hardcoded default value.
Some plugins support gathering metrics from multiple instances simultaneously. Both local and remote instances can be monitored. TCP and Unix-socket connections are supported.
It is recommended to configure plugins to keep connections to instances in an open state. The benefits are reduced network congestion, latency, and CPU and memory usage due to the lower number of connections. The client library takes care of this.
Time period for which unused connections should remain open can be determined by Plugins.<PluginName>.KeepAlive parameter.
Example: Plugins.Memcached.KeepAlive
A plugin is a Go package that defines the structure and implements one or several plugin interfaces (Exporter, Collector, Runner, Watcher):
Exporter is the simplest interface that performs a poll and returns a value (values), nothing, error. It accepts a preparsed item key, parameters and context. Exporter interface is the only interface that can be accessed concurrently. All other plugin interface access is exclusive and no method can be called when a plugin is already performing some task. Also, there is a limit of 100 maximum concurrent Export() calls per plugin, which can be reduced as necessary for each plugin.
Collector is used when a plugin needs to collect data at regular intervals. This interface usually is used together with the Exporter interface to export the collected data.
Configurator is used to provide plugin its configuration parameters from the agent 2 configuration file.
Runner interface provides the means of performing some initialization when a plugin is started (activated) and deinitialization when a plugin is stopped (deactivated). For example, a plugin could start/stop some background goroutine by implementing the Runner interface.
Watcher allows the plugin to implement its own metric polling, without using the agent's internal scheduler, for example in trap-based plugins.
Plugins by default are inactive and activated only when a metric provided by the plugin is being monitored.
Plugins are located in the plugin directory tree, grouped by meaning, for example plugins/system/uptime/uptime.go
.
A plugin must import the zabbix.com/pkg/plugin
package.
A plugin must define the structure and embed the plugin.Base
structure.
A plugin must implement one or several plugin interfaces.
func (p *Plugin) Export(key string, params []string, ctx plugin.ContextProvider) (result interface{}, err error) {
if len(params) > 0 {
p.Debugf("received %d parameters while expected none", len(params))
return nil, errors.New("Too many parameters")
}
return time.Now().Format(time.RFC3339)
}
A plugin must register itself during initialization.
func init() {
plugin.RegisterMetrics(&impl, "Time", "system.time", "Returns time string in RFC 3999 format.")
}
where RegisterMetrics
parameters are:
If logging is necessary the plugin must use the logging functionality provided by plugin.Base
(see the example above). It's basically a wrapper around standard logging, but it will prefix log messages with [<plugin name>].