Zabbix

Links

• Homepage
• compilation instructions
• Documentation
• Custom scripts
• Various scripts to automate tasks in Zabbix
• Tuning mysql for zabbix
• https://huyabbix.com
• Migrating zabbix database with minimal downtime
• Bug tracker
• Clean up database
• Zabbix on RHEL/Centos
• Grafana

FAQ

=cannot connect to IPMI host: [22] Operation canceled

Usually temporary because of broken ipmi lib, ignore it

=cannot connect to IPMI host: [125] Operation canceled

possibly authentication methor

Discover: value must be a JSON object

Could mean you need to escape slashes

Install recent zabbix on CentOS/RHEL

rpm -ivh https://repo.zabbix.com/zabbix/3.4/rhel/7/x86_64/zabbix-release-3.4-2.el7.noarch.rpm
yum install zabbix-agent

Discovery data example

Output of a discovery script should look like:

{"data":[
  {"{#VAR1}":"value11","#{VAR2":"value12"},
  {"{#VAR1}":"value21","#{VAR2":"value22"}
]}

Backing up tables

https://www.zabbix.org/wiki/Docs/howto/mysql_backup_script

cannot send list of active checks

Most likely ServerActive is defined in agent config, while not used at all

Latest 20 issues

DEFAULT_LATEST_ISSUES_CNT in/usr/share/zabbix/include/defines.inc.php

Zabbix unreachable poller processes more than 75% busy

Increase StartPollersUnreachable

More than 100 items having missing data for more than 10 minutes

Could be high load. Also check Administration->Queue

Zabbix escalator processes more than 75% busy

probably high system load overall

Check agent

zabbix_get -s my.host.com -k agent.version

ZBX_NOTSUPPORTED

Could be anything, enable logging on agent. It could be version mismatch. Check

zabbix_get -s yourhost -k agent.version

If that works, you're calling for an undefined or unsupported key.

Incorrect trigger expression. Host "xx" does not exist or you have no access to this host.

Means there's no related item.

zabbix_get returns nothing

best look at log on agent side

run playbook on single host

ansible_playbook -l somehost somplay.yml

Zabbix server is not running: the information displayed may not be current

Might be selinux: http://sysads.co.uk/2013/11/zabbix-server-running-alert/

vmware.hv.cpu.usage[{$URL},{HOST.HOST}]" became not supported: Couldn't resolve host name

Set macro {$URL} to https://your.ip/sdk/ (shouldn't discovery figure that out from {$HOST} ?

unsupported item key

This might mean it's expecting a value from the script you're calling.

echo 1

ansible or API not showing host groups

Permissions!! See Administration->User Groups