How to Have An Action Run On Hosts Where A Given Trigger Is Already “True?”

zabbixzabbix-agent

Setup: Zabbix 2.4 | RHEL 6.4

I'm configuring a method to allow Zabbix Agentd services to automatically update their own binaries when a trigger is thrown (using Git as the vessel). Two triggers are in place:

Trigger 1: "Git repo doesn't exist" – Make sure there is a git repository in place where the Zabbix Agentd's binaries are stored.
a. {Template App Zabbix Agent:vfs.file.exists[/opt/zabbix/.git/config].last()}=0
Trigger 2: "Zabbix Agent version is incorrect." – Trigger if the agent isn't the expected, compiled version. This is set as a user macro.
a. {Template App Zabbix Agent:vfs.file.cksum[/opt/zabbix/sbin/zabbix_agentd].last()}<>{$LATESTAGENT}
b. $LATESTAGENT is the cksum() value of the compiled zabbix_agentd binary.
c. Dependent on Trigger 1

Trigger 1 just ensures if Trigger 2 fires, the agent will be able to run a remote command to perform a git pull and fetch the latest binaries.

Current Situation

The triggers run fine and show my systems where the agent versions don't align. Since then, I've configured a couple actions to go with them to do the dirty work.

Trigger 1 > Run git init in the Zabbix Agentd install path.
Trigger 2 > Run git pull origin master to get the latest bins.

These are just Global Scripts I've created, so I've been able to manually test that they work by clicking any host and selecting the appropriate script for the host.

The Problem

Since I created the triggers first, they are already "triggered;" however, now that I have actions to go with those triggers, I'd like those triggers to "re-trigger" (or reinitialize or re-fire or whatever) to allow the actions to do their thing. Unfortunately, I can't really find a way to do this. Focusing on Trigger 1 (git repo exists), I tried clearing the history on that vfs.file.cksum[/opt/zabbix/sbin/zabbix_agentd] key, but that didn't affect the trigger's state. I've seen some folks mention some SQL or API work to manually modify the trigger state, but it feels like there should be something so much simpler than that!

Summary

Created triggers to deploy updated Zabbix Agentd binaries.
Created actions to go with triggers.
Can't get actions to run on triggers which are already "True"

Anyone have any idea of how I could get my triggers to "trigger again" in order to let the actions do their thing? Thanks for any help on this!

Caveats

System CM tools (e.g. Puppet) aren't an option. I need to do this within Zabbix and Git.
Cron solutions are out. Tons of servers are in play, so managing those cron jobs and changes (again, no sysCM tool) would be a manual nightmare.

Best Answer

As a workaround, you can temporarily change the trigger to "Multiple PROBLEM events generation", wait until all actions you need are executed, and then switch back.

Related Solutions

Ubuntu – Configure Zabbix to send email notifications through Exim

I figured it out! My setup was correct afterall. Here's what was happening:

Triggers are only set when a state changes, therefore all my triggers were either false OR true and were not changing state as the items I was monitoring were not changing state either.

To successfully test my triggers, I had to do this:

Make an item to ping a server every 5 sec (you can just use a default template item but I wanted results fast...)
Make an action for my server which gets executed if:

Trigger severity >= "Warning" AND Host = "My Server"
I then created a trigger which gets set if ping = 1 (which it always does...) and volia! I received an email

The problem I was having was that by creating my "test trigger" before my "test alert" the trigger was already getting set before my alert was saved. And so... i had no idea why I wasn't seeing any notifications.

Mysql – Zabbix: MySQL service monitor – invalid entry when restarting zabbix-agent

The problem with your configurations are:

You are putting the .my.cnf file at the wrong place.

Zabix agent configuration file has prescribed format and parameters defined. If you may use Include option to load additional config files, be sure it follows the same format. The zabbix documentation states it clearly:

Starting from version 1.8.6 Zabbix agent daemon will not start up if invalid (not following parameter=value notation) or unknown parameter entry is present in configuration file.

So, in your case the .my.cnf file fails to load when you put it under /etc/zabbix/zabbix_agentd.d.

You are missing the part to configure userparameter_mysql.conf file properly.

To resolve the issues:

Move the .my.cnf file from /etc/zabbix/zabbix_agentd.d directory to /etc/zabbix. And also remove any Include entry to refere to the .my.cnf file (if there is any). The content of the file may look like this:

[mysqld]
user=username
password=userpass

[mysqladmin]
user=username
password=userpass

Please make sure that the user listed here exists and have the necessary permissions in mysql.

Edit /etc/zabbix/zabbix_agentd.d/userparameter_mysql.conf file: You need to replace HOME=/var/lib/zabbix with HOME=/etc/zabbix to point to the right file (should appear three times) as mentioned in the first line of the file.
Finally restart the agent: service zabbix-agent restart