Actions not running for hosts in new group? For lower severities?

[colours]

Dear All,


I have been facing an apparent misbehavior that I have put lots of effort into diagnosing but that, for some reason, I can't fully grasp.

I'm posting with the hopes of having someone point me in the right direction -- I'm probably doing something wrong without being totally clear about what it could be.

Thanks in advance for any hints!


Observed behaviour

• There's a single trigger-based action configured. It's as simple as possible:
  • No conditions.
  • Send email to one address on detection.
  • Send email to same address on recovery.
• Actions do not run for:
  • Hosts in specific groups.
  • Triggers in some severities.

Platform

• Zabbix Server 4.0.29 on CentOS 7 with PostgreSQL 11.
• Zabbix Agents 4.0.29 on CentOS 7 hosts.
• ~20 monitored hosts.
• ~8 host groups.
• All host agents are configured minimally: Server + Hostname set. No active checks.

Under scrutiny

• Host "fe0.replica" on group "COLOURS VMs".
• Host "taste" on group "TEST".

Test and Expected Result

• Filling up /tmp.
• Zabbix trigger "filesystem nearly full" activates.
• Action on trigger runs.
• Email is sent.

Observed

• Works as expected with "fe0.replica".
• Does not with "taste": trigger activates, but action does not run.

More

• Moving/adding "taste" to the "COLOURS VMs" group makes things work for "taste".
• Moving "fe0.replica" out of "COLOURS VMs" to the "TEST" makes things not work: trigger raised, but action does not run.

Even more

• Restarting either host raises the "host has been restarted" trigger.
• ...but no action is run.

• Things I have tried
• Disabling/enabling hosts, triggers, and actions: no effect.
• Removing and re-adding the "taste" hosts: no effect.
• Renaming the "TEST" group to something else: no effect.
• Creating a new group and adding hosts to it: actions don't run for them.
• Removing the configured action, and reconfiguring it: no effect.
• Adding explicit "host in group X, Y, Z" to the action conditions: no effect.
• Adding explicit "Problem is not suppressed" to the action conditions: no effect.
• Confirmed that no maintenance periods are configured.
• All of the above with other groups (all ok except the "TEST" one) and hosts.
• Checked the server logs for any hints of what might be going on: found nothing useful to my eyes.
• Browsed these forums for similar issues, to no avail.

A bit of history

• This system has been recently(ish) upgraded from Zabbix 2.0.x.
• I noted this a few weeks ago when a filesystem full went undetected by me:
  • The host had been recently provisioned and put into a new group.
  • Initial diagnostics left me puzzled.
• Once I removed + re-added the action config could I get actions to run for it.
• Such "trick" is no longer working, for some reason.

Wrapping-up

For some reason actions are not running:

• For hosts in the "TEST" group (and any new group that I create).
• For triggers with severity level "info".

Can anyone see what I might be doing wrong? Any hints into diagnosing this behaviour?

Again, thanks in advance for any pointers or assistance you may provide.
Regards,
--
Tiago Montes

[tim.mooney]

Does the Zabbix user that's supposed to be notified have at least read access to the hostgroups in question?

See the highlighted section that beings with "Attention" here: https://www.zabbix.com/documentation...tion/operation

[colours]

Tim,

Thanks a lot, that's precisely the culprit -- after giving "TEST" group read permissions to the notified user's group, actions now run and the user is notified.
Feeling kind of silly about this, but I'm glad it makes sense and is simple.

Any thoughts on why INFORMATION severity triggers are not leading to actions being run?
Again thanks a lot for your input.
--
Tiago Montes