Dependencies and maintenance.

[quackduck]

Hi

Got 17k hosts with random names. Dependencies are set up. A parent host can have up to 1k children - and if it goes down, then up to 1k children goes down.
It does not make sense to group this complex network into host groups - we only maintain the dependencies (it's a complex and long story).

I did set the parent host in maintenance mode - it went down, and I got 1k alerts from all the children that went down.

Obviously, I'm doing something wrong since no monitoring "software" in its right mind would disregard dependencies when it comes to planned maintenance.

So, besides placing these hosts into host groups that we don't need nor want for very good reasons (it'd just be redundant work in terms of dependencies) - how do you get the dependencies to matter when you put a parent in downtime without adding each of the 1k individual hosts to the maintenance or placing these hosts in a group for the sake of effective telling maintenance mode that they're dependent on each other?

Thanks in advance.

[quackduck]

Any solution to this problem?

[tim.mooney]

Any solution to this problem?

Yes, there is a solution.

(Ask a vague question, get a vague answer).

You haven't provided enough information to be certain what the problem is, but my suspicion is that there's a timing issue with your dependency triggers. If your triggers aren't written correctly, then there's a window where the "parent" goes offline but Zabbix hasn't detected it offline yet, so any checks of the leaf nodes will fail, but they're failing before the dependency is detected.