Option to have empty maintenance groups

[Sypher]

We manage a high amount of servers for our customers. Sometimes those servers go down and to keep track of that we'd like to put them in a maintenance group.

However, it happens that hosts are back online and are then removed from the maintenance mode. Problem is that you must have a host/group in maintenance to keep it. You cannot save it with no groups/hosts assigned.

Please remove this restriction, so we can have an empty "Server down" group.

[untergeek]

I'm afraid I don't follow. Are you saying that you're using maintenance mode to track downtime? Or you want a maintenance window that you can assign servers to whenever there's downtime?

If the latter is what you're doing, consider creating a dummy host and/or host group. You WILL need a host, however. Put that in your maintenance window and you could use that to keep it "open."

[Sypher]

I'm afraid I don't follow. Are you saying that you're using maintenance mode to track downtime? Or you want a maintenance window that you can assign servers to whenever there's downtime?

Isn't it the same?

If the latter is what you're doing, consider creating a dummy host and/or host group. You WILL need a host, however. Put that in your maintenance window and you could use that to keep it "open."

I've thought of a dummy host/group, but that is not real clean. I don't know why you cannot have empty maintenance groups, it's just a silly restriction

[untergeek]

Not if you consider that Zabbix has to query the database before alerting and has to find out if a given host is in maintenance. An empty maintenance window is kind of a non-sequitur in that case, a null result.

[Sypher]

Null result = not in maintenance.

[untergeek]

Isn't it the same?

I don't see it as the same. If I truly wanted to track outages and downtime, wouldn't I want a separate window for each incident? That's how we do it in my company.

We've also created scripts which enter maintenance windows directly to the database for us. They're actually in this forum on a different thread, somewhere.

[Sypher]

I don't see it as the same. If I truly wanted to track outages and downtime, wouldn't I want a separate window for each incident? That's how we do it in my company.

We've also created scripts which enter maintenance windows directly to the database for us. They're actually in this forum on a different thread, somewhere.

We just have one window for "unexpected downtime" (with data) so everyone knows what is wrong. For expected downtime we make separate windows.

This could probably also be done with the API, but I don't really like having so many separate tools/scripts to achieve such a simple thing.

[untergeek]

If you're just going to put hosts in and out of this "temp" maintenance window, why not just acknowledge the triggers to silence them? Why bother with a maintenance window at all?

[Sypher]

Because the servers depend on other servers, and acknowledging takes more time than placing a single host in maintenance mode