Zabbix 2.2.2 Maintenance Mode not working

[Kromey]

I've got a (very?) small Zabbix monitoring setup, 1 node and 1 proxy, the latter collecting data from the 4 (current) hosts on my home LAN. One such host, Odin, happens to be my file server, and one of its routine tasks is a nightly anti-virus scan (via Clam AV); on the amount of data I have there, this pegs the CPUs to max for nearly 90 minutes, which Zabbix (rightly so) freaks out about and sends me notifications. Ditto for high disk I/O.

Obviously I don't need alerts for these, so I set up a daily Maintenance window for this period, thinking that Zabbix would put my host into Maintenance mode and therefore not send me alerts during that period. No such luck.

Here's the (relevant) setup:
Action:
I disabled the default "alert admins on all media" and created a new one, thinking maybe something wasn't quite right somehow when it was auto-generated. It has the default "maintenance status not in maintenance" condition, is set to "AND", and the usual "trigger value = problem". It evidently works because I get notifications.

Maintenance:
Active from 19/02/2014 00:00 to 20/02/2034 00:00 (NB: This is not the correct date format per my locale, but is what Zabbix's frontend seems to want to use; could that be the problem?), daily period starting at 03:30 and going for 1.5 hours (until 05:00). It has my single host, Odin, which I'm trying to disable alerts for during this period. It is configured with data collection.

Last night PROBLEM alerts arrived in my inbox a little after 04:00, with OK alerts following just before and just after 05:00. The one that came after 05:00 should be expected, but the others of course are during the maintenance period.

Am I doing something wrong here? Or are maintenance periods broken in 2.2.2? I've done extensive Googling over the last few days trying to figure out what's wrong, and found lots of complaints about maintenance mode not working, but the most recent I saw was for Zabbix 1.8.4 (or .5?); surely a feature this important would have been fixed since then?

I've found a workaround that should work, but I'd rather not have to do anything as non-standard as that -- and maintenance periods seem crucial to the basic workings of a system like this, so surely I should be able to use them as designed!

[Kromey]

Okay, I think I've found the problem.

I did further testing, creating a one-time maintenance period, and at the specified time Odin went into maintenance mode. I stopped the agent and let the requisite 5 minutes pass, and while the trigger showed up under both Triggers and Events, no alerts were sent out, as expected.

I then created a new daily maintenance period, but when the time came Odin did not go into maintenance mode. On a hunch I set the maintenance period ahead 3 hours, and like magic Odin entered maintenance mode.

In PHP I set the time zone to America/Anchorage, because I'm in Alaska and I wanted my front-end to show me times I recognize; my node server, however, is on Central time. Bizarrely, when configuring a one-time maintenance period, Zabbix uses the PHP/front-end time zone, and it works as one would expect; when configuring daily periods, however, Zabbix instead assumes you're setting up the period in the server's time zone, rather than the front-end's.

This inconsistency is, to say the least, very bizarre, and the resultant behavior is unexpected. It's also very undesirable from the perspective of someone who has to monitor very geographically diverse servers, especially if said person is in a different time zone from the Zabbix node.

At least I have a solution to my issue now...