Can Zabbix do this?

[swechsler]

I'm a Database Admin for a large media company. We currently support an environment of about 200 database servers (MySQL, Sybase, Oracle and MS SQL). Our systems group has implemented Nagios and we use RT for ticket tracking.

We find Nagios to be slow and tedious to use, and its paging escalation features appear to be basic at best. Here's what I'd like to do with paging:

We have a rotating primary on-call DBA who changes every week. Additionally, we have a secondary who takes tickets if the primary is unavailable. Currently, all alerts that get sent from outside Nagios (which is most of them) get sent to everybody, which means the secondary gets as little sleep as the primary. Nagios of course does have the primary/secondary escalation feature, but we don't have it integrated into most of our systems because of the hassles of getting it configured.

Ideally, when we get an alert (via Blackberry email) I'd like the ability to have the primary to issue a command directly to the paging system via an email reply on the blackberry. Commands would be such things as:

Acknowledged (e.g. I'm checking this out, don't page me for another 20 minutes)
Pass to (another DBA)
Close (I've finished this)
Note (Here's what I did)

Ideally it would open a ticket in RT with the relevant info, then add notes to the ticket and close it once the issue is resolved.

I was originally going to write something in Perl, but if Zabbix can do this already, then so much the better.

Anyone?

Thanks,

Steve

[richlv]

1. primary/secondary - there's no direct, automated scheduling for something like that, but you can configure two users in escalations, then alternate contact information manually. a more fancy way would be to update this information from cron via api

2. opening a ticket - trivial, either via customised email or via script;

3. stopping notifications at all/for some period - should be doable, this is something you'd configure in escalations and set it as you want. note that that stopping would not be relative, but absolute - if admin acknowledges right after receiving the notification, he gets no new notifications in 20 minutes, if he does that after 15 minutes, he gets another notification after 5 minutes.

4. stopping notifications by incoming email - probably doable, although i'm not sure about the best approach currently - if the api supports acks, that might be the best choice, if not, some frontend web call.

5. pass/close/note - not directly supported. while zabbix can send out information regarding different statuses, it does not know issue identifier etc - this would be best handled inside your ticketing system itself.