Redirect to offsite server during downtime

[ephigenie]

alright then - but its fairly easy to have some sort of zone - transfer tool (for transfering zones form one Bind to another Bind instance (if you don't use pdns) )

so at this stage it'll be easy to just create a copy of your zonefile with just an empty zone and
only 1 A record in it ...

And if i get you right - you're asking to provide a feature, to create the 2nd set of zone files by ispCP engine ? - and then ... the transfer of zones is still work in progress - so there should be something in your setup to copy zone information just abroad ...

To make the fail-over then automatically is just a matter of a few scripts as well then.
But to be honest - i thing its maybe a nice idea - but only for .01 % of our current ispCP users - even those who have hundreds of ispCP servers do have redundant-everything in their facilitys. So the possibility to have a complete disaster for a whole facility is normally below 99% - 99,9% else you're doing something seriously wrong

[kilburn]

IMHO the feature that would solve this problem is an "events system". This is, the ability to specify scripts to execute when something happens (domain added/domain deleted here).

This way, you could write a simple script that creates (or removes) the empty "failover" zone directly on the secondary server through ssh and let the panel run it when necessary. Thus, in a disaster scenario you would just need to command the backup nameserver to use the "failover" zones instead of acting as secondary for these domains.

The advantage of this "events system" is that it is a generic enough feature that would be useful for a lot of users, in many different situations.

[ephigenie]

yes for sure.
The real question that matters is : should this really go into ispCP ?
I mean there are already good frameworks to react on things (think about nagios ? )

and there of course not only sending mail - but event handlers and so on.

[kilburn]

Quote:The real question that matters is : should this really go into ispCP ?
I mean there are already good frameworks to react on things (think about nagios ? )

Hmm... maybe I obscured the idea by naming it "events system"? The thing I was proposing is supposed to provide a way to automatically take further actions when administered objects (domains, users, mails, ftp accounts, etc.) are added/updated/removed in the server.

For example, you could set up a hook that automatically creates a recurring bill in your accounting system when a new user is created. Likewise, you could create a script that, whenever a mail account is added/removed, automatically connects to your mail filtering gateway and updates the list of valid emails for your domains. The list could go on and on...

This idea has nothing to do with nagios, which provides a framework to detect "environmental" events (resource exhaustion, failing services, etc.) and react on them by alerting or taking any default actions.

[ephigenie]

its very likely that your intend will be somewhat available in upcoming versions
since when everything goes oo (not only oO ) in the codebase we'll end up with kind of objects and methods for each and everything.

Its easy then to extend the functionality for a given method with some custom code to call a billing system or something like that.

But to be honest - you've chosen the wrong thread for talking about an internal - non environmental problem related event system. Mainly because the reason for asking about such a thing was an environmental problem concrete : The facility went down because of a power outage and no website was reachable.

Idea :
Why don't you use powerdns as your 2nd DNS (it creates zones automatically in its i.e. gmysql backend.
While everything works normal, pdns receives its zone updates via axfr (yeah its unable to do ixfr - however) and
will hopefully populate its records table with your A records and so on.
In case your primary server goes down - you prepare a update statement, that replaces every A record with your failover IP. When you primary server comes up again, you just empty the records table and force a zone update on your primary - he'll then notify your pdns and transfer the zones again. (else have a table dump by the hand and just play it back -> could be easily done in 3 lines of bash code)

That way you will be able to have your dns records changed very comfortable in a short amount of time in case of a emergency.