[TCC-Mark]

Hi guys,

I've been playing around with a failover system for VPS, but one scenario I'm considering won't be caught by the DNS failover I've put in place. I don't want to go down the route of having three servers for failover so DNS failover is the best option I think.

What I could do with, though, is being able to control DNS failover using something other than ping. In the past, my VPS has gone into read only mode where it can't write to the database (of course) which crashes all of the websites, but it still responds to pings. In this instance, of course, DNS failover isn't going to trigger.

Is there a way to get failover to work by checking a custom webpage (with a back end script) on the server, for a 1 or 0, or something along those lines?

Cheers,

Mark.

[Miikka]

Create a daemon which periodically will write to the disk, if the write operation fails then it would block pings from the testing servers to trigger failover. That's what I did previously when I was using DNS failover.

This post has been edited by Miikka: 22 November 2011 - 04:22 AM

[TCC-Mark]

Miikka, that's a really good idea, thank you!

In fact, I could make the daemon do anything I wanted by way of checks and then block pings as appropriate - brilliant!

Just out of interest, though, can the VPS.net guys confirm if there is a more official way?

I was also wondering what the timing between ping check is... sometimes I've had issues where the failover / failback doesn't seem to occur as rapidly as I'd hope. I know there is a 60 second propagation time on DNS records for failover.

Cheers,

Mark.