[Linux-HA] How to calculate the failover time?

[Alan Robertson]

Warren Beldad wrote:
> Hi all!
> 
> I am just newly subscribed with this mailing list.
> I would like to ask on how to calculate the fail over time. That is
> the time between the client will be disconnected and the time that it
> can now access the new active server. I have seen a lot of results in
> the list regarding their fail over time, some says 90secs, etc but i
> cannot really find how just they do it.
> My idea would be:
> In case for a resource samba share and with a common IP resource.
> 1. Create a script that will do mount/unmount and get the time that it
> fails to mount until the time it can mount again.
> 2. Create a script that will ping the IP resource and get the time
> that it fails reach to the time that it can ping back again.
> 3. A stopwatch.
> or are there any tools that actually do this?
> Any ideas would be greatly appreciated.:-))

Here's some slightly relevant information:

	The best reliable failure detection time measured with heartbeat
		is about 400ms

	Add to that the time to start the resources (run the resource
		scripts) and that's about it (ignoring STONITH)

	If you have a STONITH device in the picture that can add
		somewhere between a fraction of a second to 30 seconds
		 to your failover time (depending on your STONITH
			device/method)


-- 
     Alan Robertson <alanr at unix.sh>

"Openness is the foundation and preservative of friendship...  Let me 
claim from you at all times your undisguised opinions." - William 
Wilberforce