elein <elein at varlena.com> writes:
>> Failover:
>> I think the master failover section should have a subsection on automatic 
>> failover that discuss the issues with having the failover decision left to 
>> a something like heartbeat and instructions on how to set it up if thats 
>> how the user decides to go.
>
> I think this would go in part of the main documentation rather than the
> tutorial.  I am firmly in the camp of having humans call a failover. No one
> has ever convinced me that there is a tool that can asses a random failure,
> as distinct from a network blip.  So let's argue elsewhere and keep it simple
> in the tutorial. 

It seems to me that it's worth pointing out that it's dangerous to use
failover automagically unless you have a good deal of "extra magic"
around to help you, like hardware that lets you say "never let that
host come up again on that IP without human intervention."

Automated failover becomes reasonable when you have an environment
that allows you to keep failed nodes from popping back up, but that
requires more than just vague assurances...
-- 
(reverse (concatenate 'string "ofni.sailifa.ac" "@" "enworbbc"))
<http://dba2.int.libertyrms.com/>
Christopher Browne
(416) 673-4124 (land)