Can someone direct me to information on exact browsers behavior when browser gets multiple A records for a given hostname (say ip1 and ip2), and one of them is not accessible.

I interested in EXACT details, like (but not limited to):

1. Will browser get 2 IPs from OS, or it will get only one ?
2. Which ip will browser try first (random or always the first one) ? Now, let's say browser started with the failed ip1
3. For how long will browser try ip1 ?
4. If user hits "stop" while it waits for ip1, and then clicks refresh
   • which IP will browser try ?
5. What will happen when it times-out - will it start trying ip2 or give error ? (And if error, which ip will browser try when user clicks refresh).
6. When user clicks refresh, will any browser attempt new DNS lookup ?

Now let's assume browser tried working ip2 first.

1. For the next page request, will browser still use ip2, or it may randomly switch ips ?
2. For how long browsers keep IPs in their cache ?
3. When browsers sends a new DNS request, and get SAME ips, will it CONTINUE to use the same known-to-be-working IP, or the process starts from scratch and it may try any of the two ?

Of course it all may be browser dependent, and may also vary between versions and platforms, I'd be happy to have maximum of details.

The purpose of this - I'm trying to understand what exactly users will experience when round-robin DNS based used and one of the hosts fails.

Please, I'm NOT asking about how bad DNS load balancing is, and please refrain from answering "don't do it", "it's a bad idea", "you need heartbeat/proxy/BGP/whatever" and so on.