I can't recall the original thread that went off on a tangent, but there
was some talk about whether or not the mysql C libs handled failover,
I was curious myself, so I asked about this on the mysql list. Apparently
the libs don't do anything special; if you hit a db that doesn't respond,
the connection fails and that's that. Has anyone thought about building
in some simple "oh, this host is down, but I've got three more A records
for 'db.blah.com' to try, so let's try the next"?
from the mysql list:
> I have a quick question about how programs linked against the mysql C
> libraries handle the following:
> -assume three mysql hosts, say 10.0.0.2, 10.0.0.3, and 10.0.0.4
> -assume a dns name "db.example.com" that returns the following:
> [EMAIL PROTECTED] host db.example.com
> db.example.com has address 10.0.0.2
> db.example.com has address 10.0.0.3
> db.example.com has address 10.0.0.4
> If my client program repeatedly connects to "db.example.com" and my
> nameserver round-robins through those IPs, will the client also keep
> cycling through those? What is the behaviour if one of those hosts does
> not respond? Will the client application then try the next one?
The MySQL C client libarary doesn't treat this case specially.
Jeremy D. Zawodny | Perl, Web, MySQL, Linux Magazine, Yahoo!
<[EMAIL PROTECTED]> | http://jeremy.zawodny.com/