|
hi,
> I was planning on using ldirectord again to my monitoring of the
> real-servers (just because it's been working nicely for me on
> my single
> director setup) and then letting heartbeat start and stop
> ldirectord...
> Does this sound like a good plan?
Scott looks like you're doing the same thing I did a while back, the only
difference being that you are using configure instead of ultramonkey (which
I'm using). I posted a summary of issues I ran into before, snippet of main
points...
<snip>
- clean up haresources example file a bit. I think the examples therein
aren't at all clear! The example config show things like
"#just.linux-ha.org 135.9.216.110 http" when something like
"primary.director.box.goes.here shared.resources.address.here http" would be
much clearer. Or even put them both down.
- going along with the 'HA' documentation, it would be nice to know that you
should delete alias IP addresses (eth0:0 for example) on both machines once
you're ready to go HA. I got a 'uh oh, shared mac address for VIP' error..
(in my case it was easy to diagnose since I'd already scoured the newsgroups
for 5 hours for other problems beforehand..)
</snip>
also you should VERY definitely expect to be using mon for your HA setup.
There's a bug (cleaned up in snapshot code?) in ha where if the node fails
over things work ok.. but if the original node pops back online it could
lead to a 'split brain' scenario where both nodes are active... not good!
naturally I've got the setup that could be bit by this bug :P ...hmm note
this only occurs if you are using 1 medium to talk back and forth between
the nodes... which will be the case if you don't use mon!
anyway I'm still in process of switching over to the 'mon' thing and I'd be
happy to share mon scripts & experiences with you. hmm especially since I'm
stuck on one of them with a 'syntax error'... :(
Peter
|
