Sametime clustering and the bad bad server put on the ’naughty mat’
Tags :Sametime
So to catch up on progress, the community
chat cluster works great behind the hardware based load balancer. When
one server is taken down (for testing), the Sametime client drops and moves
to a connecting status. Within a minute (it was at 30 seconds in
our testing) the client reconnects to the clustered chat server. Magic
at it's finest. So as long as the load balancer is working, people
can chat to their heart's content.
Now instant meetings are started on the
home server (or the one connected to in a cluster) so if that server dies,
then you lose that meeting when you fail over. Now this is where
one server became a very very bad server. One of them decided that
it would not start a meeting no matter how hard we begged. So a quick
rebuild tomorrow and we will test that last piece. I have one remaining
question. If I am on the server that stays up as the owner of the instant
meeting and the other participant was connected to the server that dies,
will they stay in the instant meeting and reconnect for chat? Oh
those begging questions to be answered.
Anyone want me to run through the steps
of how to cluster two Sametime chat servers?
blog comments powered by Disqus
On Tuesday, May 10th, 2005 by Chris Miller
