Hi James,
I really don't think that's possible, but I do think you are over
thinking the situation.
The best thing you can do for a critical deployment is to configure the
primary not to accept any connections in conjunction with not doing
things that excessively query authd (See the "Turbo CAM" post).
If you really want HA, I'd deploy the primary on a VM with something
like VMotion to move the VM between physical nodes.
James Tan wrote:
Hi,
anyone implemented a DS HA before?
E.g. a couple of servers in the FOG, should the primary goes down, there
is another to assume the task of replicating changed information to the
other slaves? So that the other slaves can still be in sync with each
other until PRI comes back up.
knowing utreplica uses hostname, can this be done using some sort of
IP-DNS round robin and have e.g. 2 PRI but using the same hostname so
all the other slaves will try at random any PRI for updates? Or can this
be done using IP bonding (Sun cluster interop with SRSS?)
My concern is even if there's clustering/round-robin dns implementation,
what happen if in a 2 PRI setup, when the failed one (with stale
information) comes back up, how will it sync with the other PRI or will
it affect the still running fine slaves (having newer data with the
still running PRI).
thanks,
James
_______________________________________________
SunRay-Users mailing list
[email protected]
http://www.filibeto.org/mailman/listinfo/sunray-users
_______________________________________________
SunRay-Users mailing list
[email protected]
http://www.filibeto.org/mailman/listinfo/sunray-users
