I'm having trouble wrapping my head around some aspects of the Sequoia
configuration and was hoping that someone here might be able to assist
in shedding some light on things.
First, my goal. I'm attempting to setup a fully redundant and easily
recoverable database cluster. Current implementation and testing is
based around MySQL.
I've successfully defined a single controller (eventually I will want to
increase this) with a virtual database consisting of two nodes. Updates
are propagated to both servers and queries work properly. This lead me
to test dropping a node, which appears to have worked well. Updates and
queries continued to work against the remaining nodes.
In investigating how to bring the failed node back into the cluster I
arrived at my conceptual problem. From what I can tell, in order to
ease the restoration of a failed node a RecoveryLog is needed or at
least highly recommended. However, this is itself stored in a database,
which brings us back to the initial problem with database redundancy.
The documentation indicates that the RecoveryLog database definition
could be pointed back to a Sequoia controller (even the self same
controller). Wouldn't this then lead me back to needing a RecoveryLog
for the new RecoveryLog virtual database? Or is there some
configuration that I'm not currently grasping?
--
Jamin W. Collins
_______________________________________________
Sequoia mailing list
[email protected]
https://forge.continuent.org/mailman/listinfo/sequoia
