On 05/24/2013 01:30 PM, Lars Marowsky-Bree wrote:
On 2013-05-24T12:15:04, Digimer <li...@alteeve.ca> wrote:
Nope. Until RHEL 5, openais was the communication layer of the cluster.
However, the full AIS API was deemed "overkill" for what HA clustering
needed, so corosync was created for RHEL 6 as a stripped-down, HA focused
version of openais. In turn, openais was made a corosync plugin for those
who actually want the full AIS API.
You know, there *is* a software community out there that doesn't use
RHEL. Really. ;-) At the time, this was actually a community decision
and not RHEL-driven. (Though of course the needs of RHEL also informed
it.)
You are absolutely correct. However, I can only speak to what I know.
/me puts on her crimson fedora :)
The AIS APIs were mostly just not deemed worth the trouble. Outside a
very narrow field (Carrier Grade), very few people used them, and those
that did preferred the opensaf implementation. Hence, openais turned out
to not be worth the effort of long-term support for a rather complex set
of services and a *huge* specification.
OCFS2 and cLVM2 exist in an incarnation that uses the AIS CKPT
(checkpoint) service for synchronization, but that has been/will be
rewritten too.
The concepts of "Active/Active", "Active/Passive", etc. is not a membership
or cluster communication layer question. Those concepts come from the
resource manager (rgmanager or pacemaker).
Yes - and no, because it neglects the detail that the AIS stuff does
include the notion of a limited cluster manager too.
Regards,
Lars
Yes, again right, but it seemed far enough out of scope to not
mention/confuse the topic. it's not practical in the open source HA
space, so far as I have ever seen.
--
Digimer
Papers and Projects: https://alteeve.ca/w/
What if the cure for cancer is trapped in the mind of a person without
access to education?
_______________________________________________
Pacemaker mailing list: Pacemaker@oss.clusterlabs.org
http://oss.clusterlabs.org/mailman/listinfo/pacemaker
Project Home: http://www.clusterlabs.org
Getting started: http://www.clusterlabs.org/doc/Cluster_from_Scratch.pdf
Bugs: http://bugs.clusterlabs.org
