[Linux-cluster] rhel6 node start causes power on of the other one

[Gianluca Cecchi]

On Fri, 25 Mar 2011 07:19:40 +0530, Rajagopal Swaminathan wrote:
> On which CPU  should service guard run on?

ServiceGuard is an HA solution with similar targets as RHCS or other
solutions, older or newer ones (aka Kimberlite, LifeKeeper, heartbeat
2.x, Pacemaker/Corosync, Pacemaker/Heartbeat and so on...)
I have not to run ServiceGuard at all at the moment.

I used it in the past on both HP-UX and Linux. What are services in
RHCS are called packages in ServiceGuard.
Last time I worked on it on Linux was about 6 years ago on Itanium 1
64bit platform for a 4 nodes configuration.
It was at that time that I noticed its behaviour (prompt at console)
when you boot a node and it doesn't "see" the other cluster nodes...
And so after my "clean_start" misunderstanding, I wanted to feed the
thread with some considerations about possible evolutions/approaches
in RHCS itself, no other intention...

Now ServiceGuard is supported only on HP-UX and HP suggests a path in
migrating existing implementations on Linux towards OS vendor cluster
suites (aka RHCS if you are using RHEL).

More information (note the domain still Compaq ;-):
http://www.compaq.com/solutions/enterprise/highavailability/linux/serviceguard/index.html
http://h18026.www1.hp.com/solutions/enterprise/highavailability/linux/serviceguard/discontinuance.html

and for commercial/technical notes about features on HP:
http://h71028.www7.hp.com/enterprise/w1/en/os/hpux11i-serviceguard-solutions-overview.html

I think in general is a good thing to see and analyze what other ones
are doing to solve your same problems...