Hi Dejan,
thanks for your answer.
I need the following -- is this possible with heartbeat ? how?
ON FIRST START
the httpd on all nodes MAY ONLY BE STARTED if
ALL MYSQL on ALL NODES are running
(I need this for testing purposes and initialization)
(how can be defined to start mysql on node1,
then mysql on node2 if node1-mysql is running,
then mysql on node3 if node2-mysql and node1-mysql are running).
ON CLUSTER RUN
when mysql on one node dies httpd on this (and only on this) node should
shutdown
ON CLUSTER RUN
when httpd on one node fails mysql should run further
NEVER should down
whole httpd or whole mysql when one node fails
How could this be achieved ?
Thanks again in advance.
Thomas.
Zitat von Dejan Muhamedagic <[EMAIL PROTECTED]>:
Hi,
On Sun, Feb 17, 2008 at 10:25:56PM +0100, www.tiri.li high
availability wrote:
Attached you find my config for a 4node cluster with
the resource for mysql and httpd for the nodes clu01..clu03, and
a configuration for "never run mysql+httpd on clu00".
This works.
But when I now (via hb_gui) stop the resource rip_httpd:1 then all of
real_mysql are restarted and real_httpd is stopped completely.
What should be changed to get
- if rip_httpd:0 is stopped, then only rip_mysql:0 should be stopped
- if rip_httpd:1 is stopped, then only rip_mysql:1 should be stopped
- if rip_httpd:2 is stopped, then only rip_mysql:2 should be stopped
Don't think that you could (or even should) reference the clone
instances in the configuration. Managing them could be OK if
really needed.
If I perform a "start default" on "real_httpd" then real_mysql is being
restarted and real_httpd is running after that fine.
If I perform a stop on rip_mysql:0 the real_httpd is being restarted
and running fine but real_mysql is being shutdown completely after this.
But this may not happen, because of the order first mysql then httpd,
so IMHO real_httpd should be dead as well. (but please only on the node
where rip_mysql fails).
Why don't you put the apache and mysql in a group? Then clone the
group. Could that help? If you describe what you want to achieve
...
Thanks,
Dejan
Thanks for your help in advance.
Thomas.
----------------------------------------------------------------
This message was sent using IMP, the Internet Messaging Program.
_______________________________________________
Linux-HA mailing list
Linux-HA@lists.linux-ha.org
http://lists.linux-ha.org/mailman/listinfo/linux-ha
See also: http://linux-ha.org/ReportingProblems
