[Linux-HA] A clone starts again

Andrew Beekhof beekhof at gmail.com
Fri Sep 21 03:02:25 MDT 2007


On 9/21/07, Andrew Beekhof <beekhof at gmail.com> wrote:
> On 9/21/07, HIDEO YAMAUCHI <renayama19661014 at ybb.ne.jp> wrote:
> > Hi,
> >
> > I constituted a cluster in two nodes.
> > The on_fail attribute of IPaddr is fence.

you also neglected to mention that you're running a custom version of heartbeat
what patches have you applied and are you planning on sharing them?

>
> which is going to have absolutely no effect since you've set
>
>            <nvpair name="stonith-enabled" value="false"/>

actually what i should have said is that it makes no sense as it will
have two effects,
1) it will force the resource to move to another node
2) it will mark the node as unclean (ready to be fenced but wont
actually fence it) which will prevent resources from running there.
which is why the clone stops on that node.

where it gets interesting is when you repeated the process on the second node.
while 1) occurs for every node the ipaddr failed on, 2) only takes
effect for active resources (ie. the most recent failure).

this could possibly be considered a bug but since the whole idea of
have stonith-enabled=false and on_fail=fence makes no sense there is
little point in making it work



More information about the Linux-HA mailing list