thanks! I will try the "on fail ignore". The other issue was not really an issue because it finally worked fine when I decided to completely erase the CIB and reconfigure constraints and resources without adding the mail resource. Maybe it was a problem due to the fact that I used "crm_resource" to remove this one resource specifically, I don't know...
I had "solved" the mail program by changing the MailTo RA, launching the mail in another process to not have to wait for the timeout to arrive, and masking the error. The advantage is that I don't have to wait the timeout - which was quite long if I remember well - to continue leaving up resources, allowing failover to execute faster. The inconvenient is that I have to change the MailTo RA... Is there any way to emulate this behaviour by setting fail_ignore? On Mon, Aug 25, 2008 at 12:21 PM, Andrew Beekhof <[EMAIL PROTECTED]> wrote: > On Tue, Aug 12, 2008 at 12:30, Laurent Yin <[EMAIL PROTECTED]> > wrote: > > Hello, > > > > I set up a DRBD-Mysql cluster with a master slave set DRBD and a mysql > > resource group containing : > > -a Filesystem > > -a mysql (5.1) > > -a virtual IP Address (IPAddr2) > > -a MailTo RA > > > > I have two constraints : > > - one colocational constraint which tells that you have to have DRBD > master > > on the machine running mysqlgroup > > - one ordering constraint which tells you have to launch mysqlgroup after > > DRBD > > > > It works fine and it does failover smoothly on machine poweroff and > stuffs. > > > > Now I would've liked it to be network-loss tolerant, eg if I unplug the > > network cable between the master node and the router, I want it to detect > > that connectivity is lost. > > For that purpose, I added two ping nodes to my ha.cf and a respawn with > > pingd. > > > > ## in HA.CF > > ping www.google.com > > ping www.yahoo.com > > > > respawn root /usr/lib/heartbeat/pingd -m 100 -d 5s -a pingd > > ## END OF in HA.CF > > > > I also added a constraint as done on the site > > http://www.linux-ha.org/pingdin the section "Only Run my_resource on > > Nodes With Access to at Least One > > Ping Node". > > > > ## CONSTRAINT ## > > <rsc_location id="my_resource:connected" rsc="mysqlgroup"> > > <rule id="my_resource:connected:rule" score="-INFINITY" boolean_op="or"> > > <expression id="my_resource:connected:expr:undefined" > > attribute="pingd" operation="not_defined"/> > > <expression id="my_resource:connected:expr:zero" > > attribute="pingd" operation="lte" value="0"/> > > </rule> > > </rsc_location> > > ## END OF CONSTRAINT ## > > > > > > I have two problems with this configuration. > > 1 ) When I unplug the network cable of the machine running mysql, after > > detecting that there is no connectivity, it tries to stop the group, > > beginning with my last resource which is MailTo. But, as there is no > > connectivity, it fails to stop, and therefore the whole group remains > > unstopped. What can I do against this? > > fix the RA or set on_fail=ignore for the resource's stop action > > > > > 2 ) When I remove the MailTo RA (just for testing purpose, to see what > > happens, but this is not an acceptable solution), it manages to stop the > > mysqlgroup, but it doesn't get started on the other node. I assume that > it > > is because DRBD is still master on this node. How can I tell Heartbeat to > > switch master/slave in DRBD when connectivity is lost? > > Or is there another solution with constraints maybe? > > create a similar pingd constraint for drbd as you used for the group > _______________________________________________ > Linux-HA mailing list > [email protected] > http://lists.linux-ha.org/mailman/listinfo/linux-ha > See also: http://linux-ha.org/ReportingProblems > -- This is the end ... beautiful friend ... This is the end .... my only friend, the end ... _______________________________________________ Linux-HA mailing list [email protected] http://lists.linux-ha.org/mailman/listinfo/linux-ha See also: http://linux-ha.org/ReportingProblems
