[Linux-HA] Constraints when a resource (group) goes down?

[Andrew Beekhof]

On 11/8/06, Hildebrand, Nils, 232 <Nils.Hildebrand at bamf.bund.de> wrote:
> Hi Andrew,
>
> > [...]
> > > Following dependencies should be met ALL TIMES:
> > >
> > > - Group "Redaktion" should be first,
> > > - Group "Master" should be next,
> > > - Group "Slave" should be last,
> > > - Group "Master" and "Slave" are never active on the same node.
> > >
> > > With the attached cib this works pretty well (on failover/failback
> > > situations).
> >
> > you might want to also assign priorities so they are always
> > processed in this order.
>
> >From the annotated DTD:
>
> "priority (integer, default=0):
>
> dictates the order in which resources will be processed. If there is an
> insufficient number of nodes to run all resources, the lower priority
> resources will be stopped to make sure the higher priority resources remain
> active. "
>
> How and why should priorities help here?
> As I said - this works pretty well as it is on failover-situations. What
> does not work is the behaviour on partial failures of resources or
> resource-groups.

as i said:

> > so they are always processed in this order.

>
> > > Next Test:
> > > When all (or certain) master resources fail the expected behaviour
> > > would have been that the slave-group gets stopped, master
> > restarts, slave starts.
> >
> > for this you need the restart_type resource attribute:
> >      http://www.linux-ha.org/v2/dtd1.0/annotated
>
> The behaviour does not change when I set restart_type to "restart". Neither
> as Group-Attribute, nor as Resource-Attribute for all or some resources.

you're replacing the existing restart_type="#default" entry in all the groups?

if this is what you've done, can you send me the result of
   cibadmin -Q
from your cluster when its running please?

with that i'll be able to simulate various failures.

>
> BTW: I am using 2.0.7-1
>
> If I fail the resource "RedWeb" it still gets just restarted.
>
> Correct would be:
> Stop VIP3,

no.  VIP3 comes _before_ RedWeb and therefore has no need to be restarted.

if you want it restarted (not sure why since its of type Dummy), then
put it after RedWeb in the group.

> Stop Slave, Stop Master,

actually these would happen first

> Start Redaktion, Start Master, Start
> Slave.

right

>
> VIP3 would be the "other end" of the ordering within the Resource-Group
> "Redaktion",
> Master would be the "other end" of the ordering constraint "order_redaktion"
> and
> Slave would then be the "other end" of the ordering constraint "order_live".
>
> > >
> > > I simulate the failure of resources by deleting the resource-files
> > > located at
> > >   /var/run/heartbeat/rsctmp
> > >
> > > But now the following happens:
> > > Whenever a resource from RG "Master" goes down, it gets restarted -
> > > even if ALL resources of the same resource group are down.
> > >
> > > To me this seems to be a bug, since the constraint "Slave after
> > > Master" is not being met.
>
> Is there any way around this problem?
>
>
> Kind regards, Nils
> _______________________________________________
> Linux-HA mailing list
> Linux-HA at lists.linux-ha.org
> http://lists.linux-ha.org/mailman/listinfo/linux-ha
> See also: http://linux-ha.org/ReportingProblems
>