[Pacemaker] Several problems with simple virtual-IP config

[Klaus Darilion]

Am 08.02.2011 18:20, schrieb Florian Haas:
> On 02/08/2011 06:03 PM, Klaus Darilion wrote:

>> Now I put server2 online again: # crm node online server2.
>> That means, server2 is online and has ping connectivity, server1 is
>> online and doesn't have ping connectivity. But the virtual-IP stayed
>> with server1:
>>
>>   Online: [ server1 server2 ]
>>
>>   failover-ip     (ocf::heartbeat:IPaddr):        Started server1
>>    Clone Set: clonePing
>>        Started: [ server2 server1 ]
>>
>> What do I have to change in the config to have here a failover to server2?
> 
> IIUC your pingd attribute is still defined, albeit with a value of 0.
> Thus the location constraint is still fulfilled and the pingd score is 0
> on both nodes. Equal score means Pacemaker gets to decide where the
> resource runs, and currently there is no reason to migrate.
> 
> Try this:
> 
> location aktiverLB failover-ip \
>         rule $id="aktiverLB-rule" -inf: pingd lte 0
> 
> or:
> 
> location aktiverLB failover-ip \
>         rule $id="aktiverLB-rule" inf: pingd gt 0
> 

Hi Florian!

I changed the config now as suggested by
http://www.clusterlabs.org/wiki/Example_configurations#pingd_location_constraint:


node server1 \
        attributes standby="off"
node server2 \
        attributes standby="off"
primitive failover-ip ocf:heartbeat:IPaddr \
        params ip="83.136.32.161" \
        op monitor interval="3s"
primitive pingtest ocf:pacemaker:pingd \
        params host_list="88.198.53.113" multiplier="10" dampen="5s" \
        op monitor interval="10s"
clone clonePing pingtest
location aktiverLB failover-ip \
        rule $id="aktiverLB-rule" -inf: not_defined pingd or pingd lte 0
property $id="cib-bootstrap-options" \
        dc-version="1.0.9-74392a28b7f31d7ddc86689598bd23114f58978b" \
        cluster-infrastructure="openais" \
        expected-quorum-votes="2" \
        stonith-enabled="false" \
        no-quorum-policy="ignore"

But the behavior is IMO still confusing:

server1 is started:

	Online: [ server1 ]
	OFFLINE: [ server2 ]
	
	 Clone Set: clonePing
	     Started: [ server1 ]
	     Stopped: [ pingtest:1 ]
	
	Migration summary:
	* Node server1:


Why are there no scores? I guess this prevents the virtualIP resource to
be started on server1.


Now I block PING. After 5 seconds the score is 10 (although ping is
blocked) and the resource gets started:

	Online: [ server1 ]
	OFFLINE: [ server2 ]
	
	failover-ip     (ocf::heartbeat:IPaddr):        Started server1
	 Clone Set: clonePing
	     Started: [ server1 ]
	     Stopped: [ pingtest:1 ]
	
	Migration summary:
	* Node server1:  pingd=10


After 5 more seconds the score becomes 0 and the resource will be removed:

	Online: [ server1 ]
	OFFLINE: [ server2 ]
	
	 Clone Set: clonePing
	     Started: [ server1 ]
	     Stopped: [ pingtest:1 ]
	
	Migration summary:
	* Node server1:  pingd=0


Now I allow PING. Even after 10 seconds the score is still 0 although
PINGs are working fine again:

	Online: [ server1 ]
	OFFLINE: [ server2 ]
	
	 Clone Set: clonePing
	     Started: [ server1 ]
	     Stopped: [ pingtest:1 ]
	
	Migration summary:
	* Node server1:  pingd=0


I think the problem is, that the pingd score is not calculated on
startup and is not updated when the connectivity problem recovers.

Any ideas what could be wrong?

Thanks
Klaus