[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:
>> Hi!
>>
>> I'm a newbie and have a problem with a simple virtual-IP config. I want
>> the virtual-IP to be either on server1 or server2, depending on which of
>> the server is having network connectivity (ping) to the outside. My
>> config is:
>>
>> node server1 \
>>         attributes standby="off"
>> node server2 \
>>         attributes standby="off"
>> primitive failover-ip ocf:heartbeat:IPaddr \
>>         params ip="11.222.32.161" \
>>         op monitor interval="3s"
>> primitive pingtest ocf:pacemaker:pingd \
>>         params host_list="11.222.53.113" multiplier="10" dampen="5s" \
>>         op monitor interval="10s"
>> clone clonePing pingtest
>> location aktiverLB failover-ip \
>>         rule $id="aktiverLB-rule" pingd: defined pingd
>> 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"
>>
>> I use Debian Lenny with packages from backports. My problems are:
>>
>> 1. After starting /etc/init.d/corosync it takes around 60s until
>> pacemaker comes online. Which timer is that? Is it safe to reduce it to
>> few seconds?
> 
> Why? That would be unsafe if your bootup times differed by, well, a few
> seconds. You'd boot up both nodes and boom, your cluster would form two
> separate single-node memberships.

This one I do not understand. E.g. consider 3 seconds timeout. Both
servers are down. Server is powered on. Pacemaker on server 1 waits 3
seconds for other cluster nodes, then it starts as single node.

Now server2 boots and pacemaker on server 2 will find the existing node
and join it.

Of course, for example if the network is down when both servers are
booting, then I will have 2 single-node cluster. But that would also
happen if the network gets partitioned later.

regards
Klaus