[Pacemaker] Problem with ClusterIP

[Anne Nicolas]

Le 26/11/2014 13:43, Michael Schwartzkopff a écrit :
> Am Mittwoch, 26. November 2014, 13:22:53 schrieben Sie:
>> Le 26/11/2014 13:07, Michael Schwartzkopff a écrit :
>>> Am Mittwoch, 26. November 2014, 12:54:20 schrieb Anne Nicolas:
>>>> Le 26/11/2014 12:23, Michael Schwartzkopff a écrit :
>>>>> Am Mittwoch, 26. November 2014, 12:01:36 schrieb Anne Nicolas:
>>>>>> Hi !
>>>>>>
>>>>>> I've been using clusterip for a while now without any problem in
>>>>>> Active/Passive clusters (2 nodes).
>>>>>
>>>>> Could you please explain, how could you use the ClusterIP in an
>>>>> active/passive cluster? ClusterIP ist for the use in an active/active
>>>>> cluster. See man iptables and look for the CLUSTERIP target.
>>>>>
>>>>>
>>>>> Please explain more detailed.
>>>>> What is your config?
>>>>> What do you expect the cluster to do?
>>>>> What really happens?
>>>>> Where is  the problem?
>>>>
>>>> Maybe my explanation was not that clear.
>>>
>>> Yes.
>>>
>>>> Here is my configuration
>>>>
>>>> crm configuration show
>>>>
>>>> node $id="17435146" pogcupsvr
>>>> node $id="34212362" pogcupsvr2
>>>> primitive apache ocf:heartbeat:apache \
>>>>
>>>>            params configfile="/etc/httpd/conf/httpd.conf" \
>>>>            op start interval="0" timeout="40s" \
>>>>            op stop interval="0" timeout="60s"
>>>>
>>>> primitive clusterip ocf:heartbeat:IPaddr2 \
>>>>
>>>>            params ip="172.16.16.11" cidr_netmask="24" nic="eth0" \
>>>>            meta target-role="Started"
>>>>
>>>> ...
>>>> property $id="cib-bootstrap-options" \
>>>>
>>>> dc-version="1.1.7-2.mga1-ee0730e13d124c3d58f00016c3376a1de5323cff" \
>>>>
>>>>            cluster-infrastructure="corosync" \
>>>>            stonith-enabled="false" \
>>>>            no-quorum-policy="ignore"
>>>>
>>>> rsc_defaults $id="rsc-options" \
>>>>
>>>>            resource-stickiness="100"
>>>>
>>>> So I've started primary node (pogcupsvr). Configuration was checked and
>>>> ok. Then started the second node (pogcupsvr2). This time all the
>>>> configuration looked ok, no error but when I checked the network
>>>> configuration, eth0 was up on both nodes with same IP address of course,
>>>> instead of having it up only on primary node.
>>>
>>> If that is your config than the start of the IP address on BOTH nodes is
>>> really bad. This should not happen and is definitely an error.
>>>
>>> BUT: I doubt that this is you complete config, because this would not work
>>> anyway. The cluster would start the IP address on one node and the
>>> Webserver in the other node.
>>>
>>> Please paste the complete config. Then the community would be able to
>>> help.
>>
>> Here is the complete configuration:
>>
>> node $id="17435146" pogcupsvr
>> node $id="34212362" pogcupsvr2
>> primitive apache ocf:heartbeat:apache \
>>           params configfile="/etc/httpd/conf/httpd.conf" \
>>           op start interval="0" timeout="40s" \
>>           op stop interval="0" timeout="60s"
>> primitive clusterip ocf:heartbeat:IPaddr2 \
>>           params ip="172.16.16.11" cidr_netmask="24" nic="eth0" \
>>           meta target-role="Started"
>> primitive drbdserv ocf:linbit:drbd \
>>           params drbd_resource="server" \
>>           op monitor interval="60s"
>> primitive fsserv ocf:heartbeat:Filesystem \
>>           params device="/dev/drbd/by-res/server" directory="/clusterfs"
>> fstype="ext4"
>> primitive libvirt-guests lsb:libvirt-guests
>> primitive libvirtd lsb:libvirtd
>> primitive mysql ocf:heartbeat:mysql \
>>           params binary="/usr/bin/mysqld_safe" config="/etc/my.cnf"
>> datadir="/clusterfs/mysql" \
>>           op start interval="0" timeout="40s" \
>>           op stop interval="0" timeout="60s" \
>>           meta target-role="Started"
>> primitive named lsb:named
>> primitive samba lsb:smb
>> group services fsserv clusterip libvirtd samba apache mysql
>> ms drbdservClone drbdserv \
>>           meta master-max="1" master-node-max="1" clone-max="2"
>> clone-node-max="1" notify="true"
>> colocation fs_on_drbd inf: fsserv drbdservClone:Master
>> order fsserv-after-drbdserv inf: drbdservClone:promote fsserv:start
>> property $id="cib-bootstrap-options" \
>>
>> dc-version="1.1.7-2.mga1-ee0730e13d124c3d58f00016c3376a1de5323cff" \
>>           cluster-infrastructure="corosync" \
>>           stonith-enabled="false" \
>>           no-quorum-policy="ignore"
>> rsc_defaults $id="rsc-options" \
>>           resource-stickiness="100"
>
> OK. Config seems to be ok. But I would make the constraints to work on the
> group, not on the fsserver. But since the fsserver is the first resource in the
> group, everything should be ok.
>
> Now:
>> This time all the
>> configuration looked ok, no error but when I checked the network
>> configuration, eth0 was up on both nodes with same IP address of course,
>> instead of having it up only on primary node.
>
> Please could you paste the output of the command "ip addr list dev eth0" on
> both nodes?

In fact I read your message and it just turned on the light in some part 
of my brain... I checked the interface coinfiguration and discovered an 
IP address on both side...
And for sure that could not work. Removing this just did the trick and 
everything is ok now.

Sorry for the noise but thanks for helping in finding this stupid mistake :)

>
> Mit freundlichen Grüßen,
>
> Michael Schwartzkopff
>


-- 
Anne
http://mageia.org