Hey Dejan,
awesome solution :) Thx a lot for the tip :)
Cya, i'll share the config for everyone ....
node $id="73288296-569b-41f5-8b14-cf2f0fac8cf7" db01
node $id="a276707f-f3fd-4726-a7c1-8147c03a6336" db02
primitive drbd0 ocf:linbit:drbd \
params drbd_resource="r0"
primitive drbd1 ocf:linbit:drbd \
params drbd_resource="r1"
primitive fs0 ocf:heartbeat:Filesystem \
params device="/dev/drbd0" directory="/data/database" fstype="ext4"
primitive fs1 ocf:heartbeat:Filesystem \
params device="/dev/drbd1" directory="/data/documents"
fstype="ext4" options="acl"
primitive ip ocf:heartbeat:IPaddr2 \
params ip="10.80.80.73" nic="eth0"
primitive samba ocf:heartbeat:anything \
params binfile="/usr/sbin/smbd"
group cluster ip samba
ms ms-drbd0 drbd0 \
meta master-max="1" master-node-max="1" clone-max="2"
clone-node-max="1" notify="true" target-role="Master"
ms ms-drbd1 drbd1 \
meta master-max="1" master-node-max="1" clone-max="2"
clone-node-max="1" notify="true"
location ms-drbd0-master-on-db01 ms-drbd0 \
rule $id="ms-drbd0-master-on-db01-rule" $role="master" 100:
#uname eq db01
location ms-drbd1-master-on-db01 ms-drbd1 \
rule $id="ms-drbd1-master-on-db01-rule" $role="master" 100:
#uname eq db01
colocation cluster-with-fs inf: cluster ( fs0 fs1 )
colocation fs0-with-drbd0master inf: fs0 ms-drbd0:Master
colocation fs1-with-drbd1master inf: fs1 ms-drbd1:Master
order drbd0-before-fs0 inf: ms-drbd0:promote fs0:start
order drbd1-before-fs1 inf: ms-drbd1:promote fs1:start
order fs-before-cluster inf: ( fs0 fs1 ) cluster
order ms-drbd0-before-cluster inf: ms-drbd0:promote cluster:start
property $id="cib-bootstrap-options" \
dc-version="1.0.8-042548a451fce8400660f6031f4da6f0223dd5dd" \
cluster-infrastructure="Heartbeat" \
stonith-enabled="false" \
no-quorum-policy="ignore" \
last-lrm-refresh="1272369431"
On 27.04.2010 12:18, Dejan Muhamedagic wrote:
> Hi,
>
> On Tue, Apr 27, 2010 at 10:51:29AM +0200, Frank Lazzarini wrote:
>
>> Hey everyone,
>>
>> I've got a technical question. In short is it possible to colocate a
>> group to two master slave resources. I know it's an awkward question.
>> But in my case I have two drbd resource to two different masterslave
>> resources (because I use two different hard drives for the two drbd
>> resources), so what I want is that both filesystems, the virtual ip and
>> the samba resource are always started and only started when both master
>> slave resources are master on the same node. (weird right)
>>
>> Maybe someone has an idea to solves this otherwise, and I am maybe not
>> thinking right. Here's my crm for the moment (would be nice to have
>> color coding in the mails somehow :) )
>>
>>
>> node $id="73288296-569b-41f5-8b14-cf2f0fac8cf7" db01
>> node $id="a276707f-f3fd-4726-a7c1-8147c03a6336" db02
>> primitive drbd0 ocf:linbit:drbd \
>> params drbd_resource="r0"
>> primitive drbd1 ocf:linbit:drbd \
>> params drbd_resource="r1"
>> primitive fs0 ocf:heartbeat:Filesystem \
>> params device="/dev/drbd0" directory="/data/database" fstype="ext4"
>> primitive fs1 ocf:heartbeat:Filesystem \
>> params device="/dev/drbd1" directory="/data/documents"
>> fstype="ext4" options="acl"
>> primitive ip ocf:heartbeat:IPaddr2 \
>> params ip="10.80.80.73" nic="eth0"
>> primitive samba ocf:heartbeat:anything \
>> params binfile="/usr/sbin/smbd"
>> group cluster ip fs0 fs1 samba
>> ms ms-drbd0 drbd0 \
>> meta master-max="1" master-node-max="1" clone-max="2"
>> clone-node-max="1" notify="true" target-role="Master"
>> ms ms-drbd1 drbd1 \
>> meta master-max="1" master-node-max="1" clone-max="2"
>> clone-node-max="1" notify="true"
>> location ms-drbd0-master-on-db01 ms-drbd0 \
>> rule $id="ms-drbd0-master-on-db01-rule" $role="master" 100:
>> #uname eq db01
>> location ms-drbd1-master-on-db01 ms-drbd1 \
>> rule $id="ms-drbd1-master-on-db01-rule" $role="master" 100:
>> #uname eq db01
>> colocation cluster-with-drbdmaster inf: cluster ms-drbd0:Master
>> order ms-drbd0-before-cluster inf: ms-drbd0:promote cluster:start
>>
> This is what I would do:
>
> # group svc samba ip?
> group svc ip samba
> colocation fs0-with-drbd0 inf: fs0 ms-drbd0:Master
> order drbd0-before-fs0 inf: ms-drbd0:promote fs0:start
> colocation fs1-with-drbd1 inf: fs1 ms-drbd1:Master
> order drbd1-before-fs1 inf: ms-drbd1:promote fs1:start
> # these two are resource sets
> # (perhaps reference svc instead of samba?)
> order fs-before-smb inf: ( fs0 fs1 ) samba
> colocation smb-with-fs inf: samba ( fs0 fs1 )
>
> This configuration allows the two filesystems to start/stop
> independently (and in parallel). However, one issue is that
> stopping one of them would stop the service too (samba). Don't
> know how to keep samba running with only one filesystem
> available. Test the configuration thoroughly!
>
> Thanks,
>
> Dejan
>
>
>> property $id="cib-bootstrap-options" \
>> dc-version="1.0.8-042548a451fce8400660f6031f4da6f0223dd5dd" \
>> cluster-infrastructure="Heartbeat" \
>> stonith-enabled="false" \
>> no-quorum-policy="ignore" \
>> last-lrm-refresh="1272034422"
>>
>>
>>
>>
>> here is crm_mon (perhaps it helps to get the idea without reading the
>> whole crm configure thingie)
>>
>>
>> Online: [ db02 db01 ]
>>
>> Master/Slave Set: ms-drbd0
>> Masters: [ db01 ]
>> Slaves: [ db02 ]
>> Master/Slave Set: ms-drbd1
>> Masters: [ db01 ]
>> Slaves: [ db02 ]
>> Resource Group: cluster
>> ip (ocf::heartbeat:IPaddr2): Started db01
>> fs0 (ocf::heartbeat:Filesystem): Started db01
>> fs1 (ocf::heartbeat:Filesystem): Started db01
>> samba (ocf::heartbeat:anything): Started db01
>>
>>
>>
>> Cheers everyone
>>
>>
>>
>> _______________________________________________
>> Linux-HA mailing list
>> [email protected]
>> http://lists.linux-ha.org/mailman/listinfo/linux-ha
>> See also: http://linux-ha.org/ReportingProblems
>>
> _______________________________________________
> Linux-HA mailing list
> [email protected]
> http://lists.linux-ha.org/mailman/listinfo/linux-ha
> See also: http://linux-ha.org/ReportingProblems
>
_______________________________________________
Linux-HA mailing list
[email protected]
http://lists.linux-ha.org/mailman/listinfo/linux-ha
See also: http://linux-ha.org/ReportingProblems
