[Pacemaker] colocation issues (isnt it always)

[Patrick H.]

So colocation is biting me in the ass again and I cant figure this one out.
I have a group of iSCSI devices that then go into an md raid device that 
then goes into an lvm device which then gets mounted and then exported 
by nfs. Thoughout this whole project I've had the resources trying to 
start on nodes where the resource it depends on wasnt even running. But 
those were usually when I failed something and they tried to migrate. 
Right now I'm having problems simply starting the final resource, the 
nfs export

All the resource are started on the same node except for the 
ocf:hearbeat:nfsserver resource. For some reason it refuses to start on 
the same node, even though I have a colocation rule.
All the other resources are on node 'nas01', but every time I start up 
the nfsserver resource it starts on 'nas02'. I've stopped it, cleaned 
it, started it back up. All have no effect. Ive even tried reversing the 
order of the resources in the colocation rule as last time I had this 
problem it was cause they were backwards.

Below is info from the crm util, and attached is the output of `cibadmin -Q`
Using the pacemaker 1.1.1-1.fc13 package on rhel6

crm(live)# configure show
node nas01
node nas02
node nas03
primitive filesystem_sdb1 ocf:etc:fs \
        params uuid="2c8b70de-09a7-4cf1-864d-346c0602d6e1" 
mount_path="/mnt/iscsi-sdb1" \
        meta target-role="Started"
primitive iSCSI_nas01_sdb1 ocf:heartbeat:iscsi \
        params portal="165.212.101.241" 
target="iqn.165.212.101.240:nas01.sdb1" \
        op monitor interval="30s" \
        meta target-role="Started"
primitive iSCSI_nas02_sdb1 ocf:heartbeat:iscsi \
        params portal="165.212.101.242" 
target="iqn.165.212.101.240:nas02.sdb1" \
        op monitor interval="30s" \
        meta target-role="Started"
primitive iSCSI_nas03_sdb1 ocf:heartbeat:iscsi \
        params portal="165.212.101.243" 
target="iqn.165.212.101.240:nas03.sdb1" \
        op monitor interval="30s" \
        meta target-role="Started"
primitive lvm-iSCSI_sdb1 ocf:etc:lvm \
        params vg_name="vg-md-iscsi-sdb1" lv_name="lv-vg-md-iscsi-sdb1" \
        meta target-role="Started"
primitive md_iSCSI_sdb1 ocf:etc:md \
        params name="nas01:iscsi-sdb1" 
uuid="abddf4ea:06524e68:8e2b8f4e:1b24c56d" \
        meta target-role="Started" \
        op monitor interval="5s"
primitive nfs_sdb1 ocf:heartbeat:nfsserver \
        params nfs_init_script="/etc/init.d/nfs" 
nfs_shared_infodir="/mnt/iscsi-sdb1/nfs" nfs_notify_cmd="/bin/true" 
nfs_ip="*" \
        meta target-role="Started"
group gr-iSCSI_sdb1 iSCSI_nas01_sdb1 iSCSI_nas02_sdb1 iSCSI_nas03_sdb1
*colocation co-sdb1 inf: nfs_sdb1 filesystem_sdb1 lvm-iSCSI_sdb1 
md_iSCSI_sdb1 gr-iSCSI_sdb1*
order or-sdb1 inf: gr-iSCSI_sdb1 md_iSCSI_sdb1 lvm-iSCSI_sdb1 
filesystem_sdb1 nfs_sdb1
property $id="cib-bootstrap-options" \
        dc-version="1.1.1-972b9a5f68606f632893fceed658efa085062f55" \
        cluster-infrastructure="openais" \
        expected-quorum-votes="3" \
        stonith-enabled="false" \
        no-quorum-policy="ignore" \
        default-resource-stickiness="INFINITY" \
        last-lrm-refresh="1292215929"



crm(live)# status
============
Last updated: Mon Dec 13 05:04:52 2010
Stack: openais
Current DC: nas03 - partition with quorum
Version: 1.1.1-972b9a5f68606f632893fceed658efa085062f55
3 Nodes configured, 3 expected votes
5 Resources configured.
============

Online: [ nas02 nas01 nas03 ]

 Resource Group: gr-iSCSI_sdb1
     iSCSI_nas01_sdb1   (ocf::heartbeat:iscsi): Started nas01
     iSCSI_nas02_sdb1   (ocf::heartbeat:iscsi): Started nas01
     iSCSI_nas03_sdb1   (ocf::heartbeat:iscsi): Started nas01
 lvm-iSCSI_sdb1 (ocf::etc:lvm): Started nas01
 filesystem_sdb1        (ocf::etc:fs):  Started nas01
 md_iSCSI_sdb1  (ocf::etc:md):  Started nas01
* nfs_sdb1       (ocf::heartbeat:nfsserver):     Started nas02*




-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.clusterlabs.org/pipermail/pacemaker/attachments/20101212/97c958a1/attachment-0001.html>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: cibadmin.xml
Type: text/xml
Size: 27216 bytes
Desc: not available
URL: <https://lists.clusterlabs.org/pipermail/pacemaker/attachments/20101212/97c958a1/attachment-0003.xml>