Hi,

I have the following (simple) configuration:

primitive iscsi0 ocf:heartbeat:iscsi \
params portal=*"172.20.250.5" *target="iqn.2010-10.nl.nas:nas.storage0"
primitive iscsi1 ocf:heartbeat:iscsi \
params portal=*"172.20.250.21" *target="iqn.2010-10.nl.nas:nas.storage1"

primitive failover-ip0 ocf:heartbeat:IPaddr2 \
params ip="172.20.60.13" iflabel="0"

primitive lvm0 ocf:heartbeat:LVM \
params volgrpname="vg0" exclusive="yes"

primitive filesystem0 ocf:heartbeat:Filesystem \
params device="/dev/vg0/lv0" directory=*"/mnt/storage"* fstype="xfs"

primitive filesystem1 ocf:heartbeat:Filesystem \
params device="/dev/vg0/lv1" directory=*"/mnt/storage2"* fstype="xfs"

primitive nfs-server lsb:nfs-kernel-server

primitive samba-server lsb:samba

group nfs-and-samba-group iscsi0 iscsi1 failover-ip0 lvm0 filesystem0
filesystem1 nfs-server samba-server

location nfs-and-samba-group-prefer-node01 nfs-and-samba-group 100: node01


So two iscsi initiators, then LVM on top of those, two filesystems (one for
nfs exports and one for a samba share).

What I noticed is that when I want to only stop the nfs-server (for doing
some maintenance for instance) the samba-server is stopped also (because it
is in a group and the order in the group seems like every primitive is
required for the next primitive) (pacemaker reads the group from left to
right)

How would I be able to stop nfs-server and/or samba-server without
interupting anything else in the group?

Should I split those two from the group? But then I would need more
constraints telling that the nfs-server and samba server can only start at
the node were the iscsi initiator/LVM is up.
And when I want to migrate the nfs-server to the other node, the
samba-server and iscsi/LVM need to migrate also because of the large vg0?

Can anyone tell me how to accomplish this?

Thanks you very much in advance,

Caspar Smit
_______________________________________________
Linux-HA mailing list
[email protected]
http://lists.linux-ha.org/mailman/listinfo/linux-ha
See also: http://linux-ha.org/ReportingProblems

Reply via email to