Share volume group

[Gabriel_23]

I currently have a volume group that lives on an array. I have a 2-node cluster set up. I would like to share this volume group at the same time between the 2 nodes with one node having read-only permissions. Is this possible? I have attempted to use the
"-S" option with vgchange, but the one node is not able to see this volume group unless I failover the package from the other node. Hopefully, this makes sense. If there are any suggestions, I would highly appreciate it.
Thanks,
Gabe

[KapilRaj]

No no no


Kaps

[KapilRaj]

You may use NFS for this...

[RAC_1]

A volume group can be shared between hosts as follows.

1 NFS
2 MCSG.

In MCSG, the volume group will be included in a package. The package will have it's own ip. At any point of time, a package will be owned/running on one node. The aplications/programs will access it with package ip. In case that nodes fails, package will be switched over to another node. A volume group needs to configured with following options for putting it in MCSG.

vgchange -c y vgxx (cluster aware)
vgchange -s y vgxx (sharable)
vgchange -a e vgxx (exclusive activation)

Anil

[Sundar_7 · Accepted solution]

This is how you do it.

From node1

# vgexport -p -v -s -m vg.map /dev/vgname
# rcp vg.map node2:/tmp

on node2

# mkdir /dev/vgname
# mknod /dev/vgname/group c 64 0x020000

- match the minor number of the group file with that of node1.

# vgimport -v -s -m /tmp/vg.map /dev/vgname

now on node2

# vgchange -a r /dev/vgname.

Now vg is activated in node2 in read only mode.


-- Sundar

[A. Clay Stephenson]

Sharing a VOLUME GROUP read-only on one or both nodes or read-write on both nodes is fine as long as all the LVOL's are raw devices --- and the software has some mechanism to coordinate activity; HOWEVER (and this is a big however), under no circumstances can two nodes share a mounted filesystem except when BOTH nodes mount it readonly. The reason is that if nodeA mounts the filesystem read-write and nodeB mounts it readonly, nodeB won't have a clue about the contents of nodeA's buffer cache ---- absolute chaos will be the result. If you wish to share filesystems, mount it read-write on one node and NFS on the other. It really make no difference whether or not this is a MC/SG cluster, the answer is the same.

[Gabriel_23]

Thank you all for taking the time out to help me out with this...