[Linux-cluster] Need a lot of help

[Bowie Bailey]

From: Marius Stoica [mailto:sly_extrem at yahoo.com]
> Patrick Caulfield <pcaulfie at redhat.com> wrote:
> > Marius Stoica wrote:
> > > Based on the documentation provided in usage.txt...nothing can be
done.
> > > Does anyone know how can I make a shared disk cluster on 2 nodes ?
> > 
> > Did you actually read it? There is a section on 2 node clusters.
> 
> My question is how can I provide redundancy (mirroring) and 
> accees the shared resource on both nodes ?...
> 
> The cman options are implemented allready.

Yea, I had that problem too.  The docs are heavy on basic setup and a
bit sparse on how to actually do anything useful.

This may help.  This is the cluster.conf file from a test cluster I'm
working on.  Note the "resources" section at the bottom.  Based on
your question, I'm not sure if you are doing GFS or not, but maybe
this will point you in the right direction.

Use lvm2 to create the logical volumes on one box and then do a vgscan
on the other box to read the volume group information.

Bowie

-----------------------------------------------

<?xml version="1.0"?>
<cluster config_version="7" name="storagetest">
        <cman expected_votes="1" two_node="1"/>
        <fence_daemon clean_start="0" post_fail_delay="0"
post_join_delay="20"/>
        <clusternodes>
                <clusternode name="test1">
                        <fence>
                                <method name="human">
                                        <device name="human"
nodename="test1"/>
                                </method>
                        </fence>
                </clusternode>
                <clusternode name="test2">
                        <fence>
                                <method name="human">
                                        <device name="human"
nodename="test2"/>
                                </method>
                        </fence>
                </clusternode>
        </clusternodes>
        <fencedevices>
                <fencedevice agent="fence_manual" name="human"/>
        </fencedevices>
        <rm>
                <failoverdomains/>
                <resources>
                        <clusterfs device="/dev/vg00/lvol01" fstype="gfs"
mountpoint="/raid" name="store1" options=""/>
                </resources>
        </rm>
</cluster>