[Linux-cluster] Duplicate Volumes, GFS fail over

isplist at logicore.net isplist at logicore.net
Fri Nov 3 15:46:12 UTC 2006


As you might be able to tell, I started doing this but it's leading to too 
many other problems. I think I'll have to learn dual path's later as it's just 
going to end up being another level of learning on top of everything else I've 
taken on. 

I appreciate the help everyone gave me on dual path's and have kept your 
replies to work with at a later date.

Mike



> #First setup multipath.conf

Done, leaving it as default.
 
> #If you are using qlogic qla2xxx series  hba's this is what i was told
> to put in modprobe.conf:
> alias scsi_hostadapter qla2xxx
> alias scsi_hostadapter1 mptbase
> alias scsi_hostadapter2 mptspi
> options qla2xxx qlport_down_retry=1 ql2xretrycount=5
> options scsi_mod max_luns=255

Done, changed file.
 
> mkinitrd -f /boot/initrd-`uname -r` `uname -r`
> init 6; exit
> #else just continue on
> modprobe dm-multipath
> modprobe dm-round-robin
> service multipathd start

Done.
 
> #Setup /etc/cluster/cluster.conf

Done.

> #Enable the nodes by running this on each box:
> ccs_tool addnodeids

This command won't run. Have to look into it.

> #at this point get on one of the boxes and do this:
> pvcreate /dev/mapper/mpath[01]
> vgcreate pri_volGrp0 /dev/mapper/mpath[01]
> lvcreate --size <JUST UNDER TOTAL SIZE>GB --name pri_lv0 pri_volGrp0
> mkfs.gfs2 -p lock_dlm -t <CLUSTERNAME>:data -j 4
> /dev/pri_volGrp0/pri_lv0
> 
> echo "/dev/pri_volGrp0/pri_lv0 <MOUNT POINT> gfs2 defaults 1 1" >>
> /etc/fstab
> mkdir <MOUNT POINT>
> 
> #Now one at a time, do each of these commands, one box then the other:
> service clvmd start
> service gfs2 start
> #if that all works enable it to run at boot:
> for x in multipathd cman clvmd gfs2; do chkconfig $x on; done







More information about the Linux-cluster mailing list