Linux and DevOps with Vijay

            I come to this solution due to one of my requirement in which i wanted to setup shared volume for my two node KVM virtual server node. I wanted to make a shared volume storage pool on both KVM node so that i can migrate VM between nodes.            Normally this design is for ovirt HA cluster. My setup was not intended to work as HA cluster but a manual cluster to move VM from one host node to another.            This setup is also usefull in case you need to setup any multimaster solution in which you dont want to use NFS and need data redundancy. So here is the setup - I have two nodes VM1  and VM2 each with a partition  /dev/vdb1 of 5GB size I created one mount point for both VM /gb1 and mounted the partition on this mount point # mkfs.xfs /dev/vdb1 # mount /dev/vdb1 /gb1 Above practice needs to be done on both VM node. Next is i will install centos gluster release on both the nodes # yum install -y centos-release-gluster5 Install gluster package no

This is quite possible that you install your Redhat 7 / Centos7 on a single path disk i.e. /dev/sda Now your storage engineer enabled all path and you need to enable multipath from now for even /boot  which is /dev/sda1  and for lvm PV /dev/sda2 which has now mounted LV on /  and on /swap So here is the simple solution from redhat -- # yum install device-mapper-multipath -y Above will install multipath package but we would not have any multipath config file at /etc/multipath.conf Create this file now with command # mpathconf --enable Now we will add wwid of our /dev/sda2 device # multipath -a /dev/sda2 wwid '3600d02300069c9ce09d41c4ac9c53200' added Now run multipath again to check .. # multipath Oct 21 09:37:19 | 3600d02300069c9ce09d41c4ac9c53200: ignoring map Now we will rebuild initrd in order to include multipath driver during kernel load. # dracut --force -H --add multipath Done ! you can simply reboot your server.