Fencing and Adding a Failover to Clustering – Part 3

In the previous two guides, we’ve discussed how to install cluster, creating a cluster and adding nodes to the cluster, also we’ve studied how cluster.conf appears to be after the necessary configurations are done.

Today, in this third part of clustering series, we are going to discuss about what is fencing, failover and how to configure them in our setup.

Fencing and Add Failover to Cluster
Fencing and Add Failover to Cluster – Part 3

First of all let’s see what is meant by Fencing and Failover.

What is Fencing?

If we think of a setup with more than one nodes, it is possible that one or more nodes can be failed at some point of time. So in this case fencing is isolating the malfunctioning server from the cluster in order to protect and secure the synced resources. Therefore we can add a fence to protect the resources shared within the cluster.

What is Failover?

Imagine a scenario, where a server has important data for an organization which the stakeholders need the organization to keep the server up and running without any down time experienced. In this case we can duplicate the data to another server (now there are two servers with identical data and specs) which we can use as the fail-over.

By any chance, one of the servers goes down, the other server which we have configured as the fail-over will take over the load and provides the services which were given by the first server. In this method, users will not be experienced the down time period which was caused to the primary server.

You can go through the Part 01 and Part 02 of this clustering series here:

  1. What is Clustering and Advantages/Disadvantages – Part 1
  2. Setup Cluster with Two Nodes in Linux – Part 2

As we’ve already discussed about our testing environment setup in last two articles, that we’re using three servers for this setup, the first server act as a Cluster server and other two as nodes.

Cluster Server: 172.16.1.250
Hostname: clserver.test.net

node01: 172.16.1.222
Hostname: nd01server.test.net

node02: 172.16.1.223
Hostname: nd02server.test.net   

Step 1: How to Add Fencing to Cluster Server

1. First we have to enable fencing on the cluster server, for this I will use below two commands.

# ccs -h 172.16.1.250 --setfencedaemon post_fail_delay=0
# ccs -h 172.16.1.250 --setfencedaemon post_join_delay=10
Enable Fencing on Cluster
Enable Fencing on Cluster

As you can see we use ccs command to add the configurations to cluster. Following are definitions of the options I have used in the command.

  1. -h: Cluster host IP address.
  2. –setfencedaemon: Applies the changes to the fencing daemon.
  3. post_fail_delay: Time in seconds which the daemon waits before fencing a victim server when a node has been failed.
  4. post_join_delay: Time in seconds which the daemon waits before fencing victim server when a node has joined the cluster.

2. Now let’s add a fence device for our cluster, execute below command to add a fence device.

# ccs -h 172.16.1.250 --addfencedev tecmintfence agent=fence_virt

This is how I executed the command and how the cluster.conf file looks like after adding a fence device.

Add Fencing Device in Cluster
Add Fencing Device in Cluster

You can execute below command to see what kind of fence options you can use to create a fence device. I used fence_virt since I use VMs for my setup.

# ccs -h 172.16.1.250 --lsfenceopts
Fence Options
Fence Options

Step 2: Add Two Nodes to Fence Device

3. Now I’m going to add a method to the created fence device and add hosts in to it.

# ccs -h 172.16.1.250 --addmethod Method01 172.16.1.222
# ccs -h 172.16.1.250 --addmethod Method01 172.16.1.223

You have to add the methods you have created while ago for the both nodes you have in your setup. Following is how I added methods and my cluster.conf.

Add Nodes to Fence Device
Add Nodes to Fence Device

4. As the next step, you will have to add the fence methods you created for the both nodes, to the fence device we created namely “tecmintfence”.

# ccs -h 172.16.1.250 --addfenceinst tecmintfence 172.16.1.222 Method01
# ccs -h 172.16.1.250 --addfenceinst tecmintfence 172.16.1.223 Method01

I have successfully associated my methods with the fence device and this is how my cluster.conf looks like now.

Add Fence to Nodes
Add Fence to Nodes

Now you have successfully configured fence device, methods and added your nodes to it. As the last step of part 03, I will now show you how to add a failover to the setup.

Step 3: Add Failover to Cluster Server

5. I use below syntax of commands to create my fail-over to the cluster setup.

# ccs -h 172.16.1.250 --addfailoverdomain tecmintfod ordered
Add Failover to Cluster
Add Failover to Cluster

6. As you have created the fail-over domain, now you can add two nodes to it.

# ccs -h 172.16.1.250 --addfailoverdomainnode tecmintfod 172.16.1.222 1
# ccs -h 172.16.1.250 --addfailoverdomainnode tecmintfod 172.16.1.223 2
Add Nodes to Cluster Failover
Add Nodes to Cluster Failover

As it is shown above, you can see cluster.conf bears all the configurations I have added for the fail-over domain.

Hope you have enjoyed the Part 3 of this series. Last part of the Clustering guide series will be posted soon which will teach you to add resources to the cluster, sync them and start-up the cluster. Keep in touch with Tecmint for the handy HowTos.

Thilina Uvindasiri
I'm a BSc Special (Hons) graduate in Information Technology and works as an Engineer-Systems. Love to work, explore and research Linux and play rugby as a hobby.

Each tutorial at TecMint is created by a team of experienced Linux system administrators so that it meets our high-quality standards.

Join the TecMint Weekly Newsletter (More Than 156,129 Linux Enthusiasts Have Subscribed)
Was this article helpful? Please add a comment or buy me a coffee to show your appreciation.

37 thoughts on “Fencing and Adding a Failover to Clustering – Part 3”

  1. This is the first time i am using Linux cluster so i need your help.

    I had setup vcs and sun cluster before, i have a doubt here how to setup up a quorum disk in centos 7. I have a requirement to setup in prod box which is in physical server it will be a 2 node cluster and few shared lun for the data.

    I am stuck in the quorum. I cannot see no one mentioning shared disk for quorum ( 500MB normally which we give ) to setup. Hope you understand my requirement and my understanding.

    Reply
  2. Hi ,
    CMAN service started fine in Node1 and Node2 but while starting cman at Master server it is getting fail.
    saying Node name cannot found.But the tags for both nodes are present .
    Has anyone came across such error.

    Reply
      • Hi Thilina,

        Below is the error:

        [root@master cluster]# service cman start
        Starting cluster:
        Checking if cluster has been disabled at boot… [ OK ]
        Checking Network Manager… [ OK ]
        Global setup… [ OK ]
        Loading kernel modules… [ OK ]
        Mounting configfs… [ OK ]
        Starting cman… Cannot find node name in cluster.conf
        Unable to get the configuration
        Cannot find node name in cluster.conf
        cman_tool: corosync daemon didn’t start Check cluster logs for details
        [FAILED]
        Stopping cluster:
        Leaving fence domain… [ OK ]
        Stopping gfs_controld… [ OK ]
        Stopping dlm_controld… [ OK ]
        Stopping fenced… [ OK ]
        Stopping cman… [ OK ]
        Unloading kernel modules… [ OK ]
        Unmounting configfs… [ OK ]

        Server details:
        10.91.18.145 master.uic.com – Management node
        10.91.18.143 node1.uic.com
        10.91.18.144 node2.uic.com

        /etc/hosts (10.91.18.145):

        127.0.0.1 localhost localhost.localdomain localhost4 localhost4.localdomain4
        ::1 localhost localhost.localdomain localhost6 localhost6.localdomain6
        10.91.18.143 node1.uic.com node1.uic.com
        10.91.18.144 node2.uic.com node2.uic.com
        10.91.18.145 master.uic.com master.uic.com

        cluster.conf (10.91.18.145):

        All the nodes (management node i.e. master.uic.com, node1, node2 can ping each other.

        Please help.

        Reply
      • Thankyou so much for your response but using this article we are not able to setup cluster on centos 7 becuase of package difference and there is lots of different in configuration as well.

        Reply
        • Vinay, Your question has been raised by another user in Part two. RHEL/CentOS 7 have deprecated the packages I’ve mentioned in the article. But that doesn’t mean you cannot do this in RHEL or CentOS 7.You can use corosync, pacemaker and pcs instead.

          Reply
  3. Thank you so much, this post gave me a better knowledge about clustering than thousands writing out there. However, I still wondering what fence options you can use to create a fence device if you used real machines for your setup?

    And, what is the point of method in fencing? Can you please put more explanation or references resource that I can work to understand them?

    Again, thank you so much for your work.

    Reply
  4. Hi Thilina,
    Fist of all, thanyou very much for this article…it so much helpfull.
    i can´t find the Part4 and…can you post the lisnk pls.
    Best regards.

    PS, sorry for my english.

    Reply
  5. Hi TecMint,
    The concept of clustering had always baffled me. However, this tutorial has cleared up all my doubts.
    Thank you for sharing this.

    Reply
  6. Hi Mr.Uvindasiri
    Thank you very much for your sharing,
    I am waiting for your last post for this serie.
    I hope you will post it soon.
    Thank you again
    Good job !

    Reply
  7. Thank you so much for all the parts ( Linux clustering with two nodes).I like the way it was explained very systematic step by step. waiting for its last part and final part with resource such as webserver,NFS etc.

    Reply
  8. Thank you so much my friend for this third part. I have a question: I tried to set a fence virtual device in a virtual environment, but it doesn’t work for me, in some part of my configuration the node doesn’t come back to the cluster after a failure. So i have added a quorum disk, and finally my cluster works ok (the node goes down and after the failure come back to the cluster), so my question is: what is the difference between a fence device and a quorum disk in a virtual environment?

    Reply
    • Hi Daniel,

      First of all you should know that a quorum disk is more useful in clusters which contains lesser nodes than a multiple node cluster which in case of a node fails. But of course having a quorum disk in any cluster would be great for its well being. In your case, what you have gone through is identified as a “fencing war” in the tech world. What i will do is, I will post a new article about the scenario you have faced in near future. It will answer all your questions and will help to understand more about fencing and quorum disks. Will put a comment here when the article is ready or send me your email and i will inform you. Thanx for your interest and nice comments.

      Reply
  9. Thank you very much for your new post about clustering. I am very much enjoying and waiting for your next post. I would like to request you that if you may have time then please make a series about “Redhat OpenStack”. Actually the world is now moving towards “Cloud Technology” and now we have to learn those cloud technology.

    Reply

Leave a Reply to Thilina Dhanushka Cancel reply

Thank you for taking the time to share your thoughts with us. We appreciate your decision to leave a comment and value your contribution to the discussion. It's important to note that we moderate all comments in accordance with our comment policy to ensure a respectful and constructive conversation.

Rest assured that your email address will remain private and will not be published or shared with anyone. We prioritize the privacy and security of our users.