Overview

As mentioned in my previous posts, PROXMOX VE (PVE) can be run as a cluster. The cluster configuration is based on one node being the master and other nodes being the slaves .  

By implementing a cluster, you will be able to balance the workload on several hosts and you will be able to increase the availability of your virtual machines.  With a PROXMOX VE Cluster, you will be able to perform “live migration” even if you do not have shared storage.  In case of hardware maintenance, you will be able to move “on the fly” the virtual machines on another node with no downtime (or limited downtime).  

Now what’s happens if a cluster node fails ?  If you haven’t implemented any shared storage, you might have no more access to some of your virtual machines. The PROXMOX VE does not migrate your VM from one node to another if a crash occurs. I haven’t tested yet the situation if a shared storage infrastructure or using the DRDB technology.At the moment, we will not dive so deeply in the product and the clustering features. We will come back to these topics in some future posts.  

In order to build the PVE Cluster, we need to install 2 PROXMOX VE servers. Follow the Installation procedure described on  Part I,  

In our example, the PROXMOX VE hosts will have the following settings :  

HOST 1   

• Hostname : NODE1.Study.lab
• IP Address : 192.168.1.10
• Subnet Mask : 255.255.255.0

HOST 2   

• Hostname : NODE2.Study.lab
• IP Address : 192.168.1.11
• Subnet Mask : 255.255.255.0

SETTING UP A 2-NODE PROXMOX VE CLUSTER

You will see that building a cluster with PROXMOX VE is really simple. However, you have to know that the setup cannot be performed through the web interface (at the moment). You will need to use the command line.  You will need to logon locally on the console of each PROXMOX VE hosts and perform the following tasks or you can use your favorite ssh client tool (I’m using Putty.exe) in order to remotely connect to your PROXMOX VE hosts and perform the configuration steps.  

Let’s start !  

Important Note :  

Before trying to create the cluster, you will need to ensure that both PROXMOX VE Hosts are in sync. You have to check that both hosts have the same time. If there is a time difference between the two hosts, you will get error messages within your console. If the time is not identical between the hosts, you might end up with the following error when creating your cluster.  

Ticket authentication failed – invalid ticket ‘root::root::1252946231::f1292eb0564a15169aaf30e36 1382247fccf718c’  

Step 1 : Check time settings on both nodes
  

In part 2, we have shown you how to check the configuration of your system. Let to this again.  

• Open your favorite browser and login into the Web interface on node 1 and node 2
• On the left menu, in the Configuration section, click on system

  

• click on the Time tab

  

• Check that the Time is identical on both hosts

If they are not, you can use the following command line (through putty or console) to set the correct time on the hosts  

date +%T -s “hh:mm:ss”  (ex. : date +%T -s”10:15:53″)   

Note: We recommend to use a NTP server in order to synchronize your clock automatically.  you can set the ntp server by editing the /etc/ntp.conf file.  

Step 2 : Go to your master Node and create your cluster  

In this demo, I’m assuming that the Node1 will be the master node of the future cluster I’m building up    

• Logon into the console or use putty to node1
• Type the following command pveca –help.  pveca is the command line utility that will be used to create the cluster.  As you can see, there not a lot of options ->meaning quite simple to use 🙂

  

•  To create the cluster, you will need to type the following command in the console,   pveca -c

  

• If everything went correctly, you will see the following output

  

Step 3 : Go to your slave  Node and add it to  your cluster  

• Login into your slave server (in this demo node2) through the console or using putty.exe
• To add the server to the cluster, you simply type the following command pveca -a -h <%IP address of the Master Server %>

  

• You will be asked if you want to continue the connection. Type yes. You will then need to provide the password for the master node.  At the end of the procedure, you will have a PVE Cluster

  

• As a final step, you can type pveca -l to check the status of your PVE Cluster

  

That’s it ! Yes, you have typed in 3 commands and you have a working PVE cluster. That’s cool !  

So, let see what’s has changed in the web interface !  

Managing your PVE Cluster through the Web Interface

In order to manager your cluster, you should always connect to the master node. Even you are able to manage both nodes, some operations can be performed only from the Master Node.  For example, I’ve tried to download an appliance to the slave node and had a message error stating that I had no Write access.  If you login into your console, you can notice on the home page  on the top section that you have the information about the cluster and status of the nodes.  

Note : It’s possible that just after creating the cluster, you will get a blinking message specifying that nodes are not in sync.  You will just need to wait a little bit in order to have the nodes to synchronize between each others.  

  

You can click on one of the node in the table and you will be redirected to the management console of the node you have selected (see figure below)  

  

In the administration section (on the left menu), you can click on the cluster link and have access to read only information about your cluster.  you can see the nodes that are member of the cluster. You can also quickly check the status and identify the master node from the other nodes.  

  

How to migrate Virtual Machines from one node to another ?

To finish with this post, i want to outline the process to migrate virtual machines from one host to another one.  For this post, i have download and created an OpenVZ container on my Proxmox VE cluster.  In the VM Manager section (on the left menu), simply click on the virtual machines link. In the right pane, you will see the list of the virtual machines (in my  demo, i have only one).  If you click on the red arrow at the corner, you have a contextual menu where you can select the option “Migrate”  

  

You will be then presented this interface. You have to specify the source node, the target node and the virtual machine to migrate.  You have two options : you can perform an offline migration or an online migration. In this demo (and because there is a known bug with openVz container), we will perform an offline migration.  When you have provided the appropriate information, you can click on the migrate link .  

  

If you migration is successful, you will have an output similar to this one.  

  

Known Issue  

If your are performing an online migration for KVM virtual machines, you should normally have no issues and the live migration will work as expected.  If you are trying to perform an online migration of an OpenVZ container, the migration will fail and you will get a message error similar to “Failed to suspend container”.  

  

This is a known issue with OpenVZ. At the moment, you cannot perform an online migration for OpenVZ. However, you can still perform an Offline migration. You will get some downtime but I have to say that the move/migration was quite fast.  

To have more info about the OpenVZ bug, see the following links  

http://bugzilla.openvz.org/show_bug.cgi?id=850  

http://bugzilla.openvz.org/show_bug.cgi?id=881  

Conclusion

That’s for this post.  To summarize, so far, we have demonstrate how to perform the installation of the PROXMOX HOST Servers. We have presented the interface that’s used to manage your newly installed Virtualization infrastructure. Finally, we have explained how to create a 2-Node Cluster.   At this point, you should have a good comprehension of the PROXMOX VE solution.  Now, it’s time for us to explain how to create virtual machines. This will be explained in Part 4.  

Till Then  

Stay Tuned