AcloudA

Building a Cluster based on VMware (part III)

Posted by Veniamin Kireev on February 19, 2018
No ratings yet.

In the first article Building a Cluster based on VMware (part I) we saw Cluster Technologies types and understand how to install VMware ESXi, in the second article Building a Cluster based on VMware (part II) we look at how to deploy the vCenter Server Appliance (VCSA) 6.5. In this, a final part we will look at how to configure VMware HA Cluster.

High Availability (HA) – the clustering technology is designed to increase the availability of the system, and in case of failure of one of the ESXi nodes, it is possible to start its virtual machines on other ESXi nodes automatically, without the participation of maintenance personnel.

To create a cluster, we need all the virtual machines to be within the same disk storage, it can be not only hard disk storage but also be software. VMware implemented this with vSAN.

First of all, let’s look on special terms we used and for a more clear understanding of the process of configuration.

  • High Availability (HA) – technology of high availability of ESXi nodes;
  • Isolation Response (IR) – a parameter that determines the action of the ESXi-host when it stops receiving cluster access signals. When creating a cluster on each ESXi-host, the HA Agent is installed, which will exchange the availability signals (Heartbeat);
  • Reservation is a parameter calculated on the basis of the maximum individual characteristics of all VMs in the cluster and is further used to calculate Failover Capacity;
  • Failover Capacity (FCap) – a parameter that determines the real fault tolerance. It is measured in integer numbers and indicates which maximum number of servers in the cluster can fail, after which the cluster itself will continue to function;
  • Number of Host Failures allowed (NHF) – the parameter is set by the administrator. Defines the target level of fault tolerance. This number of ESXi nodes can simultaneously fail;
  • Admission Control status (ADC state) – is automatically calculated as the ratio of Failover Capacity to Number of host failures allowed;
  • The Admission Control parameter (ADC parameter) is assigned by the administrator. Determines the behavior of virtual machines when there are insufficient slots to start them;
  • Restart Priority (RP) – the priority of starting machines after the fall of one of the ESXi nodes that are part of the cluster.

Now we almost ready, and need to consider the algorithm for creating a cluster:

  1. Determine the number and size of slots on the ESXi nodes (Reservation);
  2. Set the value of Number of Host Failures allowed (NHF);
  3. Calculate the parameter Failover Capacity (FCap);
  4. Compare NHF and FCap. If NHF is larger than FCap, we need:
  • – buy a new server;
  • – lower the value of NHF;
  • – optimize the Reservation;
  • – set the ADC parameter to Allow the virtual machine to be started even if they violate availability constraints;
  1. Set the Admission Control parameter to one of the states;
  2. Determine the behavior of the host in case of lost receiving of the availability signals from the remaining nodes (Isolation Response);
  3. Assign virtual machines to Restart Priority;

Isolation Response (6)

A script that runs if there are no cluster availability signals is determined by the value of Isolation Response, determines the actions of the ESXi node if it does not receive the Heartbeat signal. This happens if the ESXi host was either isolated from the cluster, as an example – a network card failure.

All events can be determined by two scenarios:

  1. Failure to send or receive signals, while everything else functions.
  2. The network ceased to work between the cluster and the individual node, but the ESXi node is still running.

In the first case, we use the value of Isolation Response – “Leave powered on”, then all VMs will continue to work.

In the second case, it is necessary to select Power off or Shutdown (used by default) if the node stops receiving signals, HA will allocate the VM to the nodes remaining in the network, and the failed node must finish the work, so there would not be a conflict between the VMs.

Reservation of Resources (Reservation)

When calculating the Failover Capacity, the cluster creates slots defined by the Reservation parameter. It is calculated according to the maximum size of VMs on all working nodes.

Failover Capacity parameter

After Failover Capacity is determined, it determines the maximum number of nodes in the cluster that can fail. In this case, all VMs will work.

And a few recommendations before we start:

  1. For a cluster with the HA, service enabled, all virtual machines and their data must be on a common data store (Fiber Channel SAN, iSCSI SAN, or SAN iSCSI). This is necessary to enable the VM on any of the cluster hosts. This also means that hosts must be configured to access the same virtual machine networks, shared memory, and other resources.
  2. Each ESXi server in the HA cluster monitors the network nodes to detect server failures. To ensure that availability signals are not interrupted, it is recommended that you set up redundant network paths. If the first network connection of the node has ceased to function, the heartbeats will begin to be transmitted through the second connection. To improve fault tolerance, it is recommended to use two or more physical network adapters on each node;
  3. If you want to use the DRS service in conjunction with HA to distribute the loads to the nodes, the cluster nodes must be part of the vMotion network. If nodes are not included in vMotion, DRS can misallocate the load to nodes;

Let’s proceed to configure the cluster.

1 Open vCentre and on the main page click “Create Datacenter”:

VMware - vCenter - Getting Started

2 Enter the Name of the Datacenter:

VMware - vCenter - Getting Started - New DataCenter Name and Location

3 Create Cluster:

VMware - vCenter - Creating a new Cluster

4 Set the Name, do not use the rest and click OK:

VMware - vCenter - New Cluster Name

5 Next, after descending on the tree to the cluster “Add a host”:

VMware - vCenter - Add a host

6 Set IP address or Network node Name:

VMware - vCenter - Add Host - Name and Location

7 Set Username and Password for ESXi nodes:

VMware - vCenter - Add Host - Connection Settings

8 Answer Yes to Security Alert:

VMware - vCenter - Add Host - Connection Settings - Security Alert

9 Let’s check the information on the added node:

VMware - vCenter - Add Host - Host Summary

10 Next, we see the license management window on the node, we can add the necessary ones and click Next:

VMware - vCenter - Add Host - Assign License

11 Determine the level of blocking the console node – the description is clear:

VMware - vCenter - Add Host - LockDown Mode

12 Location VM and click next:

VMware - vCenter - Add Host - VM Location

13 And that is all, we have done:

VMware - vCenter - Add Host - Ready to Complete - Finish

Repeat steps 5-13 for each node in the Cluster.

At this stage, we get a non-configured cluster, without a shared storage and any add-ons.

Let’s create a distributed switch for vSAN.

Right-click on our Datacenter, go to the Distributed switch and click New:

VMware - vCenter - Distributed Switch - New Distributed Switch

Set the Name for the list:

VMware - vCenter - New Distributed Switch - Name and Location

Choose the generation of the switch. The generation is determined by the oldest generation in the network:

VMware - vCenter - New Distributed Switch - Select Version

Then we set the number of ports, if necessary, specify the name of the port group:

VMware - vCenter - New Distributed Switch - Edit Settings

Next, go to the settings created by the switch, and add the Hosts participating in the cluster to it:

VMware - vCenter - New Distributed Switch - DSwitch

Click Add host:

VMware - vCenter - Add and Manage Hosts -Select task

Add a New hosts:

VMware - vCenter - Add and Manage Hosts - Select Hosts

Choose hosts that need to be added:

VMware - vCenter - Add and Manage Hosts - Select new Hosts

And click Next, after checking information:

VMware - vCenter - Add and Manage Hosts - Select Hosts - Check and Next

Determine where we want to add ports (you can leave by default):

VMware - vCenter - Add and Manage Hosts - Select network adapter tasks

Now assign uplinks for the port group:

VMware - vCenter - Add and Manage Hosts - Manage physical network adapters

Create new adapters. To create, just click New adapter. Select the port group associated with the switch, and click Next:

VMware - vCenter - Add and Manage Hosts - Manage physical network adapters - Add Networking - Select target device

Then specify the services for which the switch and the IP address of this virtual adapter are created. Click “Next” to complete the creation of the VMKernel adapter:

VMware - vCenter - Add and Manage Hosts - Manage physical network adapters - Add Networking - Connection Settings - Port properties

Check all information and if its correct – click Finish:

VMware - vCenter - Add and Manage Hosts - Manage physical network adapters - Add Networking - Connection Settings - Ready to complete

Re-create the VMKernel adapters for each ESXi host. In the end, you should have something like below:

VMware - vCenter - Add and Manage Hosts - Manage VMKernel network adapters

Before configuring, the wizard will analyze the impact. Once everything is in order, click “Next”.

Add a storage device vSAN:

Go to the cluster settings. vSAN – General – Configure.

Determine the appropriate scenario and click on it further:

VMware - vCenter - Configure vSAN - vSAN capabilities

The system will check whether vSAN is allowed on all hosts on the network:

VMware - vCenter - Configure vSAN - Network Validation

We mark disks for the cache and for storage in the manager vSAN:

VMware - vCenter - Configure vSAN - Claim disks

Then finish the settings.

Conclusion

So we got the VMware based Cluster with the vSAN storage.

This procedure is not very simple and has a lot of configuration parameter, that is why this series of articles is not a complete instruction for setting up a cluster in production, and it does not give an accurate knowledge of how everything works but will give a general understanding of the configuration.

We wish you a greater experience in Clusters technology and you always may learn more about it on official VMware site.

 

Related materials:

Views All Time
3
Views Today
6
Return to all posts

What is the difference between VMware ESXi, vCenter vSphere?
Building a Cluster based on VMware (part II)

Please rate this

The following two tabs change content below.

One Response to “Building a Cluster based on VMware (part III)”

  1. […] the next article (p III), we’ll look at how to how to configure VMware HA […]

Leave a Reply

You must be logged in to post a comment.