VSAN – Page 35

Exploring VSAN APIs Part 2 – Query available SSDs

03.05.2014 by William Lam // Leave a Comment

When enabling VSAN, users have the option to specify how the disks (SSD/HDD) will be added to VSAN either automatically (recommended) or manually claimed by the end user.

If you decide to manually claim your disks on each ESXi host within the VSAN cluster, then you will probably need to figure out which disks are SSD and HDD for you to create your VSAN disk groups. If you have used the vSphere Web Client to perform this operation, you will notice there is a toggle for displaying "eligible" disks for VSAN which can also be obtained programmatically using the vSphere API.

On each ESXi host, there is a vsanSystem manager that provides various methods for disk management and the one that we are interested in is called QueryDisksForVsan(). To demonstrate this functionality, I have created the following sample vSphere SDK for Perl script called vsanQueryDisksForUse.pl. The script will query each ESXi host in a vSphere Cluster and return a list of disks which specifies if they are eligible for use with VSAN and whether the disk is an SSD or not.

Disclaimer: These scripts are provided for informational and educational purposes only. It should be thoroughly tested before attempting to use in a production environment.

To check for the list of available SSD/HDD, run the following command:

./vsanQueryDisksForUse.pl --server vcenter55-1 --username root --cluster VSAN-Cluster

Using this information, you can then use the AddDisks_Task() method to create your own VSAN Disk Groups on each ESXi host. As you can see, it is a couple more steps if you opted for manual mode and I would recommend letting the vSphere platform automatically take care of this for you.

Exploring VSAN APIs Part 1 - Enable VSAN Cluster

03.04.2014 by William Lam // Leave a Comment

A couple of weeks back I had provided an overview on some the most common VSAN operations and the specific vSphere APIs that were required to perform the operation. It was no surprise to me that there were some folks already interested in automating the management and consumption of VSAN, I know that was one of the first thing I thought about when playing with VSAN in my home lab.

The great thing about having an API is that it can be consumed in a variety of ways, you can use one of the vSphere SDKs supporting languages like Java, .NET, Perl and even PowerShell with the new VSAN specific cmdlets. I thought it might be useful to explore some of these APIs in greater detail and even provide some working sample scripts that can help others looking to automate VSAN. I know for myself, I learn best when I can see working examples.

Disclaimer: These scripts are provided for informational and educational purposes only. It should be thoroughly tested before attempting to use in a production environment.

VSAN is enabled at the vSphere Cluster level and just like you do today with vSphere HA/DRS, it is just a couple of clicks. To demonstrate the ability to enable, disable and to check current state of VSAN on a vSphere Cluster, I have created the following sample vSphere SDK for Perl script called vsanClusterManagement.pl

There is a vsanClusterConfigInfo property that is part of a vSphere Cluster (ComputeResource) that specifies whether VSAN is enabled and if it is configured to auto-claim or manual mode. To make changes to a vSphere Cluster, you will need to use the ReconfigureComputeResource_Task() method which is demonstrated in the sample script.

To check the current VSAN configuration for a given vSphere Cluster, you can run the "query" operation:

./vsanClusterManagement.pl --server vcenter55-1 --operation query --cluster VSAN-Cluster

To enable VSAN, you can run the "enable" operation along with an optional --autoclaim option which defaults to enable for automatic claiming:

./vsanClusterManagement.pl --server vcenter55-1 --operation enable --cluster VSAN-Cluster

If we re-run the query operation, you will now notice there is more information about the VSAN Cluster including VSAN Cluster UUID, the current claim mode and all the ESXi nodes in the cluster along with their VSAN NODE UUID.

If login to your vSphere Web Client, we should see that VSAN has been successfully enabled just like you would if you were to manually perform the operation through the UI.

If you decided to select "manual" instead of having VSAN automatically claim the SSD and HDD for you, then you will probably need to know which disks are eligible for VSAN. In Part 2 of this series, we will take a look a look at how we can identify available SSDs for use with VSAN.

Why you should rename the default VSAN Datastore name

02.18.2014 by William Lam // 10 Comments

Something that I have noticed early on when working with VSAN is that the VSAN datastore name is automatically selected for you and will always default to vsanDatastore. This is not a bad thing as it simplifies the setup of VSAN which is already dead simple. To enable VSAN, it is simply checking a box in the configuration of your vSphere Cluster just like you would when enabling vSphere DRS or HA.

From an operational standpoint, I think this can be a potential issue. What happens when you have a multiple VSAN Clusters under the same vSphere Datacenter object or even across different vSphere Datacenter objects? I recently came across this while re-building one of my lab environments, can you spot the problem in the screenshot below?

Here we have two VSAN Datastores within the same vSphere Datacenter. Datastore names under a single vSphere Datacenter must be unique. I am sure some of you may have already experienced this with local datastore names using the default "datastore1". vSphere will automatically append "(n)" to the name where n is the number of instances seen in the environment.

In this next example, we have two VSAN Datastores but they are split across two vSphere Datacenters. Since unique name is enforced at the Dataenter boundary, it is even worse now because you have the same name used across both VSAN Clusters.

Even though a user will most likely drill down into a specific environment when provisioning a Virtual Machine, I think it is critical to have a good naming scheme for your infrastructure. What happens when you have someone run an audit of all your VMs and all they showed for their datastore name was "vsanDatastore" and you had multiple VSAN Clusters? Would it not be more useful to have a bit more details in the datastore name to help map it back to some logical/physical container that is a bit more meaningful? I personally would think so and would help in times of troubleshooting where time is of the essence.

Luckily this can easily be remediated since a VSAN Datastore operates just like any other datastore in which you can rename it. VSAN itself does not use the name to uniquely identify the datastore, it implements a unique UUID for each of the VSAN datastores. Simply renaming the VSAN Datastore to something more appropriate such as including the name of the vSphere Cluster will come a long when you need to correlate a Virtual Machine back to a particular compute/storage resource.

I wonder if it would be a useful to have a feature in VSAN to automatically append the vSphere Cluster name to the default VSAN Datastore name? What do you think?