Welcome to the Working with NSX Series, focused on the installation and configuration of various components within NSX. You can click me to return to the directory, or choose posts from the menu on the right.

Now that a controller cluster exists, it’s time to prepare the hosts and clusters for NSX. This will provide the necessary user space tools and kernel modules in a non-invasive fashion. I’ll walk through the steps for both a GUI and API based installation.

Cluster Installation

In order to install the NSX bits on a host, you must select an entire cluster for the installer. I would imagine this requirement exists to ensure all hosts inside of a cluster are able to provide the NSX services to the virtual machines. There would be some serious consequences if your VM moved to a host without, say, the firewall functionality and was suddenly exposed.

I’ve chosen to prepare my Lab cluster below, which consists of three ESXi 5.5 hosts. Navigate to the Web Client > Networking & Security > Installation > Host Preparation. Choose your cluster and click the Install link.

You can watch the install tasks from the web or legacy client – for flavor, I mixed it up and watched the tasks kick off in the legacy client.

Note: The ESXi hosts are never put into maintenance mode. I’m running active workloads on the host during this process.

It’s normal to see a status of Installing with a Firewall status of Unknown during the install. It took my hosts a few minutes to install and report back.

Verification

Once completed, the status will change to the version of code running in the cluster. In my case, that’s version 6.0.3. There should also be a Ready checkmark next to each host with a Firewall status of Enabled. I’ve expanded the cluster objects below so that you can see what it looks like.

API Calls

As with the NSX Controllers, you can also make an API request to the NSX Manager to complete this task. To install the software, create a POST request, or use a PUT request to upgrade the software. Note that you’ll need the cluster’s ID as shown below:

POST http://nsx-mgr-ip/api/2.0/nwfabric/configure
&lt;nwFabricFeatureConfig&gt;
&lt;resourceConfig&gt;
&lt;resourceId&gt;domain-c9&lt;/resourceId&gt;
&lt;/resourceConfig&gt;
&lt;/nwFabricFeatureConfig&gt;

POST https://nsx-mgr-ip/api/2.0/nwfabric/configure

<resourceId>domain-c9</resourceId>

</resourceConfig>

</nwFabricFeatureConfig>

Here it is in my handy Advanced REST Client for Chrome:

Next Steps

The hosts are now running all of the required NSX code. In the next post, we’ll focus on Configuring VXLAN and VTEPs so that we can start building logical switches.

36 Responses

ssujai 2014-07-04 at 13:06 | Permalink

Very useful article!

ub 2014-07-09 at 11:21 | Permalink

well done

Basha 2014-07-10 at 03:56 | Permalink

very useful article

j23evan 2014-07-16 at 14:37 | Permalink

Incredibly informative. Thanks for the article!

Raff 2014-07-21 at 17:37 | Permalink

I can’t wait to have a trial NSX copy to perform all the steps you described so well!

Brian Bailey 2014-08-13 at 20:45 | Permalink

Just finished a 2 day NSX class at EMC Office in Conshohoken PA, great potential for this.

[email protected] 2014-08-26 at 10:58 | Permalink

Very useful information.

Brett Parlier 2014-09-18 at 19:20 | Permalink

Great information in a useful format!

Gerben 2014-09-21 at 15:43 | Permalink

Nice series.thanks!

Geoff White 2014-10-02 at 13:37 | Permalink

For one of my clusters, We attempted and uninstall that just hangs at the sand clock, been like that for days. Not sure if the uninstall completed or now. Maybe this is a cosmetic bug that will disappear if I reboot vCenter (which is not something I want to do). In this state, I can’t re-install either, any ideas anyone?
http://dl.dropboxusercontent.com/u/6349310/install.png

Chris Wahl 2014-10-02 at 13:47 | Permalink
Hmm. You could try bouncing the NSX Manager appliance or the vCenter’s Web Client service. Also, check the hosts that are having their software bits uninstalled and see if they are in a “reboot pending” status.
Reply

Geoff White 2014-10-02 at 16:59 | Permalink

Hmmm, must be some persistent stale data somewhere. I bounced the NSX manager, restarted vmware-client on the VCA and migrated all VMs in that cluster and did a rolling restart of each esx host, then restarted NSX Manager and vmware-client process a second time, but the uninstalling status still exists 🙁
Any other ideas anyone?

NSX Link-O-Rama | vcdx133.com 2014-10-08 at 01:21 | Permalink

[…] Part 4 – Working with NSX – Preparing The Cluster and Hosts by Chris Wahl […]

Working with NSX - Deploying NSX Controllers via GUI and API - Wahl Network 2014-10-29 at 07:16 | Permalink

[…] That’s it! There is now a 3 node controller cluster for the NSX environment that will be used to handle control plane activities for the virtualized network. The next section focuses on Preparing The Cluster and Hosts. […]

samcer 2014-12-05 at 06:59 | Permalink

Speaking about VM migration (seeCluster Installation) I don’t think it can be moved away from the cluster since it uses a logical switch port, am I wrong?

Chris Wahl 2014-12-05 at 08:32 | Permalink
@samcer – If the VDS spans multiple clusters, than a VM could potentially move to a cluster that does not have NSX installed. This is covered in greater detail here: http://virtualclouds.biz/2014/12/01/caveat-multi-cluster-vds-design-nsx/
Reply