Tag Archives: nutanix

Upgrading CVM Foundation via API

Unconfigured Nodes can easily be upgraded

Following on from the Foundation Central introduction, the nodes should be running Foundation 4.5.1+ as minimum. There could be a scenario where the Foundation version in the CVM is older, given that the factories ship software that may be a few months behind the latest versions.

How can you get Foundation on the nodes up to a newer version before imaging?

Fortunately, upgrading is easy via the API which can be accessed via the CLI or a browser and I’ll cover both methods below.

Part 1: The CVMs have Internet access

CLI Method with Internet Access

If the CVMs have internet access the latest available update on the Nutanix Portal will be auto installed when the API is called. Here we are starting with a node with Foundation 4.4.3 installed and we want to update it to the latest available (4.5.3 at the time of writing).

To determine the current version on the node:

nutanix@NTNX-B-CVM:~$ cat foundation/foundation_version
Foundation-4.4.3-abc86afd

Or use the API to get the current version:

nutanix@NTNX-B-CVM:~$  curl -X GET --header "Accept: application/json" "http://172.16.9.57:8000/foundation/version" ; echo ""
4.4.3

Execute the API call to upgrade Foundation:

nutanix@NTNX-B-CVM:~$ curl -X GET --header "Accept: application/json" "http://172.16.9.57:8000/foundation/auto_update_foundation"

That’s all there is to it. Now just wait ~2 to 15 minutes depending on your connection speed to the Nutanix Portal for the binary (~1.6GB) to download and install, then verify:

nutanix@NTNX-B-CVM:~$cat foundation/foundation_version
foundation-4.5.3-815d6c96

Done!

GUI Method with Internet Access  

  1. Navigate to a CVM via http://your-cvm-ip:8000/docs
    If a node (unconfigured) has DHCP and has registered with Foundation Central you can determine the IP address of the CVMs from the Foundation Central home page
  2. You can check if an Foundation update for the CVM is available using this API :  /is_update_available
    Find that on the API Explorer page, expand it and click “Try it out!” button
  3. Verify you get Response Code 200 after being patient (~10 to 30 seconds)
  4. If a Foundation upgrade is available, you can now upgrade. Upgrade the CVMs Foundation to the latest version using this API :  /auto_update_foundation and again expand it and click “Try it out!” 

Same as before it can take several minutes depending on your connection to the Nutanix Portal. Verify you get a Response Code 200.

If the CVMs have internet access, you don’t need to provide any parameters to these APIs, all you have to do is to hit “Try it out!” on that screen. Find the /version API on the Explorer page after the upgrade to verify the latest version is now on the node. 

Done!

Part 2: If the CVMs have no direct Internet access

If the nodes cannot contact the Nutanix Portal directly, you will need the foundation upgrade binary from the Nutanix Portal and apply the update manually.

For example, upgrading from v4.4.3 to v4.5.3 using a manual file:   

  1. Get the “Foundation Upgrade for CVM or Standalone Foundation VM” file from the Nutanix Portal (eg. foundation-4.5.3.tar.gz
  2. Create a directory on the CVM you want to be upgraded : /home/nutanix/foundation_updates
  3. Copy the foundation-<version>.tar.gz file to this location

CLI method via manual update file 

Now that we have the binary uploaded, we can initiate the upgrade.

nutanix@NTNX-B-CVM:~$ cat foundation/foundation_version
foundation-4.4.3-abc86afd

nutanix@NTNX-B-CVM:~$$ ls ~/foundation_updates/
foundation-4.5.3.tar.gz

nutanix@NTNX-B-CVM:~$ curl -X GET --header "Accept: application/json" "http://172.16.9.58:8000/foundation/auto_update_foundation?tar_file=foundation-4.5.3.tar.gz"

<wait ~1-2 minutes>

nutanix@NTNX-B-CVM:~$cat foundation/foundation_version
foundation-4.5.3-815d6c96

Once complete, remember to delete the file you uploaded to /home/nutanix/foundation_updates to conserve CVM space.

Done !

GUI method via manual update file  

  1. Navigate to the CVM via http://your-cvm-ip:8000/docs and the API Explorer will appear. Expand the /auto_update_foundation section. 
  2. Make the API call via the /auto_update_foundation and add the filename uploaded and click “Try it out!”.
  3. Foundation will be upgraded after a few minutes. Verify by either querying the API  /version or via the CVM:  ‘cat ~foundation/foundation_version’
  4. Done!

Once complete, remember to delete the file you uploaded to /home/nutanix/foundation_updates to conserve CVM space.

Browsing to the update Foundation API via the Explorer on http://cvm_ip:8000/docs

We plan to incorporate upgrading CVM Foundation on nodes detected by Foundation Central soon, so you can centrally update all your nodes easily.

Nutanix Foundation 4.1

Foundation 4.1 brings a bunch of improvements to make Nutanix deployments even easier.

The Foundation Golden Rule

Always upgrade and use the latest available version of Foundation” prior to your deployments.  Each new version has new features (like the below) as well as additional new platforms and software supported by Nutanix. If you are using an old version you are missing out !

foundationinstallmethod

Foundation Release Notes help determine which method to use

Upgrading is easy. Download the tar.gz from the support portal. In Standalone VM Foundation, click the version number and follow the prompts. In CVM Foundation use the java applet to upgrade the inbuilt Foundation on the discovered nodes.

Let’s get into what is in 4.1.

 

 

Foundation UI Changes

CVM Foundation and Standalone VM Foundation in 4.1 have a more unified UI and workflow. However, Standalone Foundation VM is the only method that allows you to manually add nodes (instead of using discovery) or abort an in-progress imaging session.

Range Auto-fill option has been moved to the “Tools” menu to keep the default UI clean.

Screenshot 2018-06-23 14.08.47

Foundation now generates auto-filled host names without a hyphen by default.

CVM Foundation now supports imaging nodes without forming a cluster. This was already supported in Standalone VM Foundation.

Standalone Foundation VM (via discovery) can image nodes or create a cluster without IPMI being required. This mirrors the CVM Foundation experience in Standalone VM Foundation.

Faster Deployments via Skipping AOS Imaging

On discovered nodes, Foundation 4.1 can now skip imaging AOS (if all nodes are the same AOS version) and allow you to just change the underlying hypervisor whilst keeping the AOS version as-is. This saves downloading and re-uploading the AOS image file when the AOS version isn’t changing. This is a time saver.

Lets say your nodes arrive from the factory with AHV and AOS version 5.5.1. You do not need to upload a new AOS image during the Foundation process if you just wanted to change to ESXi for example.

Screenshot 2018-06-24 11.34.22

 

Screenshot 2018-06-24 11.34.44

As an example, below I selected ESXi 6.5U1 to change the primary hypervisor for the cluster. By default Foundation will image all nodes to ESXi, but in my case I wanted to select nodes C and D to remain on AHV in ‘Storage Only’ or ‘Storage Node’ mode. This way I could have all nodes participate in the storage pool but only need to license ESXi on nodes A and B.

Screenshot 2018-06-24 11.35.55

Once you click “Start” you can relax – it will take about an hour to complete the process. Note that if you aren’t changing AOS or hypervisor images at all, an install takes just a few minutes total (just applying your IP addresses).

41ImagingESXIandAHV

Imaging nodes with different hypervisors (ESXi and AHV for “Storage Nodes”) at the same time !

Side note : of course the AHV nodes finish first! (Though no one is surprised right?! :)

Screenshot 2018-06-24 12.30.57

All done! 

Screenshot 2018-06-24 12.51.55

Pre-Configuration Portal JSON file creation and import

Foundation can import configuration files generated at https://install.nutanix.com. This can be used to generate a Foundation JSON file that can be imported into Foundation 4.1+ on site. You can create a configuration manually ,or if you have placed an order for Nutanix NX nodes, you can see your order and auto-populate the nodes (while the nodes are being shipped for example).

Below are some screenshots from the install.nutanix.com site. The idea is that the UI should look very similar to the normal Foundation process.
EOS1

 

EOS2.PNG

Note that the “Import New Order” function only applies to Nutanix NX orders. We are investigating ways to expand this to other appliance types in a future release.

Once you’ve completed the fields, click the ‘Download’ link at the bottom of the page to generate your JSON file.

Other Fixes in 4.1:

Multi-homing is supported when IPMI configuration is skipped.

Foundation no longer fails if a NTP server is supplied but not reachable with AOS 5.6+

Summary:

Nutanix Foundation continues to evolve. We’ve got some more UI changes and features planned around better networking options and more central deployment methods, with the eventual goal that anyone would be able to deploy their Nutanix cluster with any hypervisor they choose, any model of node they choose, in any configuration they want and eventually any cloud with just a few clicks, and in an automated fashion.

Would love to hear your suggestions and feedback on Foundation to make it better.

 

“Remote” Bare Metal Foundation

One of the little known options when using “Bare Metal” Foundation is doing so over a layer 3 network, instead of the traditional “same layer 2 network + MAC address” method.

This allows Foundation imaging of Nutanix nodes over a (good!) WAN link or across different subnets in your DC for example.

Foundation-SiteA-SiteB

This method can be used to remotely ‘Bare Metal’ any hardware vendor platform running Nutanix via IPv4 – Nutanix NX, Lenovo HX, Dell XC, Software Only Cisco and HPE and others.

Foundation-Remote-Quote
Quick Summary of the “Remote Bare Metal Foundation” procedure:

  1. Rack and cable the nodes, and configure the IPMI ports on the network with an IPv4 address (eg. via BIOS see below). Do this first. 
  2. Deploy the Foundation VM on the network – ensuring it has IPv4 connectivity to the IPMI ports. The VM does not need to be on the same subnet as the IPMI ports and could be in a different site over a WAN.
  3. Go through the Bare Metal install process via the Foundation VM, skipping discovery and instead manually adding blocks/nodes via selecting the “I have configured their IPMIs to my desired IP addresses” option.

Critical Note on WAN Bandwidth Requirements

With this method you will copy AOS + Hypervisor image files over the network in parallel to each and every node – so consider available bandwidth and network utilisation as well as the AOS / Hypervisor image sizes that will be transferred from your Foundation VM to the nodes during the imaging process.

These files can be several GB in size. Foundation pushing images to nodes will time out after 15 minutes – so you will likely need a WAN link minimum of 50Mbit/s to copy the 4GB AOS file to a SINGLE node…and a better link if you are changing to ESXi (additional ~350MB) or HyperV (additional ~4GB) or if you are imaging more than one node.

If you have 4 nodes – multiply that by 4 of course. Clearly, this method is not for your small branch ROBO link. Use a tool like https://techinternets.com/copy_calc to see if your WAN link can handle the workload within that timeframe.

At time of writing you cannot modify the timeout setting.

Foundation-FileXfer-WAN2


In summary, ensure your network link is capable of respecting the timeout value taking into account the number of nodes you are imaging. For example, if you were imaging 4 nodes over the WAN, you will be copying at least 16GB in total over that link within 15 minutes.

Screenshot 2018-05-23 09.46.39

If you had a 1Gbit link (or local 1 Gbit switch), 20 nodes would take ~12 minutes just for the AOS images. If you are imaging HyperV nodes, you could only image 10 nodes (as you need to include the 4GB HyperV ISO as well) on 1Gbit links. This is why old 100Mbit switches or USB adapters won’t suffice when you are imaging multiple nodes. 

Site A and Site B can be different L3 subnets. Make sure Site A’s Foundation VM subnet and Site B’s IPMI subnet and Site B’s CVM/Hypervisor subnet are all routable to each other. That is, every subnet involved must be routable to each other.

Setting the IPMI Ports Manually

If you are unsure how to set the IPMI IP addresses manually, see “Setting IPMI Static IP Address” section in the Foundation Field Installation Guide for instructions for configuring via BIOS on each node.  The Foundation Field Installation Guide can be found on the Nutanix Support Portal.

Foundation-BIOS-IPMI

The above screenshot is from one node’s IPMI settings via BIOS. You would repeat this for each and every node you want to deploy, then use Foundation to image the nodes.

Quick UI Walkthrough

Below is a walkthrough of the initial screens in Foundation v4.1 for the Bare Metal via IPv4 process. Note that the IPMI addresses you type should match the IP addresses you’ve manually assigned to the nodes of course  :

FNDN-41-ipv4IPMI

We are also developing a “Foundation Central” microservice within Prism Central which will allow for ‘zero touch’ deployments at scale, including using a local (to the nodes) file store to avoid pushing files over the WAN – but for now this ‘bare metal’ method works if you have the luxury of bandwidth.