Hello Everyone! I hope you’re having a good end of the year.
On today’s post, I will talk about a specific use case and one of the ways to solve it:
The business need was to do the following:
Have a single master template with dozens of inputs, to fullfill every deployment need.
This master template will be consumed by multiple projects and different users
Availability of inputs need to change based on the consumer / project
Visibility of inputs needs to change based on the consumer / project
Format of inputs needs to change based on the consumer / project
Source (in case of external data) for input data needs to change based on the consumer / project
You can see that we’re hitting a few limitations in the vRA OOB code:
The ‘conditional’ values for existence, visibility, format, etc, for a service broker form inputs don’t allow for this level of customization
A single cloud template can’t have more than one service broker form attached to it
So how do we solve this business need? Here comes the API solution!
Important: all the API information in swagger format (everything that was used in making this solution) is available in your vRA instance at https://VRA_FQDN/automation-ui/api-docs
API Documentation URL
What does the solution need to accomplish?
I will make a quick summary of what the code needs to do
Grab the inputs from a service broker form (and the requester ID, this is important)
Save the actual deployment name, but make the API deployment (mapped to the vRO workflow) use a temporary
Use the inputs to deploy the master template blueprint via the API (since we’re using the API and we don’t have access to the requester’s credentials, the API call will be executed by an administrative account configured in vRO)
Poll the master template blueprint deployment until it is successful
Once it is successful, change the owner to the original requester
Once that is done, destroy the temporary deployment that generates the API one.
These inputs (and a bit more) are the ones that are going to be used in the API call to the blueprint-request API -> part of that code. I create the body of the call with the inputs
Now those inputs need to be part of the action and the workflow we’re going to use, as you can see here:
vRO Action Inputs
You can see that we have some extra inputs that are not in the blueprint!
All the REST configuration will be variables of the workflow that is calling the action. This will use the previously configured REST host and the credentials
ProjectId is needed for the deployment, and we get that from the Service Broker Form
ownerId is needed for the deployment change owner action, and we get that from the execution context of the workflow that is being called by the service broker form
realDeploymentName is the actual deployment name (remember that the Service Broker form will use the temporary deployment name, and the actual deployment will use the name you input.
So what does the structure look like? In this case, we have two offerings for two projects:
Each of them have different configurations for the fields, for example:
projectId and temporary deployment name are visible fields, the rest is editableprojectId and temporary deployment name are not visible, plus, a bunch of fields aren’t editable
These two service broker forms map to two vRO workflows:
The two workflows
Why do I need two workflows? because as mentioned before, I can’t use two service broker forms for the same catalog item. However, these two vRO workflows are just wrappers of the ‘main’ workflow
The ‘offering’ workflow is just a wrapper for the base workflow
However, there is an important caveat here. The ownerId will only be part of the workflow that is called by Service Broker, in this case, the wrapper one. So the information for the requester is on the execution context of this workflow.
And I need to pass that to the base workflow. So how do I do that? By extracting the property from the execution context, and then passing it to the base workflow.
The only thing left now is to destroy the temporary deployment after it is done with the API one. So how do we do that?
We have a workflow that will delete a deployment via API – It basically calls the the deployments API with a DELETE action
//delete Deployment
System.log(deploymentId)
var request = restHost.createRequest("DELETE", "/deployment/api/deployments/"+deploymentId);
request.contentType = "application/json";
request.setHeader("accept", "application/json");
request.setHeader("Authorization", "Bearer " + tokenResponse)
//Attempt to execute the REST request
try {
response = request.execute();
jsonObject = JSON.parse(response.contentAsString);
System.log(response.contentAsString)
}
catch (e) {
throw "There was an error executing the REST call:" + e;
}
The deploymentId comes from the subscription run. Since this is run from the Event Broker state ‘Deployment Completed’, we have the deploymentId available there
deploymentId parameter on the event broker state
So we just extract it (the same way we did with the execution context) but from the inputProperties (payload used in the event broker subscriptions)
So how does all of this look in an actual run? Let’s show it in a video! (takes around 6 minutes, you will see the temporary deployment being generated, then the actual one via the API, the owner name change, and the destruction of the original one.
Deployment DEMO!!!
And I’m going to attach the code of the two most important actions that form the workflows (deployViaApi and deleteDeployment) here:
I hope you enjoyed reading and understanding this as much as I did while trying to come up with this solution and then making this a reality. This idea can be heavily customized to suit other use cases (and grow this one way more) but the principles used here should still apply!
Please leave your feedback in the comments if you liked it, and share it!
On today’s post, we will dive into a practical example to use vRealize Orchestrator Configuration Elements to help with a business need for vRA 8 Custom Forms!
The Problem
Customer X is using a single cloud template with multiple inputs backed by vRO Actions. The main input, and what defines pretty much all the rest, is the project selected. Given the complexity of the inputs, the cloud template can be used by all projects and by many different use cases.
Customer X was trying to improve the form loading times, which were around 10 seconds for the initial loading, plus 10 more seconds every time they changed the project in the form. This heavily impacted the user experience since it was giving a sensation of ‘slowness’ overall to anyone that was requesting the items.
The project defines, for example (there are more fields, but these are the ones we will use as example):
Hostname prefixes
Puppet roles
AD OUs
Portgroups
vCenter Folders
Each project has a ‘Project Map’ which contains different modifiers to then perform a different search in an external API, which has a cache of the objects needed, to reduce the time needed to gather the data (for example, sending API calls to vCenter to get folders)
However, the fact that the Project Map does not have all the information and needs to be processed in real time ends up adding more loading time to the form than desired.
A solution: vRO Configuration Elements
Emphasis on ‘A solution’ and not ‘THE solution’ since there could be other (even better!) ways to solve this problem, but this is how I approached it and will show it in this blog post.
vRO configuration elements, are originally used for example, for sets of variables that will be used in multiple internal actions/workflows, to avoid having the same data in many places, and for ease of managing. The configuration elements can be referenced in workflows or actions and the information is only changed in a single place.
However, there is another use we can give to configuration elements and that is using them as a Database!
All the configuration elements reside in the vRO DB, and the elements used can be of any of the types that exist within vRO.
Creating a Configuration Elements category called ‘Projects’
Create one configuration element per project, within that category. The easiest way to accomplish this is to create one configuration element, define all the needed attributes, and then just duplicate that configuration element to match all the projects you need – In this case, since we need to retun this to vRA Custom Forms, mostly in drop-down form, we will be using string arrays
One configuration element per project, with the variables mentioned
An action that will return the values to the custom forms, using two inputs, the Project we want to get the information from, and the value that we want to get. That makes the action reusable by multiple fields in the form: In this case, I called it getConfigurationElementValue and it can be seen on the following link: https://github.com/luchodelorenzi/scripts/blob/master/getConfigurationElementValue.js
An action or workflow that will:
Get the data from the external API
Populate the configuration elements with that data
This action/workflow can be scheduled to run every minute, every 5 minutes, depending on the need.
The data will be persisted in the vRO DB so that’s why I’m calling this a ‘database’ instead of a cache, however, it could very well be called a ‘persistent cache’ since all it is doing is to make the data available to the user as fast as possible but not doing any processing.
This workflow runs every 5 minutes and updates the values on all the existing projects (Configuration Elements)
The important thing to note here is that there isn’t any processing from the Custom Form to the vRO configuration elements when the user requests a catalog item!. Getting the data directly from the vRO DB without any processing at request time is what is going to give us the fastest loading times. All the processing is done in the background, without none of the requesters noticing!
The last step is to refer to the getConfigurationElementValue action in our custom form
A small caveat – the way vRA 8 and the ‘Project’ field works is that even though the project shows the user the names to be chosen, it is actually processing the IDs, so in this case I added a hidden field called ProjectName which is what I will be actually using to convert the IDs to names (since the configuration elements are based on the name)
Mapping the Project IDs to namesUsing the getConfigurationElementValue action to get the values needed in the form
This is a small demo of how this works, take a look at the loading times for the form and changing the project! (And this is on a nested lab!)
Video Demo
Summary
To reiterate, the important things are:
No (or as little processing as possible if there is a field that cannot be used with configuration elements) should be done in the actions that are returning the data to the custom form
All the data should be processed in the background – The requester won’t be aware of it
Adding new projects it is as simple as duplicating one of the existing ones and changing the name. The way the workflows and actions are coded in this example will always look for every project (configuration element) below the ‘Projects’ folder
Getting the data out of the vRO DB directly via configuration elements instead of going to external sources, is the fastest way to get the values in the form.
Closing Note
I hope you found this interesting! It is using configuration elements in a way that might not be the most common usage, but it can bring great benefits to user experience when interacting with vRA requests. Having the data processed in the background and having really short form loading times will give the sensation of having more ‘speed’ to the tool itself!
Feel free to share this or leave a comment if you thought it was interesting!
On today’s post, we will go through creating an ABX Action to change the DNS Servers for a Deployment in vRA8. This might be needed to do in scenarios in which, even though the network has DNS servers configured, a specific deployment might require to use other DNS Servers while still being on the same network, for example, to join a different AD domain
The same idea can be used to edit other fields of the deployment, such as the IP Address, search domains, etc.
The post will be divided in 5 sections:
Cloud Template
Event Topic
ABX Action
ABX Subscription
Test
Cloud Template
In the template we’re going to need two inputs – dnsServers (comma separated list of DNS Servers) as well as an input to manage the amount of VMs in the deployment, we can call it ‘instances’
instances:
type: integer
title: Amount of VMs
default: 1
maximum: 5
dnsServers:
type: string
description: Comma separated list of DNS Servers
title: DNS Servers
default: '1.1.1.1,8.8.8.8'
These two values will be custom properties on the VM Object
In addition to this, the network assignment for the VM resource should be set to ‘static’. A customization specification profile is optional, since using a ‘static’ assignment will auto-generate a ‘ghost’ customization specification profile at the time of provisioning
The event topic that we need to use to make changes to the Network Configuration is the one that has the object that we need to edit being presented to the workflow in an editable state, as in, not read-only.
For this specific use, the state is Network Configure
Pay special attention to the ‘Fired once for a cluster of machines’ part
The dnsServers object is a 3D Array, so that is what we need to use in the ABX Action Code
So from this point we learn that:
The action will run once for a cluster of machines, so if we were to do a Multi-VM deployment we need to take this into account, otherwise, it will only run for a single VM and not all of the VMs in the deployment
a 3D array needs to be used to insert the DNS Servers into the object at the event topic
ABX Action
For this example, I will use Python, and I will not use any external libraries for array management (such as numpy) since I wanted to see if it could be done natively. Python has way better native support for lists than it does for arrays, but in this case, given the schema of the object in the event topic, we’re forced to use a 3D Array.
The first thing we need to do when creating the action, is to add the needed inputs. In this one, I will add the custom properties of the resources as an input
Adding the custom properties
Once we have that as an input, we can use it to get the data we need (amount of instances and DNS servers)
To pass data back to the provisioning state, we will use and return the ‘outputs’ object
This is the code of the action itself, I will explain it below
def handler(context, inputs):
outputs = {}
dnsServers = [[[]]]
instances = inputs["customProperties"]["count"]
inputDnsServers = inputs["customProperties"]["dnsServers"].split(",")
if (len(inputDnsServers) > 0):
outputs["dnsServers"] = dnsServers
outputs["dnsServers"][0][0] = inputDnsServers
for i in range(1,int(instances)):
outputs["dnsServers"] = outputs["dnsServers"] + [[inputDnsServers]]
return outputs
Define the outputs object
Define the 3D Array for DNS Servers
Assign the inputs as variables in the script
Convert the comma separated string into a List
If the list is not empty (this means that the user did enter a value in the DNS Servers list on the input form) then we add the 3D Array to the outputs object.
Why am I asking to see if it is empty? Because if the user did not put anything on the field, we will be overwriting the output with an empty array, and that will overwrite the DNS that were read from the network in vRA. We only want to overwrite that if we’re actually changing the DNS Servers.
Also in the same condition, we want to add the DNS Servers array to each VM, so we iterate through the amount of VMs.
The way to add it without using numpy (we have no append method) is not elegant, but it does the trick. Basically, we initialize the first element and then we add other elements to the same array using the same format.
Return the outputs object
This can also be done in javascript and powershell, the idea would be the same.
So how does this object look like in an actual run?
In this example, I changed the DNS for 3 VMs – You can see that we’re using the 3D Array Structure
Lastly, we need to configure a subscription for it to run at this specific state.
ABX Subscription
This is the most straightforward part – We create a blocking subscription in the Network Configure state, and we add the action we just created
The ABX subscription can be filtered via a condition (for example, to run only on specific cloud templates) as well.
So let’s do our deployment now!
The network i’m going to select has the 8.8.8.8 DNS configured
This will be overwritten by whatever we put on the input form. I’m going to use 1.2.3.4 and 5.6.7.8 for this example, and there will be 2 VMs in the deployment
We can check the output of the action before the deployment finishes
Action run started by an Event (Subscription)Action output
In there we can see the actual code that run, if it was successful or not, the payload, and the output the action had. In this case, our two DNS Servers for our two VMs with a successful output.
Checking the DNS for one of the VMs, we can see the two DNS Servers we selected as inputs!
Success!!!
Summary
I hope you found this post useful! The same idea can be used to change several other properties at provisioning time. Also, it was a learning experience for me to learn how to play with arrays in Python natively, and how to interact with ABX properly.
More posts will be coming soon!
If you liked this one, please share it and/or leave a comment!
On today’s post, I will focus on a Dynamic Multi-NIC configuration for a Cloud Template in vRA 8.x
This allows customers to reuse the same cloud templates for virtual machines that could have a different amount of NICs, and this amount is defined at the time of the request. If this wasn’t dynamic, then a cloud template with three networks, will always need to have three networks configured at the time of the request, which might not be the case.
Using a Dynamic construct allows for less cloud template sprawl, since multiple application configurations can use the same cloud template.
Since this configuration is not trivial, this post will be a step by step guide on how to achieve this result.
Current Environment
For this Lab demonstration, we will use a vSphere Cloud Account, 4 NSX-T segments that are part of a Network Profile with a capability tag named “env:production” – In doing so, when using that constraint tag in the cloud template, we can guarantee our deployment will use that specific network profile.
The 4 NSX-T segments also have a single tag that refers to the type of network it is. In this scenario, Application, Frontend, Database and Backup are our 4 networks.
NSX-T Segments tagged and defined in the network profile‘env:production’ tag in the network profile
Creating the Cloud Template
To get the Dynamic Multi-NIC configuration on the Cloud Template to work, we need the following things:
Inputs for Network to NIC mapping based on tagging
Inputs for NIC existence
Network Resources
VM Resource and Network Resource assignment
In addition to this, we can do customization in Service Broker to change the visibility of the fields. This is done to only allow the requester to choose a network mapping for a NIC what will actually be used.
Inputs for Network to NIC mapping based on tagging
This cloud template will allow for configurations of up to 4 NICs, and since we have 4 networks, we should let the requester select, for each NIC, what networks can be used.
We can see that each of the inputs allows for any of the networks to be selected.
Inputs for NIC Existence
Other than the first NIC (which should always exist, otherwise our VM(s) wouldn’t have any network connectivity, we want to be able to deploy VMs with 1, 2, 3, and 4 NICs, using the same Cloud Template.
To achieve that, we will create 3 Boolean Inputs that will define if a NIC should be added or not.
To manage the configuration of the NICs and networks, the network resources for NICs 2, 3 and 4 will use a count property, and this property’s result (either 0 if it doesn’t exist, or 1 if it does exist) will be based on the result of the inputs. Network 1 will not use that property
Also, we will use the deviceIndex property to maintain consistency with the numbering – So the network resources look like this
The constraint tags that are used are the Network Input (to choose a network) and the ‘env:production’ tag to make our deployment use the Network Profile we defined earlier.
VM Resource & Network Resource Assignment
This is the tricky part – Since our networks could be non-existent (if the needNic input is not selected) we cannot use the regular syntax to add a network, which would be something like:
This will fail on the Cloud Template validation because the count for Network2 could be zero, so to do the resource assignment, we need to use the map_by syntax.
This allows for any combination of NICs, from 1 to 4, and if the count of one of the resources is 0, it won’t be picked up by the assignment expression.
This is what the Cloud Template looks on the canvas. You can see that Networks 2, 3 and 4 have the appearance of possible multiple instances. This is because we’re using the count parameter.
Canvas view of the Cloud Template
If we were to deploy this Cloud Template, it will look like this:
Doesn’t make much sense to select networks that we won’t assign, right?
How do we fix this? We can leverage Service Broker to manage the visibility of the fields based on the boolean input!
Using the inputs as conditional value for the visibility of the network field
So now, from Service Broker, it looks like this:
No extra NICs selectedNICs 2 and 3 selected
So if we deploy this, it should have three networks assigned. The first NIC should use the Application Network, the second NIC should use the Frontend Network and the 3rd NIC should use the Database Network.
Let’s test it!
TA-DA!
We can see that even if the Cloud Template had 4 Network Resources, only 3 were instantiated for this deployment! And each network was mapped to a specific NSX-T segment, thanks to the constraint tags.
Closing Note
I hope this blog post was useful – The same assignment method can be used for other resources such as Disks or Volumes – the principle is still the same.
Feel free to share this if you found it useful, and leave your feedback in the comments.
On today’s post, we will go through the process of updating an onboarded deployment in vRA 8.x
The onboarding feature allows customers to add VMs that were not deployed from vRA, to the vRA infrastructure. This means that these VMs are added to one or more deployments, and once they exist within vRA, operations such as power cycling, opening a remote console, or resizing CPU/RAM are now available.
However, there are scenarios in which customers would want to expand these deployments, not with new onboarded VMs, but with newly deployed VMs (or other resources) from vRA! These deployments will use an image, a flavor, could use a multitude of inputs, tagging, networks, etc. So how do we do this?
Onboarding the VMs using an auto-generated Cloud Assembly Template
The first thing we need to do, is to create an onboarding plan, select a name for our deployment, and select the VMs we’re going to onboard initially.
Creating the Onboarding PlanAdding two VMs to be onboarded
On the deployments tab, we can rename the deployment if needed, but the most important part is to select Cloud Template Configuration and change it to Create Cloud Template in Cloud Assembly Format– this will allow us to have a source for our deployment, that we can edit afterwards to allow for future growth
Cloud Template in Cloud Assembly format
It is important to note that the imageRef has no image available. Since this is not a vRA Deployment but an Onboarding, none of the resources are being deployed from any of the images. We will come back to this item later.
After saving this configuration and clicking on Run, our deployment will be onboarded
Updating the onboarded deployment to add a new VM in a specific network
If we check on the onboarded deployment, we will see that it is mapped to a specific Cloud Template (the one that was auto-generated earlier by the Onboarding Plan)
So if we were to do an update on this deployment, we need to edit that Cloud Template
I will now add a vSphere Machine resource as well as a vSphere Network:
This is what our template looks like now. So the next thing we should do is click on Update, right?
Update is Greyed out!
The update task is greyed out because ir Cloud Template does not have inputs. Since we don’t have inputs, what we need to do is to go to the Cloud Template, and instead of selecting Create a New Deployment we should select Update an Existing Deployment and then click on the onboarded deployment.
Updating the Onboarded Deployment
After clicking on Next, the plan is presented.
Notice something wrong here?
The update operation will attempt to re-create the onboarded VMs! That’s not something we want, and also, in this scenario, it will fail since there is no image mapping to deploy from!
What we want is to leave all the VMs that were previously onboarded, untouched, and only add our new VM and network. So how do we achieve this?
This is achieved by adding the ignorechanges parameter with a value of true to every resource in the cloud template that was previously onboarded – In this scenario, this would be our 2 DevTools VMs
Adding the ignoreChanges parameter
If we re-try updating the deployment now, the only tasks that should appear will be the ones for the new resources (VM and Network)
Update deployment showing the new tasks
After clicking on ‘deploy’ and waiting for it to finish, our deployment will now like this
Deployment updated with our new VM and network! Hooray!
Offboarding/Unregistering limitations
It is important to note that vRA’s limitations for unregistering VMs are still present. The only VMs that can be unregistered from vRA are the ones that were previously onboarded. VMs that were deployed from vRA will not be able to be unregistered without deletion. The fact that the deployment VMs are part of an Onboarded Deployment does not change this.
Closing Note
I hope you enjoyed this post! When I started working on this use case I figured it was not as trivial as I thought, and after doing research and testing, found this walkthrough/solution.
On today’s post, as a continuation of the previous post (in which we talk about the VCF MGMT Domain) I will show a step by step guide of how to do a complete deployment of a VCF Workload Domain, subject to some specific constraints based on a project I was working on, using VCF’s API!
What’s this non-standard architecture like?
In this specific environment, I had to play around the following constraints
4 hosts with 256GB of RAM using vSAN, check the previous post for information about the MGMT domain!
3 Hosts with 256GB of RAM, using vSAN
3 Hosts with 1.5TB of RAM, using FC SAN storage
Hosts using 4×10 NICs
NIC Numbering not being consistent (some hosts had 0,1,2,3 – other hosts had 4,5,6,7 – even though this can be changed editing files on the ESXi, it is still a constraint and can be worked around using the API)
With this information, the decision was to:
Separate the Workload Domain into 2 clusters, one for NSX-T Edges and the other one for Compute workloads, given the discrepancies in RAM and storage configuration, they could never be part of the same logical cluster.
This looks something like…
It is impossible to deploy this using the GUI, due to the following:
Can’t utilize 4 Physical NICs for a Workload Domain
Can’t change NIC numbering or NIC to DVS uplink mapping
So we have to do this deployment using the API! Let’s go!
Once we have the token, we can use it in other API calls until it expires and we just either refresh it or create a new one. All the VCF API calls that are generated to SDDC manager (not internal API calls) will require the usage of a bearer token.
List of steps to create a workload domain
Commission all hosts from SDDC manager and create network profiles appropriately to match the external storage selection – In this scenario, we will have a network profile for the vSAN based hosts, as well as another network profile for the FC SAN based hosts. Hosts can also be commissioned via API calls (3.65 in the API reference) instead of doing it via the GUI, but the constraints I had did not prevent me from doing it via GUI.
Get all the IDs for the commisioned hosts – The API Call is “2.7.2 Get the Hosts” and it is a GET call to https://sddc_manager_url/v1/hosts using Bearer Token authentication
Create the Workload Domain with a single cluster (Compute) – The API Call is “2.9.1 Create a Domain”
Add the Secondary Cluster (Edge) to the newly-created workload domain – The API Call is “2.10.1 Create a Cluster”
Create the NSX-T Edge Cluster on top of the Edge Cluster – The API Call is “2.37.3 – Create Edge Cluster”
For each of these tasks, we should first validate our JSON body before executing the API call. We will discuss this further.
You might ask, why don’t you create a Workload Domain with two clusters instead of first creating the Workload Domain with a single cluster and then adding the second one?
This is something I hit during the implementation – If we check the Clusters object on the API, we can see it is an array, so it should be able to work with multiple cluster values.
"computeSpec": { "clusterSpecs": [
The info on the API call also points to the fact that we should be able to create multiple clusters on the “Create Domain” call.
Even worse, the validation API will validate an API call with multiple clusters
However, I came to learn (after trying multiples times and contacting the VCF Engineering team, that this is not the case)
For example, if our body looked something like this (with two clusters), the validation API will work!
However, when we go ahead and try to create it, it will fail, and we will see the following error on the logs.
ERROR [vcf_dm,02a04e83325703b0,7dc4] [c.v.v.v.c.v1.DomainController,http-nio-127.0.0.1-7200-exec-6] Failed to create domain com.vmware.evo.sddc.common.services.error.SddcManagerServicesIsException: Found multiple clusters for add vi domain. at com.vmware.evo.sddc.common.services.adapters.workflow.options.WorkflowOptionsAdapterImpl.getWorkflowOptionsForAddDomainWithNsxt(WorkflowOptionsAdapterImpl.java:1222)
So, as mentioned earlier, we need to first create our domain (with a single cluster), and then add the 2nd cluster!
1: Create a Workload Domain with a Single Cluster
We will first create our Workload Domain with the compute cluster, which in this scenario, uses external storage, and will use the secondary distributed switch for overlay traffic.
This is my API call body based on the API reference, to create a Workload Domain with a single cluster of 3 hosts, using two VDS, 4 physical NICs numbered from 0 to 3 and external FC storage, using the host IDs that I got after the previous step.
The DVS that is going to be used for overlay traffic must have the isUsedByNsxt flag set to true. In the case of a 4 NIC and 2 VDS deployment such as this one, it shouldn’t have any of the management, vMotion or vSAN traffic.
With the body, to execute the VALIDATE and EXECUTE api calls, we will do the following: (high level overview since we can use any REST API tool such as Postman, curl, invoke-restmethod, or any wrapper from any language that can execute REST calls)
The list of steps will be the same for all the POST API calls, changing the URL to match each specific call.
We should continue editing and retrying in case of errors until we get the validation to pass, do not attempt to execute the API call without validating it first!
The deployment will start and after a couple minutes we will see in the SDDC console that it was successful.
If it were to fail for whatever reason, we can troubleshoot the deployment by checking where it failed on the SDDC console as well as checking logs, but as long as the validation passes, it should not be a problem with the body we’re sending.
2: Adding a 2nd Cluster to the existing workload domain
To add a cluster to an existing domain, the first thing we need is to get the ID of the domain, that can easily be done with a GET call to https://sddc_manager_url/v1/domainsand selecting the ID of the workload domain we just created.
Once we get the ID, this is the body (following the API reference) to add a new cluster to an existing domain.
Even though we don’t need the cluster to be prepared for NSX-T (since it will only be used for Edges) setting the isUsedByNSXT flag to true will make the secondary VDS used by the uplink portgroups once we create a T0, which is what we want in this scenario – otherwise, we would not be using the 3rd and 4th NICs at all.
As discussed earlier, we should first run the POST call to validate in this case, the URL is https://sddc_manager_fqdn/v1/clusters/validations and after the body is validated, proceed with the creation removing validation from the URL
Last but not least, we need to create our NSX-T Edge Cluster on top of the 2nd cluster on the domain!
3: Create NSX-T Edge Cluster
The last piece of the puzzle is creating the NSX-T Edge Cluster, to allow for this workload domain to leverage overlay networks and communicate to the physical world.
To create the NSX-T Edge Cluster, we first need to get the Cluster ID of the cluster we just created (how many times can you say cluster in the same sentence?)
Now that we have the ID, this is the body to create two Edge Nodes, configure management, TEP and uplink interfaces, configure a T0 and a T1 instance, as well as configuring BGP peering on the T0 instance!
As mentioned before, please run the VALIDATE call first, in this scenario, a POST call to https://sddc_manager_fqdn/v1/edge-clusters/validations – after validation is passed, proceed to execute the call without the validations on the URL.
After this procedure is finished, we will have our workload domain with two clusters as well as a T0 gateway completely configured and ready to go! Simple and quick, isn’t it?
Closing Note
Leveraging APIs for VCF can help us not only to work with architectures or designs that are not able to be implemented due to GUI restrictions, but also greatly speed up the time we take in doing so!
I hope you enjoyed this post, and if you have any concerns, or want to share your experience deploying VCF via API calls, feel free to do so!
Hello Everyone! It’s me again, trying to maintain a weekly post cadence!
Today I’m going to talk about some roadblocks I hit while doing a 4.2 VCF Deployment in a real, customer environment. Hopefully this will prevent these issues from happening to you or help you to solve them quickly if they do arise!
Password Policy for Cloud Builder
In VCF 4.2, several changes to password strength were made. It seems that using 8 character passwords are hit/miss (you could get a valid deployment and then immediately a non-valid deployment if you deploy another Cloud Builder with a password like “VMw@r3!!” – I haven’t been able to fully grasp the cause for this behaviour.
In addition, VMware is now a dictionary word, so it wont be allowed. So “VMware1!” and “VMware1!VMware1!” will also fail.
The password that i’ve been using successfully for the initial deployment is “VMw@r3!!VMw@r3!!” – That one works 100% – You can go ahead and use that one.
Hostnames in uppercase
This one is really, really strange – If the hostnames of your ESXi hosts are in uppercase, you will get a ‘Failed to connect to lowercase_hostname’ for all of your hosts when running the validation, and the validation will stop and won’t query any of the host configuration
I spent some time trying to figure this out, at first I thought it was DNS records, but then on a different environment, 3 of the 4 hosts had their hostname in upper case and one of them in lower case, and the one in lower case was the only one connecting, so that made me test the change and suddenly the new host in lowercase was also connecting!
To clarify, ESXI1.VSPHERE.LOCAL will fail, esxi1.vsphere.local will work – Make sure your hostnames are in lowercase
Heterogeneous / Unbalanced disk configuration across hosts
This one is really interesting, let’s say you’re doing an all flash VCF and you have 20 disks per host – The best way to configure it would be 4 Disk groups of 1 Cache + 4 Capacity, so that you would use all 20 disks.
Since you can have at a maximum 5 Disk groups of 1 Cache + 7 Capacity, 40 is the maximum number of disks you can have.
However, make sure that you’re following these two rules for your deployment
Make sure that the amount of disks follows a multiple of a homogeneous disk group configuration so that all your disks can be used and all the disk groups have the same amount of disks – I.e, if you have 22 disks, there is no way you can use all disks while maintaining all disk groups with the same amount of disks. If you have 22 disks, you can do 3 (1+6) and one won’t be used, or 4(1+4) and two won’t be used.
Make sure that all your hosts have the same amount of disks. You can check this before installing – In my scenario, validation was passing but it was setting the cluster as hybrid instead of all flash. After checking that all devices were SSD and were marked as SSD I was really confused. Then I checked and two of the hosts had 2 more disks than the rest. Fixing that made the validation pass and marking the cluster as all flash.
EVC Mode
This one almost made me reinstall the whole cluster…
BE REALLY SURE that you’re selecting the correct EVC mode for your CPU family if you’re selecting an EVC mode in the Cloud Builder spreadsheet.
If you select the wrong EVC mode, Cloud Builder will fail in this deployment, and you won’t be able to continue from the GUI at all. The only way around it is via the API. Otherwise, it is wiping the cluster and starting from scratch!
I’m going to show you how to fix this issue but the method applies in case you need to edit the configuration and then re-attempt a deployment.
First of all, you need to get your SDDC Deployment ID, you can get it with this API call (I will be using curl for this example but you can also use something like invoke-restmethod in powershell or even a GUI based REST client such as Postman)
You can export the output to a file or to a text viewing tool such as less, and then search for the sddcId value
Editing the JSON File
Once you have the sddcId, you need to edit the JSON file that CB generated from the spreadsheet so you can then use it in the API call. I recommend that you copy the file and edit the copy. The file is located at /opt/vmware/sddc-support/cloud_admin_tools/resources/vcf-public-ems/
#COPY THE FILE
cp /opt/vmware/sddc-support/cloud_admin_tools/Resources/vcf-public-e ms/vcf-public-ems.json /tmp/newjson.json
#REPLACE STRING ON FILE
sed -i "s/cascadelake/haswell/g" /tmp/newjson.json
You can also edit the file using vi – in this case I used sed because I knew the string will only appear once in the file and it was faster
Restarting the deployment
Now that you have the sddcId and you’ve edited the JSON file, it is time for you to restart the process using another API call
And if you log in to the Cloud Builder web interface, your deployment should be running again! Phew, you saved yourself from reinstalling and preparing 4 nodes! Go grab a beer while the deployment continues 😀
Driver Issue when installing NSX-T VIBs
I ran into this issue after waiting for multiple hours for the NSX-T Host Preparation to finish, and seeing all the hosts on the NSX-T tab being marked as failed.
When checking the debug logs for Cloud Builder, I saw errors like:
2021-04-07T23:06:44.700+0000 [bringup,196c7022580bfc32,5a84] DEBUG [c.v.v.c.f.p.n.p.a.ConfigureNsxtTransportNodeAction,bringup-exec-7] TransportNode esxi1.vsphere.local DeploymentState state is {"details":[{"failureCode":260
80,"failureMessage":"Failed to install software on host. Failed to install software on host. esxi1.vsphere.local : java.rmi.RemoteException: [DependencyError] VIB QLC_bootbank_qedi_2.19.9.0-1OEM.700.1.0.15843807 requires qe
dentv_ver \u003d X.40.17.0, but the requirement cannot be satisfied within the ImageProfile. VIB QLC_bootbank_qedf_2.2.8.0-1OEM.700.1.0.15843807 requires qedentv_ver \u003d X.40.17.0, but the requirement cannot be satisfied within the Im
ageProfile. Please refer to the log file for more details.","state":"failed","subSystemId":"eeaefa1e-c5a2-4a8a-9623-994b94a803a9","__dynamicStructureFields":{"fields":{},"name":"struct"}}],"state":"failed","__dynamicStructureFields":{"fi
elds":{},"name":"struct"}}
This is related to QLogic drivers that are included in the HP custom image that was being used in this deployment (and was patched to 7.0u1d which is the pre-requisite for VCF 4.2)
None of these drivers were in use, and none of the hosts were using QLogic hardware – So these drivers could be removed without issues, however, it is best to unconfigure the hosts from NSX-T first since that also prompts for a reboot.
Go to the Transport Node tab in NSX-T, select the cluster, and click on “Unprepare” – This will likely fail and prompt you to run a force cleanup – This one will work and the hosts will disappear from the tab.
In my scenario, none of the NSX-T VIBs were installed so no NSX-T VIB cleanup was necessary
Now, it is time to delete the drivers from the hosts and reboot them. You can run this one by one on the hosts (since you already have vCenter, vCLS, and NSX Manager VMs running, you can’t just blindly power-off all your hosts)
Edge TEP to ESXi TEP validation when using Static IP Pool
VCF 4.2 removes the need of having a DHCP server on the ESXi TEP network (as long as you’re not using stretched cluster) which is a lifesaver for many, since setting up the DHCP server was usually a light stopper for customers (the other one being BGP)
However, the validation still attempts to search for a DHCP server (it doesn’t matter that you configured a Static IP Pool on the spreadsheet) and since there isn’t any, you get a 169.254.x.x IP and the validation fails. For example:
VM Kernel ping from IP '172.22.17.2' ('NSXT_EDGE_TEP') from host 'esxi1.vsphere.local' to IP '169.254.31.119' ('NSXT_HOST_OVERLAY') on host 'esxi2.vsphere.local' failed
You can see the IP is on the 169.254.x.x range
Luckily, this is just a validation bug, it is reported internally, and will likely be fixed in the latest VCF release. The issue will not present itself while actually doing the deployment and the TEP addresses will be set up correctly using the static IP Pool
BGP Route Distribution Failure
If your BGP neighboring is not configured correctly on your upstream routers, you will see the task “Verify BGP Route Distribution fail”
021-04-08T05:09:54.729+0000 [bringup,42ba3b72e2ee4185,395f] ERROR [c.v.v.c.f.p.n.p.a.VerifyBgpRouteDistributionNsxApiAction,pool-3-thread-13] FAILED_TO_VALIDATE_BGP_ROUTE_DISTRIBUTION
com.vmware.evo.sddc.orchestrator.exceptions.OrchTaskException: Failed to validate the BGP Route Distribution result for edge node with ID 123b3404-bab6-4013-a9f7-eba3b91b4faf
This means that the BGP configuration on the upstream routers is incorrect, usually, there is a BGP neighbor missing. The easiest way to figure out what’s missing is to check the BGP status on the Edge Nodes
In my case, the Upstream switches only had one neighbor configured per uplink VLAN, so node 1 showed:
BGP neighbor is 172.22.15.1, remote AS 65211, local AS 65210, external link
BGP version 4, remote router ID 172.22.15.1, local router ID 172.22.16.2
BGP state = Established, up for 09:09:51
And node 2 Showed:
BGP neighbor is 172.22.15.1, remote AS 65211, local AS 65210, external link
BGP version 4, remote router ID 0.0.0.0, local router ID 172.22.15.3
BGP state = Connect
You can see that the BGP session for node 2 is not established. After configuring the neighbor correctly on the upstream routers, the issue was resolved!
Conclusion
Deploying VCF 4.2 in this environment has been a rollercoaster but luckily, all the issues were able to be solved.
I hope this helps you either avoid all of these issues (by pre-emptively checking and fixing what could go wrong) or in case it does happen to you, to fix them as quick as possible)
Stay tuned for more VCF 4.2 adventures, next time, with workload domains!
After a long hiatus, I decided to write a new blog post (and hopefully improve the frequency of them :D) – This will be based on a 2-hour presentation that I did for VMUG (VMware User Group) Argentina last week, which was done in spanish, and I will link it down below
However, for all of the non-spanish Speakers, I will do a breakdown of everything you need to check before attempting a vSphere upgrade from the vCenter & PSC perspective to pass the upgrade wth flying colors! – Buckle up!
Where is our environment currently standing?
First of all, you need to assess the current situation of your vCenters and PSCs – Is replication working correctly for example? This article goes really really deep into checking that:
If you have any replication issues, this is the first thing you need to fix, otherwise, as shown in the previous article (and the video) you risk completely destroying your environment.
The 2nd thing you need to check is your current topology – How many PSCs and vCenters are actually in my environment? Am I using PSC HA? Is everything converged? Depending on your current topology, it might be a pretty trivial migration or it would need multiple steps over the course of a weekend.
What happens in the upgrade process?
First of all, the external PSC is deprecated in vSphere 7.0 – That means that, as a part of the upgrade process, any environment with an external PSC is converged. Even though this process might be straightforward, it can cause multiple problems before, during and after the migration. It’s easier and more convenient to break it up in parts
So if we’re good with replication (check and re-check previous article, I can’t stress this enough) then we need to figure out an upgrade and migration plan
Planning the upgrade process based on our topology
Let’s start with something simple:
What would be the correct steps here?
Let’s break it down:
1: Offline snapshots of all three VCs (with embedded PSCs) – offline means with all the SSO domain powered off- this is done from the ESXi nodes that are hosting the VMs.
2: Upgrade vCenter 1
3: Check functionality and replication
4: Offline snapshots of all three VCs (with embedded PSCs)
5: Upgrade vCenter 2
6: Check Functionality and replication
7: Guess what?
8: Upgrade vCenter 3:
9: Check Functionality and replication
10: Delete all snapshots
Why am I taking snapshots at every step? Why don’t I just take a single round of snapshots and then upgrade all at once?
Well, because if you had any issue at any point of the 2nd or 3rd upgrade, you would have to roll back everything and start from scratch. If you do it this way, you have multiple points to go back and avoid having to re-do the upgrade process! This can get even worse if instead of 3 vCenters you have 9 or 10 – If let’s say, you had an issue with upgrade 7, you would have to revert everything!
Now let’s make this a little bit more complicated!
So let’s picture this scenario (which is not too uncommon, i’ve seen this is in the real world)
What do we have?
First of all, blue lines symbolize good replication and red lines symbolize that replication is not working – So, as discussed earlier, this will be the first thing to fix – in the process of fixing this (most likely with a GSS ticket), multiple rounds of offline snapshots will be taken!
Now, onto the topology:
6 External PSCs in a ring topology
3 PSC HA VIPs being used by 2 vCenters each
6 vCenters
So what should we do here? This not only involves the upgrade of the vSphere environment, but also, the re-pointing of 2nd and 3rd party tools to the new converged PSCs – Think of NSX and SRM for example.
The biggest pain point in this scenario, however, is PSC HA – how do we get rid of this prior to the upgrade?
Even though there is a KB for converging PSC HA (https://kb.vmware.com/s/article/65129) in practice, this is not the best approach due to how error prone it is.
What is the best approach? There are two ways to approach this, depending on downtime and operations.
The cleanest approach, would be to deploy 6 new PSCs, then repoint the vCenters to those 6 PSCs, and then decomission all the PSC HA nodes (as well as the VIP) – However, this might be complicated because of lack of IP addresses in the management segment, time, etc.
You could also leverage lsdoctor (https://kb.vmware.com/s/article/80469) to unconfigure PSC HA and then repoint the vCenters to each of the nodes – This introduces a little bit more downtime per vCenter (downtime when unconfiguring PSC HA + downtime until the repoint is complete) but removes the need of deploying new PSCs.
If you ask me, I recommend the first option, to make this as clean as possible.
So in this scenario, what would you do?
Offline snapshots of all vCenters and PSCs
Deploy PSC 7 pointed to PSC 6
Deploy PSC N pointed to PSC N-1 until all PSCs are deployed.
Check replication among the new PSCs
So now we have something like this
You can see that by deploying the PSCs in that order, we have a “semi-ring” already, with way less operational hassle than if we were deploying them pointed to a single PSC and then having to remake the replication agreements
So what’s next?
We need to repoint the vCenters to these new PSCs – Since the repoint is a pretty short process, you can get away with taking a single round of offline snapshots at the beginning and just repoint everything
Offline snapshots of all vCenters and PSCs
Repoint all vCenters to the new PSCs, 1:1
Check correct functioning
End result:
Lovely, right?
Now, we need to get rid of all the PSCs that were forming the PSC HA (nodes and VIPs)
So we did all this and we haven’t even started upgrading or converging… but believe me, taking due diligence in doing this as clean as possible will save you from multiple headaches when you actually upgrade!
So what is left?
Form a ring creating an agreement between PSC12 and PSC7
Take a new round of offline snapshots
Converge PSC7
Check correct functioning
Take a new round of offline snapshots
Converge PSC8
….
….
Until all PSCs are converged
In case there is any issue with the convergence, you can just go back to the latest functioning snapshot so you don’t have to redo everything!
You should be here now:
And from here, you can finally do the upgrade process – as discussed previously and in the first scenario, you should take a round of offline snapshots per each upgrade, to avoid having to re-do upgrades
Last but not least, you should repoint all 2nd and 3rd party solutions to the new converged (and upgraded) PSCs that are now living inside the vCenter appliance!
Closing note
I hope you enjoyed this post – If you have even limited knowledge of spanish, I encourage you to watch the youtube video in which I go over this in detail, and also I analyze and fix replication issues the same way it would be done if you contacted GSS.
Feel free to share this with peers, customers, partners – If we generate awareness about these processes and a clean and correct way of doing them, we will have way more succesful upgrades!
Recently, we’ve been working on a global issue affecting all customers that had deployed a vCenter Server as version 6.5 Update 2 or later. The Security Token Service (STS) signing certificate may have a two-year validity period. Depending on when vCenter was deployed, this may be approaching expiry.
Since currently there is no alert on vCenter for this certificate, and also it is a certificate that prior to 6.7u3g had no way to be replaced by customers in case of expiration (required GSS involvement to execute internal procedures / scripts) and it generates a production down scenario, silently.
Within the GSS team, we’ve come up with three scripts to help with this situation.
Checksts.py
Checksts.py is a python script that is mentioned in KB https://kb.vmware.com/s/article/79248. This script will proactively check for expiration of the STS certificate. It works on Windows vCenters as well as vCenter Server Appliances.
To use it, you can download it from the KB mentioned:
Once it is downloaded, you can copy it to any directory on your vCenter. After that, you will run it like this:
Windows: "%VMWARE_PYTHON_BIN%" checksts.py
VCSA: python checksts.py
This is an example for VCSA:
If you get the message “You have expired STS certificates” and/or your certificate expiration date is in less than 6 months, we recommend to move onto the next step, replacing the STS certificate! If your expiration date is in more than 6 months, then you don’t have to worry about any of this!
The idea is the same for both, replacing the STS certificate with a new, valid one. This can be done proactively (cert has not expired yet) as well as reactively (cert has already expired and you’re in a production down scenario)
The steps for these two KBs are mentioned in the articles. They’re pretty much identical, with minor differences in running the commands due to the Guest OS, and super straightforward to run.
Once the STS is replaced, in case it was done proactively, you will be good to go!
YOU CAN STOP READING FROM THIS POINT ON – hope you liked this blog entry!
However, if this was done reactively, then it is likely that you will need to replace more certificates in your vCenter Server, especially if you were using VMCA certs (which could have the same expiration date as the STS certificate if they were never replaced)
Replacing other certificates
How do I know if which of my other certificates are expired?
On the KBs mentioned, there are two one-liners provided to check for certificates
Windows: $VCInstallHome = [System.Environment]::ExpandEnvironmentVariables("%VMWARE_CIS_HOME%");foreach ($STORE in & "$VCInstallHome\vmafdd\vecs-cli" store list){Write-host STORE: $STORE;& "$VCInstallHome\vmafdd\vecs-cli" entry list --store $STORE --text | findstr /C:"Alias" /C:"Not After"}
VCSA:for i in $(/usr/lib/vmware-vmafd/bin/vecs-cli store list); do echo STORE $i; /usr/lib/vmware-vmafd/bin/vecs-cli entry list --store $i --text | egrep "Alias|Not After"; done
These commands will show, for each of the VECS (VMware Endpoint Certificate Store) stores, the expiration date for all certificates. If the certificates have an expiration date prior to today, then they’re expired. Also, you will have issues with services if certificates are expired. Services such as vpxd-svcs, vpxd or vapi-endpoint will be pretty verbose with expiration date of certain certificates.
For example:
root@vcsa1 [ /tmp ]# for i in $(/usr/lib/vmware-vmafd/bin/vecs-cli store list); do echo STORE $i; /usr/lib/vmware-vmafd/bin/vecs-cli entry list --store $i --text | egrep "Alias|Not After"; done STORE MACHINE_SSL_CERT Alias : __MACHINE_CERT Not After : Apr 6 11:57:19 2029 GMT STORE TRUSTED_ROOTS Alias : c96d3301505316ccc1b295276ece31318ad79ec7 Not After : Apr 6 11:57:19 2029 GMT Alias : 8a11418d5ae2b87b7e8a5cb8646fbfae41503f9d Not After : Dec 13 21:50:49 2029 GMT Alias : cb5a495d34f3f2f75d357b47aac3799346665258 Not After : Sep 25 20:32:57 2022 GMT Alias : 229a64a3dff7417d0b38fb011c692a55b7bee5c2 Not After : May 16 20:21:12 2030 GMT Alias : 2f0e8e4f1658e61bef5004cb5efd159b90396838 Not After : May 16 20:45:07 2030 GMT STORE TRUSTED_ROOT_CRLS Alias : 4504400e4bcbdab5a34a9bc2555abd55327369c1 Alias : 31b2b5a18d89d90dadff901400a60d45ca3356e9 Alias : e7840a7cbbe7fcdd7a13d9159ff97443cc53fb5e Alias : 985d7e55183635f13e2c6469eee9c72f68334615 STORE machine Alias : machine Not After : Apr 6 11:57:19 2029 GMT STORE vsphere-webclient Alias : vsphere-webclient Not After : Apr 6 11:57:19 2029 GMT STORE vpxd Alias : vpxd Not After : Apr 6 11:57:19 2029 GMT STORE vpxd-extension Alias : vpxd-extension Not After : Apr 6 11:57:19 2029 GMT STORE APPLMGMT_PASSWORD STORE data-encipherment Alias : data-encipherment Not After : Apr 6 11:57:19 2029 GMT STORE SMS Alias : sms_self_signed Not After : Apr 12 12:04:48 2029 GMT STORE BACKUP_STORE
In this case, none of the certificates are expired. But if we had expired certificates we will need to replace them!
Let’s group them in three groups. All of them are replaced using the same tool, certificate-manager, detailed on KB https://kb.vmware.com/s/article/2097936, but the option you will use will depend on the scenario
Group 1: Machine SSLCertificate (Front facing certificate, on port 443)
If only Machine SSL is expired, you will run Option 3 (Replace the Machine SSL certificate with a VMCA Generated Certificate) of this KB, with the following caveats
The “comma separated list of hostnames” you will be prompt to complete, should contain the PNID of the node as well as any additional hostname or alias you might be using. How do we get the PNID for the node?
The value of “VMCA Name” should match the PNID obtained in the prior step
Group 2: Root certificate (VMCA root certificate)
If there is any certificate expired in the TRUSTED_ROOTS store, it will be safer to just run Option 8 (Reset all certificates)on the KB mentioned above. This will reset all certificates to VMCA signed. The same caveats mentioned for Option 3 apply
Group 3: Solution Userscertificates(vpxd, vpxd-extension, machine, vsphere-webclient)
If there is any certificate expired in the stores vpxd, vpxd-extension, machine or vsphere-webclient, run Option 6 (Replace Solution User Certificates with VMCA generated Certificates)on the KB mentioned above. The same caveats mentioned for Option 3 apply
Once all this is done, you should be back up and running with regenerated certificates, and out of the production down scenario!
Closing note
This is a pretty concerning issue, so I’m really happy to have been part of the team to help fix so many environments across the globe.
Please, use this information to proactively check for the STS certificate, as well as replacing without having to get into a production down scenario. You can share this with customers, partners, or whoever you feel might be benefited from this information!
Lucho's Blog! 