Host Commissioning — Build Guide
Before an ESX host can go into a new workload domain, a new cluster or an existing cluster, it has to be commissioned: added to the VCF instance’s global inventory as a free host. TechDocs: “Adding ESX hosts to the global inventory is called commissioning. You must commission ESX hosts before you can use them to create a new workload domain or add them to an existing VCF domain.”
This page covers that one step end to end: preparing the host, commissioning
it from the vSphere Client, through the SDDC Manager API or with a script,
checking the result, decommissioning and recommissioning, and the failures you
are most likely to hit. The runbooks that consume commissioned hosts point
here:
23-workload-domain-creation.md,
24-cluster-creation.md,
25-cluster-expansion.md and
22-stretch-execution.md.
Hosts for the management domain at bring-up are not commissioned: the VCF
Installer takes them straight from the deployment spec (see
35-bringup-json.md). Hosts for clusters on NFS 4.1,
iSCSI, FCoE or NVMe over Fabrics can’t be commissioned at all; see
§1.3.
Evidence level: TechDocs 9.1 (Commission ESX Hosts, Decommission ESX Hosts, Network Pool Management, all last updated 2026-11-09), the SDDC Manager API 9.1.1 reference, the configmax tool and Broadcom KBs, read verbatim on 2026-11-10. The
SDDCAdminsfix (KB 415254) is field-verified 2026-10-22;administrator@vsphere.localas the break-glass account (§1.5) is field-verified 2026-11-10. The Commission Hosts wizard in the vSphere Client (§2: the pages and fields of Host Addition and Validation) is lab-verified 2026-11-10. Everything else on this page is not lab- or field-verified in this repo.
The sequence, end to end
network pool → prepare each host → commission → check: Unassigned Hosts
(once) (DNS, cert, disks) (UI, API or / UNASSIGNED_USEABLE
script)
- Prepare: network pool, host, storage type, access
- Commission in the vSphere Client
- Commission through the SDDC Manager API
- Scripted: VCFHostPreparation
- Check the result
- Decommission and recommission
- Common failures
1. Prepare
1.1 The network pool comes first
TechDocs lists it as a host prerequisite: “A network pool must be created and available before host commissioning.” and “The network pool must support the storage type you select for a host (vSAN, vSAN ESA, vSAN Storage Client, NFS, VMFS on FC, vVols).” Every host is bound to one pool when you commission it, and the pool hands out the vMotion and storage VMkernel addresses: “you only need to enter an IP address manually for the vmk0 management interface. Network pools automatically assign static IP addresses to vSAN, NFS, iSCSI, and vMotion VMkernel ports.”
Which networks the pool needs, from Network Pool Management:
| Storage being used | Required networks in the pool |
|---|---|
| vSAN | vMotion and vSAN |
| vSAN compute | vMotion and vSAN |
| vSAN storage (formerly vSAN Max) | vMotion, vSAN, vSAN Storage Client |
| NFS | vMotion and NFS |
| vSAN and NFS | vMotion, vSAN, and NFS |
| VMFS on FC | vMotion only, or vMotion and NFS |
| vVols on FC / iSCSI / NFS (deprecated) | vMotion only or vMotion and NFS / vMotion and iSCSI / vMotion and NFS |
- Size each range for every host that will use the pool, plus growth. The default pool created at bring-up belongs to the management domain; TechDocs: “To create a workload domain with hosts in a different layer-2 network than the management domain, you must create a new network pool. Also, if you want to use external NFS or VMFS on FC storage, you must create a new network pool.”
- In 9.1 the storage and vMotion networks can be IPv6: “You cannot mix IP versions for the same traffic type within a cluster.”
- Where the pool’s networks come from (VLAN, MTU, gateway, range per network):
intake
H8and23-workload-domain-creation.md§1. For a stretched cluster’s second AZ:22-stretch-execution.md§2. - To see existing pools and their ranges: vSphere Client → Global Inventory
Lists → Hosts → Network Pools, or
GET /v1/network-pools.
1.2 Prepare each host
Install a supported ESX build, then work through Broadcom’s Preparing ESX Hosts for VCF or vSphere Foundation. The Commission ESX Hosts page then checks these criteria, quoted from TechDocs 9.1 (grouped here):
| Area | What TechDocs requires |
|---|---|
| DNS and name | ”Has both forward and reverse lookup entries (FQDN) configured on the DNS server.” “Host name must be same as the FQDN.” In the wizard, “The host FQDN must match the host name (including capitalization) provided during DNS configuration.” |
| Certificate | ”Self-signed certificate regenerated based on FQDN of host.” ESX generates its certificate at install time, before the hostname is set, so its CN is localhost.localdomain; set the FQDN (esxcli system hostname set --fqdn=…), run /sbin/generate-certificates, reboot (Regenerate the Self-Signed Certificate on the ESX Hosts, which runs this “After you have configured the NTP”). External-CA hosts: see §1.4 |
| NICs | ”Hosts can have a single physical NIC or multiple physical NICs. For hosts with multiple physical NICs, only one can be connected to a standard switch, and the NIC must support all management traffic.” “Management IP address is configured on a VMkernel interface connected to the Management Network on the standard switch.” The 9.0 page also says “All physical NICs must have a minimum speed of 10 Gbps”; the 9.1 page no longer lists that line |
| Gateway | ”Host is configured with an appropriate gateway. The gateway must be part of the management subnet.” |
| Disks | ”All disk partitions on HDD and SSD are deleted.” For vSAN: “SSD disks or NVMe disks without pre-existing disk partitions must be available with each host” (Create a New Workload Domain). A host that was in a vSAN cluster before needs its disks wiped: §6.2 |
| Hardware | Certified on the Broadcom Compatibility Guide for the storage type, “Host has the drivers and firmware versions specified in the VMware Hardware Compatibility Guide”, “Hardware health status is healthy without any errors.” |
| ESX build | ”A supported version of ESX is installed on the host.” For a new workload domain: “If the hosts in the management domain have an express patch applied, the hosts for the new workload domain must be imaged with the same express patch.” (Create a New Workload Domain) |
NTP is part of the same host preparation (and the order matters: NTP before
the certificate). Plan the servers in
01-network-dns-plan.md.
Single pNIC, or a management vNIC that’s the only uplink on vSwitch0:
commissioning is the same, the domain/cluster build afterwards is API-only.
See 29-single-pnic-build.md §8
and the single-uplink note in prerequisites.md.
1.3 Pick the storage type per host
The storage type is chosen per host at commission time and decides where the host can go. TechDocs:
- “Hosts that use vSAN storage can only be used with vSAN-based clusters.”
- “Hosts that are commissioned for vSAN HCI with ESA enabled can only be used for vSAN ESA clusters.”
- “Hosts that are commissioned for vSAN HCI without ESA enabled can only be used for vSAN OSA clusters.”
- “Hosts that are commissioned for vSAN Storage (formerly vSAN Max) can only be used for vSAN storage clusters (a subset of vSAN ESA).”
- “Hosts that are commissioned for vSAN Compute Cluster can only be used for vSAN compute clusters.”
- NFS, VMFS on FC and vVols hosts likewise only go into clusters of that type.
| You want | UI choice | JSON / API value |
|---|---|---|
| vSAN OSA | vSAN → vSAN HCI, ESA off | VSAN |
| vSAN ESA | vSAN → vSAN HCI, ESA on | VSAN_ESA |
| vSAN storage cluster (vSAN Max) | vSAN → vSAN Storage | VSAN_MAX |
| vSAN compute cluster | vSAN → vSAN Compute Cluster | VSAN_REMOTE |
| NFS v3 | NFS | NFS |
| VMFS on FC | VMFS on FC | VMFS_FC |
| vVols (deprecated since 9.0) | vVol + protocol FC / iSCSI / NFS | VVOL + vvolStorageProtocolType |
Values from the wizard’s JSON template description (“VSAN, VSAN_REMOTE,
VSAN_ESA, VSAN_MAX, NFS, VMFS_FC, VVOL”, “Use VSAN_REMOTE for a vSAN compute
cluster. Use VSAN_MAX for a vSAN storage cluster.”). Write them in
uppercase: a lowercase vsan_esa makes the API validation fail with an
HTTP 500 (KB 453788, VCF 9.x).
OSA or ESA is your decision, not something the host can tell you. A fresh host has unclaimed disks, so nothing on it says which architecture you intend (VCFHostPreparation’s README makes the same point: “vSAN OSA vs ESA cannot be auto-detected”). Decide per cluster in the design, and commission every host of that cluster with the same value; a host commissioned with the wrong one has to be decommissioned and commissioned again (§6).
Per type, before you commission:
- VMFS on FC: “the hosts must have supported FC cards (Host Bus Adapters)
and drivers installed and configured … The VMFS datastore must be mounted on
all the hosts before commissioning.” Without a VMFS-6 datastore on the LUN
the hosts don’t show up later in the domain/cluster wizard’s host list
(KB 432318,
VCF 9.x). Zoning and the datastore:
prerequisites.md→ VMFS on Fibre Channel. - vVols (deprecated): FC: “all ESX hosts have access to the FC array”; NFS: the hosts “must be able to reach the NFS server from the NFS network assigned in the IP pool”; iSCSI: “the iSCSI software initiator must be enabled on each ESXi host and the VASA provider URL must be listed as the dynamic target.” Add the VASA provider to the inventory before you create the domain or cluster.
- vSAN ESA on disks that aren’t ESA-certified: “With VMware Cloud Foundation 9.0.1 and later, you can use non-certified ESA disks during vSAN ESA cluster configuration for proof-of-concept purposes only” (KB 408300; see §7).
Can’t be commissioned: NFS 4.1, iSCSI, FCoE, NVMe over Fabrics. The
wizard has no storage type for them. Clusters on those types are built in
vCenter and imported; their hosts are added in vCenter, not commissioned
(KB 405095).
The route is
25-cluster-expansion.md §5;
why those types are converge-only is in
prerequisites.md → Principal storage.
1.4 Certificates follow the management domain
TechDocs: “If you used ESX hosts with certificates generated by an external CA for deployment, hosts used for expanding VCF domains or for creating workload domains must also use certificates generated by an external CA.” (Configure ESX Hosts with Signed Certificates). Get this wrong and validation fails with “Host SSL certificate is not trusted by SDDC Manager”; §7 has both fixes from KB 415767.
1.5 Who can commission, and from where
Console. TechDocs documents commissioning in the vSphere Client; the
SDDC Manager UI (deprecated in VCF 9) is the documented fallback. There is no
commissioning wizard in VCF Operations: the Commission Hosts wizard lives in
the vSphere Client (management domain vCenter), lab-verified 2026-11-10. TechDocs 9.0
and 9.1 both: “To perform this task in the vSphere
Client, you must have access to the management domain vCenter, or your VCF
Instance must be configured with VCF SSO and vCenter linking. … As an
alternative, you can perform this task using the SDDC Manager UI”. Linking:
Linking vCenter Instances in VCF Operations;
fleet SSO: 12-sso-configuration.md.
Role. “Access” means an ADMIN role in SDDC Manager. Without it,
Hosts → Unassigned Hosts and Network Pools in the vSphere Client show
“Authorization Error : Unauthorized access. There is a restriction with the
page you are trying to reach and it cannot be displayed.” Assigned Hosts
still works, and being a vCenter administrator isn’t enough. SDDC Manager’s
/var/log/vmware/vcf/commonsvcs/vcf-commonsvcs.log shows
PERMISSION_NOT_FOUND.
Fallback: administrator@vsphere.local. The management vCenter’s own
administrator@vsphere.local always has access to Unassigned Hosts and the
Commission Hosts wizard. Use it as the break-glass account when a named
user hits the error, then fix that user’s access with the table below.
Field-verified 2026-11-10.
Pick the fix by who is logged in:
| Who / where | Fix | Source |
|---|---|---|
| Domain user (e.g. AD), management vCenter | Management vCenter: Administration → Single Sign On → Users and Groups → Groups → SDDCAdmins → Add Members. Add the user, or better a group it belongs to. Log out and back in | KB 415254 (VCF 9.x); field-verified 2026-10-22 |
| VCF SSO user with vCenter linking: the error, or Add Hosts asks you to log in to another management vCenter whose Global Inventory stays blank | Give the user the ADMIN role in each SDDC Manager of the linked environment. The KB’s path is the SDDC Manager UI as administrator@vsphere.local: Administration → Single Sign On → + USER OR GROUP → pick the user → ADMIN → ADD. That console is deprecated in VCF 9, but it’s the documented route | KB 440968 (VCF 9.x) |
| VCF SSO user: the views are simply missing, and password management shows “Cannot convert a Symbol value to a string” | Not a permission problem: the SAML token is too large for SDDC Manager’s nginx (400 Request Header Or Cookie Too Large in vsphere_client_virgo.log). The workaround enlarges nginx’s header buffers on SDDC Manager | KB 425591 (9.0.x) |
| The same message in VCF Operations (Fleet Management → Lifecycle) | The VMware Cloud Foundation adapter is in a warning collection state. Stop and start its collection | KB 447493 (VCF Operations 9.0.2) |
The first two rows are the same gap reached by different routes: neither user
has an ADMIN role in SDDC Manager. That SDDCAdmins maps to ADMIN is
inferred, not stated by Broadcom.
Limits per VCF instance (configmax, SDDC Manager 9.1.0, checked live 2026-11-10): 50 hosts per commission operation (Per VCF Instance → Hosts in a Commission Operation), 10 concurrent commission operations (Concurrent Operations → ESXi Host Commission); the same 50 / 10 for decommission (Hosts in a Decommission Operation, ESXi Host Decommission). TechDocs: “You can also run multiple commission hosts tasks at the same time.”
2. Commission in the vSphere Client
TechDocs 9.1 Commission ESX Hosts (the 9.0 page has the same click path):
- “In the vSphere Client, click Global Inventory Lists → Hosts → Unassigned Hosts.”
- Click Commission Hosts.
- On the Checklist page, review the prerequisites (§1) and confirm by clicking Select All, then Proceed.
- The wizard has two pages, Host Addition and Validation and
Review (lab-verified 2026-11-10). Under Add Hosts, choose Add
new (one host at a time) or Import (many hosts from a JSON file; the
template is the JSON link in the page text “You can either choose to add
host one at a time or download JSON template and perform bulk
commission.”):
- Add new fields: Host FQDN; Storage Type (vSAN, NFS, VMFS on FC, or vVol). For vSAN, the vSAN Type (vSAN HCI, vSAN Storage, or vSAN Compute Cluster) and, for vSAN HCI, whether to use ESA. For vVol, the protocol (FC, iSCSI or NFS). Network Pool Name (drop-down); User Name and Password (the host’s root credentials). Click Add, and repeat for the next host.
- Import: download the JSON template from the link, fill in per host the FQDN, root user name and password, storage type (§1.3 table), network pool name and, for vVols only, the protocol; Browse to the file and Upload.
- The hosts appear under Hosts Added (columns FQDN, Network Pool,
Confirm FingerPrint, Validation Status; lab-verified). “Verify that the server
fingerprint is correct for each host and then activate the Confirm All
Finger Prints toggle.” Compare against a value you read from the host
yourself rather than clicking through: a wrong fingerprint means you are
not talking to the host you think. The wizard shows the SHA-256
fingerprint of the host’s SSL certificate (port 443), in the form
SHA256:<base64>.HostPrep.ps1(§4) writes exactly that value per host into its HTML report, so you can compare them one to one. If you regenerated the certificate (§1.2), the report shows the new one. - Click Validate All. “Each host is marked as Valid or Invalid. For invalid hosts, you can correct the problem and validate again, or select the host and click Remove to proceed with commissioning the valid hosts.”
- Click Next.
- Skip failed hosts during commissioning: “This setting is enabled by default. When enabled, hosts that fail are skipped and commissioning continues on the remaining hosts.”
- Click Commission.
Validate vs commission. Validate All only checks; nothing is added to the inventory until Commission. The commission task then validates and commissions each host again: “After you specify host details and select the network pool to associate the hosts with, the platform validates and commissions each host.”
What the task does to the host, per TechDocs:
- “the platform temporarily activates SSH on the host in order to retrieve the host key. After it retrieves the host key, the platform deactivates SSH on the host.”
- “Multiple VMkernels are created to test the vMotion network. If MAC address filtering is enabled on your physical infrastructure, this may cause issues such as vMotion network connectivity validation failure.”
When it finishes, “The commissioned hosts are added to the Unassigned Hosts table.”
3. Commission through the SDDC Manager API
Validate first, then commission, with the same body. Endpoints from the SDDC Manager API 9.1.1 reference — Hosts (the same paths exist in the 9.0 and 9.1 references):
| Step | Call | Returns |
|---|---|---|
| Token | POST /v1/tokens | Bearer token (as in 21-config-artifacts.md) |
| Pool ID | GET /v1/network-pools | id of the pool from §1.1 |
| Validate | POST /v1/hosts/validations | 202, a Validation with id, executionStatus, resultStatus, validationChecks[] |
| Poll validation | GET /v1/hosts/validations/{id} | same object; wait for executionStatus: COMPLETED, then read resultStatus |
| Commission | POST /v1/hosts | 202, a Task (type HOST_COMMISSION); Location: /v1/tasks/{id} |
| Poll task | GET /v1/tasks/{id} | status until SUCCESSFUL or FAILED |
| Check | GET /v1/hosts?status=UNASSIGNED_USEABLE | the new hosts with their id |
POST /v1/hosts/validations/commissions is marked Deprecated in the 9.1.1
reference; use POST /v1/hosts/validations.
The body is an array of
HostCommissionSpec,
one entry per host:
[
{
"fqdn": "sfo01-w01-esx01.sfo.rainpole.io",
"username": "root",
"password": "<ESX root password>",
"storageType": "VSAN_ESA",
"networkPoolId": "<network pool ID>",
"networkPoolName": "sfo-w01-np01"
},
{
"fqdn": "sfo01-w01-esx02.sfo.rainpole.io",
"username": "root",
"password": "<ESX root password>",
"storageType": "VSAN_ESA",
"networkPoolId": "<network pool ID>",
"networkPoolName": "sfo-w01-np01"
}
]
Field by field (9.1.1 reference):
fqdn,username,password,storageType,networkPoolId— required.networkPoolNameis optional; the pool is bound by ID.storageType— the schema lists “VSAN, VSAN_ESA, VSAN_REMOTE, VSAN_MAX, NFS, NFS41, VMFS_FC, VVOL, VMFS, FC”. The extra values (NFS41,VMFS,FC) appear in the schema, but TechDocs’ commission procedure offers only the seven in §1.3. Stick to those; this repo hasn’t tested the others.vvolStorageProtocolType—ISCSI,NFSorFC; vVols only.sshThumbprint,sslThumbprint— optional: “SSH thumbprint of the host”, “SSL thumbprint of the host”. This is the API’s counterpart of the wizard’s fingerprint confirmation.skipHclCompatibilityPrecheck— optional: “This option bypasses vSAN ESA HCL validation during host commissioning. Set this flag only when hosts lack vSAN ESA certified disks or when SDDC Manager cannot verify disk status. (Using vSAN ESA certified disk(s) is recommended.)”
The file holds root passwords: treat it as a secret, and don’t commit it.
Where to run it without a script: VCF Operations’ API Explorer. TechDocs (in the stretch procedure): “Log in to VCF Operations. In the navigation pane, click Developer Center APIs & SDKs. Click the API Explorer link on the SDDC Manager API.”
Decommission is the reverse: DELETE /v1/hosts with an array of
HostDecommissionSpec ([ { "fqdn": "…" } ]), returning a
HOST_DECOMMISSION task.
4. Scripted: VCFHostPreparation
VCFHostPreparation (Paul van Dieen) wraps §1.2 and §3 in two PowerShell scripts. Background and a walkthrough: the blog post Automating ESXi host preparation for VCF 9 with PowerShell. Run them in order:
HostPrep.ps1reads a text file of host FQDNs and, per host, checks forward and reverse DNS, sets and checks NTP, setsConfig.HostAgent.ssl.keyStore.allowSelfSigned, detects the storage type, regenerates the certificate when its CN doesn’t match the FQDN (offering to fix a host still calledlocalhost.localdomainfirst), and can reset the root password to a VCF 9 compliant value.-WipeDiskclears old partitions on vSAN hosts (§6.2);-DryRunand-WhatIfReportchange nothing. Output: an HTML report with each host’s SSL thumbprint in theSHA256:<base64>format the wizard shows (§2 step 5), and a commissioning CSV.Commission-VCFHosts.ps1reads that CSV, asks for the SDDC Manager, the network pool and the root password, runsPOST /v1/hosts/validations(with an HTML validation report; it stops on a failure), thenPOST /v1/hosts, polls the task, and lists each host’s SDDC Manager ID.-ValidateOnlystops after validation.
The storage type is the one manual step. HostPrep.ps1 writes VMFS_FC when
it finds an FC HBA, NFS when an NFS datastore is mounted, and VSAN
otherwise. Edit the CSV’s StorageType column to VSAN_ESA (or VVOL)
before running Commission-VCFHosts.ps1: it sends the value as is.
Requirements and all parameters are in the repo’s README.
5. Check the result
- vSphere Client: Global Inventory Lists → Hosts → Unassigned Hosts lists the new hosts (View ESX Host Inventory).
- API:
GET /v1/hosts?status=UNASSIGNED_USEABLEreturns them, with theidthat the domain, cluster, expansion and stretch specs ask for. The other states areASSIGNEDandUNASSIGNED_UNUSEABLE. Filter further withstorageType=ornetworkpoolId=to check a host landed with the right type in the right pool. - Progress or failure detail: VCF Operations → Fleet Management →
Tasks, or
GET /v1/tasks/{id}.
Then go back to the runbook you came from: 23 (new domain), 24 (new cluster), 25 (add to a cluster) or 22 (stretch).
6. Decommission and recommission
6.1 Decommission
From Decommission ESX Hosts: “The hosts that you want to decommission must not be assigned to a VCF domain. If a host is assigned to a VCF domain, you must remove it from the VCF domain before you can decommission it.” So remove the host from its cluster first, then:
- “In the vSphere Client, click Global Inventory Lists → Hosts → Unassigned Hosts.”
- Select one or more hosts and click Decommission Selected Hosts.
- Skip failed hosts during decommissioning (on by default).
- Click Confirm.
API: DELETE /v1/hosts (§3).
If a decommission doesn’t complete, SDDC Manager can keep the host’s record, and a later commission of the same FQDN fails (“Host(s) already present in inventory”, §7).
6.2 Recommission a former vSAN host
Re-image when the host was in a domain. TechDocs: “If the decommissioned host was previously part of a VCF domain, you must re-image it before commissioning it again and adding it to a VCF domain. If the decommissioned host was not part of a VCF domain, you do not have to re-image it before commissioning it again.” The same applies to moving a host between domains: “the host must be decommissioned, re-imaged, and commissioned before it is added to the workload domain” (KB 414650, VCF 5.x and 9.x). Commissioning is also where you correct a wrong storage type (OSA vs ESA).
Wipe the old vSAN partitions. A reinstall doesn’t necessarily clear the data disks, and leftover vSAN partitions fail validation with “vSAN Partition found on the host, needs clean-up” (KB 399567). First make sure “the host is not part of any active vSAN cluster and does not contain any active objects” (same KB). Then either:
- Host Client: Storage → Devices → the vSAN-eligible device → Actions → Clear partition table, for each device (KB 399567), or
- ESX shell:
vdq -ilists the disks; ESA hosts first needesxcli vsan storagepool remove -u <device UUID>per device andesxcli vsan cluster leave(KB 399567), or - Scripted:
.\HostPrep.ps1 -WipeDisk(add-DryRunto see first). Per its README it runs only for hosts detected asVSAN, always excludes the boot disk, asks Y/N per host, unmounts VMFS on the target disks and runspartedUtil mklabel <device> gptover SSH.
Wiping is destructive: check the device list before you confirm.
7. Common failures
Logs: commission and validation errors land in
/var/log/vmware/vcf/operationsmanager/operationsmanager.log on SDDC Manager
(per the KBs below).
| Symptom | Cause and fix | Source |
|---|---|---|
| ”Authorization Error : Unauthorized access” on Unassigned Hosts / Network Pools | No ADMIN role in SDDC Manager; see the table in §1.5 | KB 415254 (field-verified), KB 440968, KB 425591, KB 447493 |
”Host SSL certificate is not trusted by SDDC Manager” (SDDC_TRUST_HOST_SSL_CERT_FAILED) | The host has a default certificate while the management hosts use external-CA certificates. Bring-up with external certs: replace the new host’s certificate with an external one, then commission. Default at bring-up, external later: switch the ESX certificate mode in vCenter to vmca, commission, switch back | KB 415767 (VCF 5.2.x, 9.x) |
| “Unable to get the Host Fingerprint, verify ESXi host connectivity or credentials or SSH” | SSH unavailable, a firewall between SDDC Manager and the host, or a certificate CN that’s the short name. Check the CN with openssl s_client -connect <host-fqdn>:443 -showcerts from SDDC Manager; fix the FQDN and regenerate the certificate (§1.2) | KB 388823 (VCF 9.0.x) |
| “vSphere standard switch vSwitch0 must have only one NIC as Uplink” | More than one NIC on vSwitch0; the KB names a NIC below 10 Gbps as the cause. Leave one uplink on vSwitch0 (§1.2); you can remove the failing host and commission the rest | KB 412127 (VCF 9) |
| “vSAN Partition found on the host, needs clean-up” | Old vSAN partitions or an ESA storage pool on the disks; wipe them (§6.2) | KB 399567 |
VSAN_ESA_HOST_NOT_HCL_COMPATIBLE, or “Failed to validate vSAN HCL status” with vSAN ESA | The disks can’t be verified as vSAN ESA certified against SDDC Manager’s HCL. Check the HCL data on SDDC Manager is current; for a proof of concept on non-certified disks, KB 408300’s property vsan.esa.sddc.managed.disk.claim=true in /etc/vmware/vcf/operationsmanager/application.properties, or the API’s skipHclCompatibilityPrecheck | KB 408300 (9.0.1+, 9.1.x) |
Http failure response for …/ui/api/v1/hosts/prechecks: 500 OK with “Failed to validate vSAN HCL status” on a JSON import | The storage type is in lowercase (vsan_esa); write it in uppercase | KB 453788 (VCF 9.x) |
| “Host(s) already present in inventory, validation failed” after re-imaging | An earlier decommission was incomplete and left the FQDN in the SDDC Manager database. The fix is a database clean-up: open a Broadcom support request | KB 424242 (VCF 9.x) |
| “[Applicable to skipped ESXi Hosts]Perform set of validations for current workflow” | The guardrail check can’t reach NSX Manager by name. Fix forward and reverse DNS for the NSX Manager FQDN, as resolved from SDDC Manager, VCF Operations and vCenter | KB 423908 (VCF 9.x) |
| The ping test from SDDC Manager to the host’s management network fails (for example, ICMP blocked between them) | 9.1 and later can skip the test: validation.disable.ping.connectivity.check=true in /etc/vmware/vcf/operationsmanager/application.properties, then systemctl restart operationsmanager | KB 432391 (VCF 5.2.3, 9.1+) |
| vMotion network connectivity validation fails | The commission creates “Multiple VMkernels … to test the vMotion network”; MAC address filtering on the physical network can block them. Also check the pool’s vMotion VLAN, MTU and gateway | TechDocs Commission ESX Hosts |
| Commissioned VMFS_FC hosts missing from the domain/cluster wizard (“No suitable hosts available”) | No VMFS-6 datastore on the presented LUN; create it on the hosts and refresh the wizard | KB 432318 (VCF 9.x) |
| Host on an imported cluster: “Cluster is marked as imported. This operation is not allowed on imported clusters” | Converge-only storage; don’t commission, add in vCenter | KB 405095, 25 §5 |
KB 408300 and KB 432391 change SDDC Manager’s validation behaviour. Use them for what the KB describes, not to push an unsupported host through.
Sources
- TechDocs 9.1: Commission ESX Hosts, Decommission ESX Hosts, Network Pool Management, View ESX Host Inventory (all last updated 2026-11-09, read 2026-11-10)
- TechDocs 9.0: Commission ESX Hosts (same click path; adds the 10 Gbps line)
- TechDocs 9.1: Preparing ESX Hosts for VCF or vSphere Foundation, Regenerate the Self-Signed Certificate on the ESX Hosts, Create a New Workload Domain (express patch, vSAN disks)
- API: SDDC Manager API 9.1.1 — Hosts, HostCommissionSpec, Validate Host Commission Spec, Commission Hosts, Decommission Hosts, Get Hosts
- configmax: SDDC Manager 9.1.0 (Hosts in a Commission Operation 50, ESXi Host Commission 10 concurrent; same for decommission), checked live 2026-11-10
- KB: 399567, 405095, 408300, 412127, 414650, 415254, 415767, 423908, 424242, 425591, 432318, 432391, 440968, 447493, 453788, 388823
- Tool: VCFHostPreparation (
HostPrep.ps14.4.0,Commission-VCFHosts.ps13.1.4); blog post: Automating ESXi host preparation for VCF 9 with PowerShell - Lab: Commission Hosts wizard in the vSphere Client, screenshot of Host Addition and Validation, 2026-11-10