Academy/Holodeck Lab Setup & Operations/Holodeck Instance Lifecycle Management
This lab targets VCF 9.0.2

Holodeck Instance Lifecycle Management

VCF 9.0.2Intermediateadminarchitectvcdx⏱ 150 min

Lifecycle cmdlets are the same for VCF 9.0.0.0 through 9.0.2.0 on Holodeck 9.0.2 (Stop-/Start-HoloDeckInstance, Remove-HoloDeckInstance [-ResetHoloRouter], New-/Get-/Import-HoloDeckConfig, Reset-HoloDeckState, -DeveloperMode). Task 4 compares a VCF 5.2.2 instance (Cloud Builder, vSAN OSA only, 60-day License Later) with the VCF 9.0.2 instance (VCF Installer, ESA or OSA, 90 days) using the documented support for both on Holodeck 9.0. Source-checked 1 Oct 2026 against the Holodeck 9.0.2 documentation at vmware.github.io/Holodeck/9.0.2/ (index: Holodeck Environment Overview, Holorouter Overview, Concepts, Pre-requisites, Deployment, Post Deployment, Day 2 Operations, How To; cmd_reference; FAQ) and, where VCF behaviour is described, the VCF 9.0.1 documentation. In Holodeck 9.x the toolkit is a PowerShell module built into the HoloRouter OVA (Photon OS): the operator SSHes to the HoloRouter, runs pwsh, and works under /holodeck-runtime. There is no Windows HoloConsole, no C:\Holodeck path, no -ConfigFile parameter and no separate Prepare phase. Rewritten for v8.2.2 after the 360-degree audit.

Objectives

  • Design and execute a multi-checkpoint snapshot strategy for a Holodeck instance using Stop-HoloDeckInstance, outer-layer vSphere snapshots of the nested VMs, and Start-HoloDeckInstance, and explain why Holodeck itself has no snapshot cmdlet
  • Remove an instance with Remove-HoloDeckInstance, decide when -ResetHoloRouter is needed, and verify that the nested ESX hosts and the VCF Installer VM are gone from the outer layer while the HoloRouter and its config survive
  • Redeploy from the existing config with Import-HoloDeckConfig and the same New-HoloDeckInstance command, and compare redeploy time (hours) with snapshot revert time (minutes) to choose a reset method per scenario
  • Manage configs the documented way: one config per deployment (1:1), never edit /holodeck-runtime/config/<config-ID>.json by hand, New-HoloDeckConfig for each new deployment, Get-/Import-HoloDeckConfig to switch, and keep a versioned record of the commands that produced each instance
  • Deploy a VCF 5.2.2 instance beside the 9.0.2 instance (second HoloRouter and trunk port group) and articulate the architectural differences: Cloud Builder versus VCF Installer, SDDC Manager as the sole control plane versus the VCF Operations fleet model, vSAN OSA only versus ESA or OSA, 60 versus 90 days License Later, BOM versions
  • Automate repeatable deployments with -DeveloperMode environment variables and a reset script, and document a lab reset runbook that names the cmdlet, the guard rail and the expected duration of each step

Prerequisites

holodeck-02 complete: a management-only VCF 9.0.2 instance (InstanceID holo) healthy behind a HoloRouter, with the holodeck-02-complete snapshot set already taken at the outer vSphere layer. Outer datastore headroom for two further snapshot sets (budget 200 GB or more). For Task 4: physical capacity for a second pod per the VCF 5.2 single-site row (16 CPU, 384 GB RAM, 2 TB) and the VCF 5.2.2 binaries (ESXi 8.0 U3g ISO and Cloud Builder 5.2.2 OVA) staged in /holodeck-runtime/bin/5.2.2/ on the second HoloRouter.

Prior labs: holodeck-02

Required skills:

  • HoloRouter SSH and pwsh (holodeck-01, 02)
  • Outer-layer vSphere snapshot operations
  • PowerShell scripting basics
  • Reading the VCF 9.0 and 5.2 Bills of Materials
📸 Starting State: S3-9x — Management Domain Deployed (VCF 9.x)

Lab Environment

Pod 1: HoloRouter (holorouter-a) on External plus Trunk-A port groups with the VCF 9.0.2 instance holo (esx-01a..04a, vcfinstaller-a, vc-mgmt-a, sddcmanager-a, nsx-mgmt-a, ops-a) on 10.1.0.0/20. Pod 2 (Task 4 only): a second HoloRouter on its own trunk port group (Trunk-52) with a VCF 5.2.2 instance (Cloud Builder, SDDC Manager 5.2.2, vCenter 8.0 U3g, NSX 4.2.3) also on the default 10.1.0.0/20, isolated from Pod 1 by the separate port group as the Holodeck overview documents for multiple pods.

Mermaid source
graph TB
  OUTER[Outer vSphere: snapshots, power, datastore] --- P1[Pod 1: holorouter-a + Trunk-A]
  OUTER --- P2[Pod 2: holorouter-52 + Trunk-52]
  P1 --- I1[VCF 9.0.2 instance holo: VCF Installer, SDDC Manager, VCF Operations]
  P2 --- I2[VCF 5.2.2 instance: Cloud Builder, SDDC Manager 5.2.2]
  P1 -.- CFG[/holodeck-runtime/config/<config-ID>.json, state, logs/]

IP Addressing

NetworkPurposeVLAN
10.1.0.0/20Default Holodeck Site A master CIDR, subdivided by New-HoloDeckNetworkConfig into the Site A subnets (management, vMotion, vSAN, host overlay, edge uplinks, and others)VLANs 0 and 10-25 (Site A default)
10.2.0.0/20Default Holodeck Site B master CIDR (dual-site only)VLANs 40-58 (Site B default)
<external management network>HoloRouter external management IP (DHCP or static, set at OVA deployment); the only external IP a Holodeck environment needsExternal port group
10.1.0.0/20 (Pod 2)Default Site A CIDR reused by the VCF 5.2.2 pod; no conflict because each pod sits on its own trunk port group behind its own HoloRouterVLANs 0 and 10-25 on Trunk-52

Credentials

SystemUsernamePassword
HoloRouter (SSH and pwsh)rootSet in the OVA Additional settings at deployment
HoloRouter webtop(no authentication in 9.0.2)http://<holorouter-ip>:30000; the docs warn not to expose it externally
VCF Installeradmin@localHolodeck documented default (Post Deployment table)
VCF OperationsadminHolodeck documented default (Post Deployment table)
Management vCenteradministrator@vsphere.localHolodeck documented default (Post Deployment table)
SDDC Manageradministrator@vsphere.localHolodeck documented default (Post Deployment table)
Management NSXadminHolodeck documented default (Post Deployment table)
Nested ESX hostsrootHolodeck documented default (Post Deployment table)
Cloud Builder (VCF 5.2.2 pod)adminHolodeck documented default (Post Deployment table: 'admin for Cloud Builder')

Tasks

Task 1 Design and execute a multi-checkpoint snapshot strategy

Recoverability

Holodeck 9.0.2 documents Stop-HoloDeckInstance and Start-HoloDeckInstance for power, and nothing for snapshots: checkpoints are taken at the outer vSphere layer on the nested VMs. A consistent set needs the instance powered off first.

Step 1

Define the checkpoint plan in your notebook: C0 holodeck-01-baseline (HoloRouter only, taken in holodeck-01), C1 holodeck-02-complete (management domain up), C2 holodeck-07-pre-change (before any destructive task in this lab), C3 holodeck-07-complete. For each: which VMs are in the set (all nested ESX hosts, the VCF Installer VM and the HoloRouter), the required power state, the retention and the datastore cost estimate.

Four checkpoints defined with VM set, power state, retention and cost.
Step 2

SSH to the HoloRouter as root, start pwsh, import the config of the instance (Get-HoloDeckConfig to list, Import-HoloDeckConfig -ConfigId <id>) and record the instance: Get-HoloDeckInstance -InstanceID holo.

ssh root@<holorouter-ip>
pwsh
Get-HoloDeckConfig
Import-HoloDeckConfig -ConfigId <config-id>
Get-HoloDeckInstance -InstanceID holo
Config imported into $config; instance components and deployment state listed.
Step 3

Power the instance off cleanly with the documented cmdlet. It asks for confirmation; -Force bypasses the prompt (use it only in scripts).

Stop-HoloDeckInstance -InstanceID holo
Nested hosts and the installer VM powered off; confirm in the outer vSphere Client.
Step 4

In the outer vSphere Client take snapshot holodeck-07-pre-change on every nested VM of the instance (esx-01a..04a, vcfinstaller-a) and on the HoloRouter. Do not snapshot memory (the VMs are off); note the start time of the first snapshot and the end time of the last.

Snapshot set present on all VMs; elapsed time recorded.
Step 5

Power the instance back on with Start-HoloDeckInstance -InstanceID holo and time the cold start until SDDC Manager, vc-mgmt-a, nsx-mgmt-a and ops-a answer on their FQDNs (Post Deployment table).

Start-HoloDeckInstance -InstanceID holo
All management components reachable; cold-start time recorded (typically tens of minutes).
Step 6

Measure the snapshot cost: in the outer vSphere Client compare datastore free space before and after the set, and read the delta-disk sizes in each VM's Snapshots view after an hour of uptime. Record the growth rate.

Snapshot footprint and growth rate recorded against the 200 GB budget.
Step 7

Write the snapshot policy: maximum two sets per VM in the lab, revert only with the instance powered off (Stop-HoloDeckInstance first, then revert all VMs in the set, then Start-HoloDeckInstance), never revert a subset of the nested hosts (vSAN and SDDC Manager inventory would diverge), and delete sets older than the current lab plus one.

Policy written with the revert ordering and the subset rule.

Validation Gate

Check: Checkpoint plan written; holodeck-07-pre-change set exists on all nested VMs and the HoloRouter; cold-start time and snapshot footprint recorded; revert ordering documented.

Expected: All items present in the lab notebook: revert ordering documented.

Common Errors

Snapshot taken with the instance running
Cause: Nested vSAN and SDDC Manager state captured mid-flight
Fix: Stop-HoloDeckInstance first, then snapshot
Reverting only the failed host
Cause: vSAN cluster and SDDC Manager inventory diverge from the reverted host
Fix: Revert the whole set in one operation
Datastore fills during the lab
Cause: Delta disks grow with vSAN resync traffic
Fix: Keep to two sets; consolidate after each lab

Task 2 Remove the instance and verify the cleanup

Manageability

Remove-HoloDeckInstance deletes all nested ESX hosts and the VCF Installer (or Cloud Builder) VM of the instance. -ResetHoloRouter additionally removes the networking configured for the instance and reboots the HoloRouter; New-HoloDeckInstance configures it again. The config file and the binaries stay.

Step 1

Confirm the holodeck-07-pre-change snapshot set from Task 1 is complete (this task is destructive). Record the outer datastore usage and the VM list of the instance (esx-01a..04a, vcfinstaller-a) before removal.

Pre-removal inventory and datastore usage recorded.
Step 2

In pwsh with the config imported, remove the instance without resetting the HoloRouter first, so you can observe what remains: Remove-HoloDeckInstance. Confirm the prompt.

Remove-HoloDeckInstance
Nested ESX hosts and the VCF Installer VM deleted from the outer layer; HoloRouter still running.
Step 3

Verify at the outer layer: the nested VMs are gone, the HoloRouter VM remains, datastore space is released (snapshots of the deleted VMs go with them). Verify on the HoloRouter: /holodeck-runtime/config/<config-ID>.json and /holodeck-runtime/bin/9.0.2.0/ are untouched; Get-HoloDeckSubnet -Site a and Get-HoloDeckDNSConfig still list the instance's subnets and records because the networking was not reset.

ls -l /holodeck-runtime/config/ /holodeck-runtime/bin/9.0.2.0/
Get-HoloDeckSubnet -Site a | ft -AutoSize
Get-HoloDeckDNSConfig | select -First 20
Config and binaries present; subnets and DNS records still present.
Step 4

Decide the -ResetHoloRouter rule for your runbook: use it when the next deployment changes the network plan (custom CIDR or VLAN range, single to dual site) or when the networking state is suspect; skip it for a same-plan redeploy to save the HoloRouter reboot. Record that the reset reboots the HoloRouter automatically and drops its DNS until New-HoloDeckInstance rebuilds it.

Reset rule written with both cases.
Step 5

Record the Task 2 timings: removal duration, outer datastore reclaimed, and whether anything of the instance survived that you did not expect (compare against the Task 1 inventory).

Cleanup verified and timed.

Validation Gate

Check: Nested VMs removed; HoloRouter, config and binaries intact; networking state observed intact without -ResetHoloRouter; reset rule written.

Expected: All items present in the lab notebook: reset rule written.

Common Errors

Remove-HoloDeckInstance 'no config loaded'
Cause: New pwsh session without Import-HoloDeckConfig
Fix: Get-HoloDeckConfig then Import-HoloDeckConfig -ConfigId
HoloRouter unreachable after -ResetHoloRouter
Cause: Documented automatic reboot
Fix: Wait for the reboot; reconnect by its external IP
Expecting the config file to be deleted
Cause: Remove-HoloDeckInstance removes VMs, not configs
Fix: Configs stay under /holodeck-runtime/config/

Task 3 Redeploy from the existing config and compare reset methods

Recoverability

Each config maps to exactly one deployment; the same New-HoloDeckInstance command with the same config redeploys the instance, and Holodeck's idempotent state lets a failed run resume. Timing the redeploy against the snapshot revert gives you the reset decision table.

Step 1

In a fresh pwsh session import the same config (Import-HoloDeckConfig -ConfigId <id>) and check $config shows the description and ID of the original deployment.

pwsh
Import-HoloDeckConfig -ConfigId <config-id>
$config
Original config loaded.
Step 2

Run exactly the New-HoloDeckInstance command recorded in holodeck-02 (same -Version 9.0.2.0, -InstanceID holo, -ManagementOnly and any -CIDR, -VLANRangeStart, -DNSDomain or -vSANMode values). Note the start time. Because the Task 2 removal did not reset the HoloRouter, observe whether the networking phase is skipped or re-applied.

New-HoloDeckInstance -Version 9.0.2.0 -InstanceID holo -ManagementOnly
Pre-checks pass; nested hosts rebuilt; VCF Installer deployed; bring-up runs. Start time recorded.
Step 3

If the run fails, do not start over: re-run the identical command and watch it resume at the failed step (idempotency). Only a corrupted state justifies Reset-HoloDeckState (irreversible) followed by a clean run. Record which happened.

Resume behaviour observed or not needed.
Step 4

When the deployment completes, record the end time and validate with Get-HoloDeckInstance -InstanceID holo plus the four management FQDNs. Compute the redeploy time (the docs give 4-5 hours for -ManagementOnly on 9.0.0.0 as an indication).

Get-HoloDeckInstance -InstanceID holo
Redeploy time recorded; instance healthy.
Step 5

Now time the alternative: Stop-HoloDeckInstance, revert the whole holodeck-07-pre-change set at the outer layer (if you kept a copy of the set on a clone, or use the holodeck-02-complete set of the new VMs taken now), Start-HoloDeckInstance, and measure until the management FQDNs answer.

Revert-and-start time recorded (minutes to low tens of minutes).
Step 6

Build the reset decision table: scenario (failed lab step, broken networking, new network plan, new VCF version, expired 90-day License Later window) against method (snapshot revert, re-run same command, Remove-HoloDeckInstance then redeploy, Remove-HoloDeckInstance -ResetHoloRouter then redeploy, new config plus redeploy) with the measured or documented duration of each.

Decision table with five scenarios and durations.

Validation Gate

Check: Instance redeployed from the same config; redeploy and revert times measured; decision table complete, including that the License Later expiry can only be addressed by redeploy or licensing.

Expected: All items present in the lab notebook: decision table complete, including that the License Later expiry can only be addressed by redeploy or licensing.

Common Errors

Changing parameters on the redeploy
Cause: Custom CIDR or VLAN values omitted get overwritten by defaults
Fix: Reuse the recorded command verbatim
Reset-HoloDeckState after every failure
Cause: Throws away the idempotent state
Fix: Re-run first; reset only on corrupted state
Reusing one config for a second concurrent deployment
Cause: Documented 1:1 rule; state conflicts
Fix: New-HoloDeckConfig per deployment

Task 4 Deploy a VCF 5.2.2 pod beside the 9.0.2 pod and document the differences

Manageability

Holodeck 9.0 supports VCF 5.2, 5.2.1, 5.2.2 and 9.0.0.0 through 9.0.2.0. A second pod on its own trunk port group with its own HoloRouter isolates the 5.2.2 instance so both can run with default addressing. The comparison is the deliverable.

Step 1

Create Trunk-52 on the outer vSS/vDS with the same settings as Trunk-A (VLAN trunk, MTU 9000, the documented security settings) and deploy a second HoloRouter OVA (4 vCPU, 8 GB RAM from Holodeck 9.0.1, 75 GB; External plus Trunk-52; SSH enabled) with its own external IP.

Second HoloRouter reachable over SSH.
Step 2

Stage the VCF 5.2.2 binaries on the second HoloRouter: scp the ESXi 8.0 U3g ISO and the Cloud Builder 5.2.2 OVA to /holodeck-runtime/bin/5.2.2/. Note that the offline depot is not used for 5.2.x (Cloud Builder carries the bundle).

scp /<local-path>/<ESXi ISO> root@<holorouter-52-ip>:/holodeck-runtime/bin/5.2.2/
scp /<local-path>/<Cloud Builder OVA> root@<holorouter-52-ip>:/holodeck-runtime/bin/5.2.2/
Two binaries in bin/5.2.2/.
Step 3

On the second HoloRouter create a dedicated config and deploy a management-only 5.2.2 instance. vSAN OSA is the only option for 5.2 (-vSANMode ESA is not supported there); -VVF yields no result with 5.2; -NsxEdgeClusterMgmtDomain would include AVN on 5.2. Record the start time.

pwsh
New-HoloDeckConfig -Description 'vcf522-compare' -TargetHost <target> -Username <user> -Password <password>
New-HoloDeckInstance -Version 5.2.2 -InstanceID holo52 -ManagementOnly
Pre-checks pass; Cloud Builder deployed instead of VCF Installer; bring-up runs.
Step 4

While it deploys, open the Pod 1 (9.0.2) consoles and list the components that exist only in 9.0: VCF Operations (ops-a) as the fleet management and lifecycle UI, VCF Installer, the offline or online depot, the Identity Broker option; then list what the 5.2.2 pod will have instead: Cloud Builder (login admin), SDDC Manager as the lifecycle and inventory UI, bundles downloaded by SDDC Manager, no VCF Operations fleet layer.

Component lists for both versions.
Step 5

When the 5.2.2 pod is up, log in to Cloud Builder and SDDC Manager 5.2.2, record the BOM (vCenter 8.0 Update 3g build 24853646, ESXi 8.0 Update 3g build 24859861, NSX 4.2.3 build 24866349, Cloud Builder and SDDC Manager 5.2.2 build 24936865 per the VCF 5.2.2 BOM) against the 9.0 BOM (vCenter 9.0 build 24755230, ESX build 24755229, NSX 9.0.0.0 build 24733065 for 9.0.0; your 9.0.2 instance shows its own builds), and the License Later windows (60 days for 5.2, 90 days for 9.0).

BOM and licensing comparison recorded from live consoles.
Step 6

Write the comparison table: Installer (Cloud Builder vs VCF Installer), control plane (SDDC Manager alone vs SDDC Manager plus VCF Operations fleet management), storage (OSA only vs ESA or OSA), depot (bundle in Cloud Builder vs online or offline depot), licensing window, BOM, upgrade path (5.2.x in-place to 9.0 per vcf-evolution labs), Holodeck binaries path (bin/5.2.2 vs bin/9.0.2.0), identity (vCenter SSO vs VCF SSO with embedded or appliance Identity Broker).

Nine-row comparison table.
Step 7

Decide the fate of Pod 2: keep it for the vcf-evolution labs (an in-place upgrade candidate) or remove it with Remove-HoloDeckInstance -ResetHoloRouter on the second HoloRouter. Record capacity released.

Decision recorded; capacity noted.

Validation Gate

Check: Second pod isolated on Trunk-52; 5.2.2 instance deployed with Cloud Builder; comparison table complete and sourced from the live consoles and the two BOMs.

Expected: All items present in the lab notebook: comparison table complete and sourced from the live consoles and the two BOMs.

Common Errors

Pre-check 'binary not found' on the 5.2.2 run
Cause: Files staged in bin/9.0.2.0 or wrong file names
Fix: Stage ESXi ISO and Cloud Builder OVA in bin/5.2.2/
Trying -vSANMode ESA with 5.2.2
Cause: 5.2 is OSA only on Holodeck
Fix: Omit -vSANMode or set OSA
Both pods on one trunk port group
Cause: Duplicate 10.1.0.0/20 and VLANs collide
Fix: One trunk port group per pod
Looking for VCF Operations on 5.2.2
Cause: Fleet management arrives with 9.0
Fix: Use SDDC Manager 5.2.2 for lifecycle

Task 5 Automate deployments with -DeveloperMode and build the lab reset runbook

Manageability

-DeveloperMode takes every interactive input (depot, proxy, datastore, trunk port group, vCenter cluster and datacenter) from environment variables, so a deployment or a reset becomes a script. The HoloRouter also documents a job-scheduling service for recurring commands.

Step 1

In pwsh on the Pod 1 HoloRouter set the documented variables for your depot type. Online: $env:brcm_build_token and $env:enable_proxy (plus proxy_protocol, proxy_ip, proxy_port, enable_proxy_auth, proxy_username, proxy_password when a proxy is used). Offline: offline_depot_ip, offline_depot_port, offline_depot_username, offline_depot_password, offline_depot_protocol. Always: datastore_name, trunk_port_group_name; with a vCenter target also cluster_name and dc_name. Keep secrets out of the saved script (prompt for them or read from a file outside the repo).

$env:brcm_build_token = '<token>'
$env:enable_proxy = 'n'
$env:datastore_name = '<datastore>'
$env:trunk_port_group_name = 'Trunk-A'
$env:cluster_name = '<cluster>'
$env:dc_name = '<datacenter>'
Variables set in the session.
Step 2

Run New-HoloDeckConfig and New-HoloDeckInstance -DeveloperMode with the same parameters as the manual run and confirm no interactive prompt appears. For dual site the docs require two pwsh sessions each with the variables set.

New-HoloDeckConfig -Description 'auto-9022' -TargetHost <target> -Username <user> -Password <password>
New-HoloDeckInstance -Version 9.0.2.0 -InstanceID holo -ManagementOnly -DeveloperMode
Deployment proceeds without prompts.
Step 3

Write reset-lab.ps1 on the HoloRouter (outside /holodeck-runtime/config): import the config by ID, Stop-HoloDeckInstance -Force, then either (a) call the outer vCenter with PowerCLI to revert the named snapshot set and Start-HoloDeckInstance, or (b) Remove-HoloDeckInstance and re-run New-HoloDeckInstance -DeveloperMode. Parameterise the mode and the snapshot name; log start and end times to a file under /holodeck-runtime/logs/.

Script written with two modes and timing output.
Step 4

Test mode (a) once end to end and record the duration; dry-run mode (b) only to the Remove step if capacity or time is short, noting the Task 3 redeploy time.

Mode (a) proven; mode (b) timing taken from Task 3.
Step 5

Explore the HoloRouter's job-scheduling service named in the Holorouter Overview ('Job scheduling: Allows users to schedule commands/scripts to be run recursively', vmware.github.io/Holodeck; the documentation does not describe its interface): check what the appliance exposes (for example crontab -l and systemctl list-timers as root) and record how a recurring reset-lab.ps1 run could be scheduled. Mark anything you could not confirm as ASSUMED in your runbook.

crontab -l
systemctl list-timers --all | head
Scheduling mechanism identified or marked ASSUMED.
Step 6

Write the lab reset runbook: for each step the cmdlet, the guard rail (snapshot set present, instance powered off, config imported, -Force only in scripts, Reset-HoloDeckState only on corrupted state, -ResetHoloRouter only on plan change), and the expected duration from your measurements.

Runbook complete with cmdlet, guard rail and duration per step.

Validation Gate

Check: -DeveloperMode deployment runs without prompts; reset-lab.ps1 works in snapshot mode; scheduling option recorded with its evidence status; runbook complete.

Expected: All items present in the lab notebook: runbook complete.

Common Errors

Prompt still appears in -DeveloperMode
Cause: A variable missing or misspelt for the chosen depot type
Fix: Set the full documented variable set for online or offline
Secrets committed with the script
Cause: Token and depot password hard-coded
Fix: Prompt or read from a protected file
Scheduled reset collides with a running lab
Cause: No guard
Fix: Check Get-HoloDeckInstance state and a lock file before acting

Final Validation

The instance lifecycle is under control: a consistent snapshot strategy with Stop-/Start-HoloDeckInstance, a verified Remove-HoloDeckInstance with a documented -ResetHoloRouter rule, a timed redeploy from the same config against a timed snapshot revert, a VCF 5.2.2 pod beside the 9.0.2 pod with a sourced comparison, and an automated reset runbook.

✓ Snapshots → holodeck-07-pre-change and holodeck-07-complete sets taken with the instance powered off; footprint within budget

✓ Removal → Nested VMs gone; HoloRouter, config, binaries intact; networking state understood

✓ Redeploy → Same command, same config; redeploy and revert times recorded; decision table complete

✓ Comparison → 5.2.2 pod deployed via Cloud Builder; nine-row table sourced from consoles and BOMs

✓ Automation → -DeveloperMode run without prompts; reset-lab.ps1 tested; runbook with guard rails

Cleanup / Restore

Snapshot: holodeck-07-complete

• Consolidate or delete the holodeck-07-pre-change set once holodeck-07-complete exists

• Decide whether Pod 2 (5.2.2) stays for the vcf-evolution labs

• Clear the -DeveloperMode variables that hold secrets from the session

Design Reflection (VCDX)

Lifecycle discipline in a nested lab mirrors the operational design chapter of a VCDX submission: recovery objectives (revert in minutes versus redeploy in hours), change control (one config per deployment, commands recorded verbatim), and the cost of state (snapshot growth, idempotent state, HoloRouter networking state). The 5.2 to 9.0 comparison is the version-evolution story panels expect you to tell without notes: what Cloud Builder and SDDC Manager did alone, what VCF Operations and the VCF Installer add, and what that changes in your availability, manageability and recoverability arguments.

Requirements

  • Consistent checkpoints of a nested VCF instance
  • Clean removal and redeploy from a recorded config
  • Measured recovery times per reset method
  • Documented 5.2 versus 9.0 architectural differences

Constraints

  • No Holodeck snapshot cmdlet; snapshots at the outer layer only
  • 1:1 config to deployment; config files never hand-edited
  • License Later: 90 days (9.0), 60 days (5.2); redeploy or license afterwards
  • One trunk port group per pod

Assumptions

  • Outer datastore headroom for two snapshot sets
  • Capacity for a second pod during Task 4
  • Recorded holodeck-02 deployment command available

Risks

  • Partial revert — Impact: vSAN and inventory divergence — Mitigation: whole-set revert with the instance off
  • Snapshot sprawl — Impact: datastore exhaustion — Mitigation: two-set limit and consolidation
  • Secrets in automation — Impact: token and depot credentials leaked — Mitigation: prompt or protected file
  • Unscheduled reset — Impact: lab destroyed mid-exercise — Mitigation: state check and lock file

Self-Assessment Discussion Prompts

  1. When is a redeploy preferable to a snapshot revert even though it costs hours?
  2. What does the 1:1 config rule protect, and how would you version the commands that produced each instance?
  3. Which 9.0 components would you present as the reason to upgrade a 5.2 estate, and which 5.2 behaviours disappear?
  4. How does the License Later window shape a training lab's reset calendar?

Extensions

Dual-site lifecycle

Repeat Tasks 1-3 on the dual-site pod from holodeck-06 and record the per-site Stop/Start and removal behaviour from two pwsh sessions.

harder

5.2.2 to 9.0 in-place upgrade

Keep Pod 2 and run the vcf-evolution-02 upgrade sequence on it (SDDC Manager, NSX, vCenter, ESX), timing each stage.

harder

Provision-only redeploy

Redeploy with -ProvisionOnly and complete bring-up manually from /holodeck-runtime/specs/ to see what the automated path hides.

harder

Offline depot automation

Stand up the Offline Depot appliance and switch the -DeveloperMode variables to the offline set.

same

⚠ Known Pitfalls

Expecting a Holodeck snapshot cmdlet COMMON
Problem: None is documented.
Resolution: Stop-HoloDeckInstance, then outer-layer snapshots, then Start-HoloDeckInstance.
Hand-editing /holodeck-runtime/config/<id>.json COMMON
Problem: Documentation says not to.
Resolution: Use New-/Get-/Import-HoloDeckConfig.
-ResetHoloRouter on every removal COMMON
Problem: Needless reboot and DNS loss.
Resolution: Reset only when the network plan changes.
Two pods on one trunk port group COMMON
Problem: Addressing collides.
Resolution: One port group and one HoloRouter per pod.
ESA on VCF 5.2.2 COMMON
Problem: OSA only on Holodeck for 5.2.
Resolution: Compare OSA to OSA, or OSA to ESA knowingly.

References

Was this page useful?
Type to search. ↑ ↓ to move, Enter to open, Esc to close.