Skip to content

Gen3 Admin Advanced Management

GT AI OS Release edited this page Jun 15, 2026 · 2 revisions

Advanced Management

Start Here

  1. Confirm GT AI OS is already installed and you can sign in to the Control Panel and tenant app—this page is for day-two operations, not first-time onboarding (see Getting Started).
  2. Replace <namespace> below with your install namespace (for example gt-ai-os-prod).
  3. Use gt-ai-os-admin validate when you need a health check after deploy or upgrade.
  4. Use gt-ai-os-admin update (or manage-ai-os.sh) to move an existing namespace to a newer release.
  5. Use gt-ai-os-admin bootstrap-creds when you need operator login credentials again.

Why this matters

The guided installer keeps its finish message short for new operators. After install, you need reliable cluster and release commands for validation, upgrades, credential recovery, and pod inspection. This page collects those operator tools in one place so advanced tasks do not depend on memorizing install-time terminal output.

Details

These commands run on the RKE2 server (or any host with a working kubeconfig and gt-ai-os-admin on PATH from a prior install-ai-os.sh run). They manage application releases inside an existing namespace—they do not reinstall RKE2.

Cluster pod view (k9s)

Interactive terminal UI for pods, deployments, services, and related cluster objects:

k9s

Scope to your GT AI OS namespace when you know it:

k9s -n <namespace>

From the install operator menu (gt-ai-os-operator):

  • Menu option 5) Cluster view (k9s)
  • Or: gt-ai-os-operator --k9s --namespace <namespace>

k9s is recommended in the operator package but may need a separate install on some hosts (brew install k9s on macOS dev clusters). Set KUBECONFIG to your RKE2 kubeconfig before launching.

Validate deployment health

Runs deploy validation and smoke checks for the namespace:

gt-ai-os-admin validate --namespace <namespace>

Shortcuts:

  • gt-ai-os-operator --validate --namespace <namespace>
  • Install operator menu 3) Validate & sign in

Use after install, upgrade, or when troubleshooting pod readiness before declaring the deployment healthy.

Upgrade to a newer release

Upgrades application manifests and images in the existing namespace:

gt-ai-os-admin update --namespace <namespace> --to <release-tag> --yes

Omit --to when you want the tool to prompt for or resolve the target tag interactively.

gt-ai-os-admin update:

  • Loads stored instance state for the namespace
  • Fetches release assets for the target tag
  • Refreshes Helm values and manifests
  • Runs deploy validation and smoke checks (unless disabled)
  • Saves updated state

It upgrades apps in place; it does not replace the Kubernetes distribution.

For a guided flow, use the day-two management script (below) or gt-ai-os-operator --update-interactive.

See also Updates and Self-Hosted Updating for release-channel context.

Day-two management script

Interactive menu for namespace install, upgrade, local-dev hostname changes, namespace deletion, and orphaned-namespace cleanup:

sudo bash "/var/lib/gt-ai-os/operator-scripts/manage-ai-os.sh"

Non-interactive upgrade subcommand (same flow as menu option 2):

sudo bash "/var/lib/gt-ai-os/operator-scripts/manage-ai-os.sh" upgrade

Optional environment variables to reduce prompts:

Variable Purpose
NAMESPACE or GT_AI_OS_NAMESPACE Target namespace
TO_VERSION or GT_AI_OS_UPGRADE_TO_VERSION Target release tag

Shared operator auth and instance state live under /var/lib/gt-ai-os/admin. Any sudo user on the host shares GitHub and Cloudflare profiles stored there.

manage-ai-os.sh expects gt-ai-os-admin on PATH from a prior install. It does not bootstrap a host with no cluster.

Retrieve login credentials again

Print bootstrap operator credentials for the namespace (Control Panel and related sign-in material):

gt-ai-os-admin bootstrap-creds --namespace <namespace>

Use when credentials were lost after install or when onboarding another operator who needs the initial login path. Rotate passwords through normal account policy after recovery when your security posture requires it.

Roll back after a bad upgrade

When a release regression is confirmed:

gt-ai-os-admin rollback --yes --namespace <namespace> --to <older-tag>
gt-ai-os-admin validate --namespace <namespace>

Confirm URLs and sign-in with gt-ai-os-admin report --namespace <namespace>.

When to use which tool

Need Tool
First install on a fresh RKE2 host install-ai-os.sh / gt-ai-os-operator install flow — Getting Started
Health check gt-ai-os-admin validate
Release upgrade gt-ai-os-admin update or manage-ai-os.sh upgrade
Credential recovery gt-ai-os-admin bootstrap-creds
Live pod inspection k9s or gt-ai-os-operator --k9s
Interactive namespace lifecycle menu manage-ai-os.sh

Related pages

Clone this wiki locally