PodWarden
User ManualPodWarden

Technical Support

AI-powered diagnostics and support ticket creation for your PodWarden cluster

Overview

The AI Support Assistant helps you investigate cluster issues and open support tickets without leaving PodWarden. It automatically collects diagnostics from your cluster, then lets you choose between AI-powered issue detection or a fast-track reporting flow.

Prerequisite: PodWarden must be connected to PodWarden Hub. Configure the connection in Settings → Hub Connection. Without a valid Hub API key the support panel cannot reach the AI analysis or ticket endpoints.

Opening the panel

Click the support icon in the top navigation bar to open the assistant. A badge on the icon shows the number of open tickets. The panel slides in from the left and stays open while you navigate — closing it does not discard your session.


Phase 1 — Diagnostics collection

When the panel opens, diagnostics collection starts automatically. You do not need to do anything during this phase.

The panel shows live progress as it gathers data from your cluster:

  • Cluster nodes and their status
  • Active deployments
  • Ingress rules
  • System messages (active health check alerts)
  • Pod logs and Kubernetes events (when you are viewing a specific deployment)

Each collected block appears as a collapsible item in the panel. Click any block to expand and read the raw data.

When collection finishes, the panel transitions to the choice step.


Phase 2 — Choose your path

After diagnostics complete, a text input appears along with two buttons:

  • Analyze with AI — runs the full AI analysis using your diagnostics and optional description
  • Report a Problem — lightweight fast-track flow for when you already know what the issue is

You can optionally type a description of what you are experiencing before choosing either path. The text carries forward into whichever flow you select.


AI Analysis path

Click Analyze with AI to have the assistant examine your diagnostics automatically.

  1. The AI scans the collected data and identifies issues (CrashLoopBackOff pods, DNS mismatches, missing ingress rules, etc.).
  2. Detected issues appear as cards. Select the one that matches what you are experiencing.
  3. The assistant may ask one or two clarifying questions to refine the ticket.
  4. A ticket preview appears with a generated title and body. You can edit the text directly before proceeding.
  5. Click Submit to move to the attachments step.

If the AI detects no matching issues, select Something else to describe the problem manually. The same clarifying questions (area and timing) guide the session to a ticket preview.

If the AI returns guidance rather than issue cards — for example, a known how-to or FAQ answer — review the answer and indicate whether it resolved your question. If it did not, you can continue to a ticket.


Fast-Track path

Click Report a Problem to skip AI analysis and go straight to ticket creation.

  1. If you already typed a description, it is used as-is. Otherwise, you are asked: "What are you experiencing?" (free text, required).
  2. Which area? — select the affected area: Deployment, Networking, Storage, Auth, UI, or Other.
  3. When did this start? — select: Just now, Today, After a change, or Ongoing.
  4. The AI rewrites your description into a structured ticket with a concise title and formatted body.
  5. The ticket preview appears with three options:
    • Edit — edit the title and body inline
    • Rewrite — send your edited text back to the AI for another pass (available up to 3 times per session)
    • Submit — proceed to the attachments step

After 3 rewrites, only Edit and Submit remain.


Attachments and submission

Before the ticket is created, the attachments step shows what will be included:

  • diagnostics.md — the full cluster diagnostics collected in Phase 1 (always attached)
  • pod-logs.txt and k8s-events.txt — included automatically when viewing a deployment
  • screenshot.jpg — optional; click Add Screenshot and capture your screen

Click Continue to Submit when you are ready.

The assistant checks for duplicate tickets before creating a new one. If similar open tickets are found, you can:

  • Update an existing ticket — adds your diagnostics and description as a new reply, and optionally reopens the ticket
  • Create new anyway — opens a separate ticket regardless

Once submitted, the panel shows the ticket number. Click it or go to View Tickets (link in the panel header) to open /support and see all your tickets.


Viewing tickets

URL: /support

The support page lists all tickets associated with your account. Each row shows the ticket number, title, status, and last update time.

URL: /support/[id]

The ticket detail page shows the full ticket body and the conversation thread with the support team. Ticket content is rendered as formatted markdown.


Resetting a session

While a session is active (during collection or AI interaction), a Clear button appears in the panel header. Clicking it and confirming discards the current session and its diagnostics. The next time you open the panel a fresh collection run starts automatically.

If a collection or analysis step fails, a Retry button appears in the panel. Click it to restart diagnostics from the beginning.