> ## Documentation Index > Fetch the complete documentation index at: https://docs.insight.nobly.dk/llms.txt > Use this file to discover all available pages before exploring further. # Monitoring & administration > Track workflow instances, read the history timeline, handle failures with error nodes and replay, and manage environment variables and secrets. All administration pages live under **Settings → Admin → Workflow** and require administrator access. ## Workflow Definitions The **Workflow Definitions** page lists every workflow with its name, ID, status badge (**Draft** / **Published**), anchor type, and creation and publish timestamps. Filter by name, status, or anchor type; click a row to open it in the [designer](/workflows/designer). ## Workflow Instances The **Workflow Instances** page searches across every run of every workflow. Filter by: * **Workflow** (multi-select) * **Status** — `Running`, `Failed`, or `Completed` * **Anchor** — document or WorkView, optionally a specific ID (to answer "what workflows touched this document?") * **Date range** Results show the workflow, the current step (by display name), status, timestamps, and the anchored document or object, loading more as you scroll. ### Instance detail Opening an instance shows everything about the run: * **Variables** — the instance's current variable bag * **Live tasks** — task transitions currently waiting for a user * **Timers** — armed timer transitions and when they fire * **History** — the full append-only audit timeline Each history event records what happened, where (node and action), who caused it (the user, or the service account for automated steps), and when: | Event | Meaning | | ------------------------------------------------------- | ----------------------------------------------------------------------------- | | `instance.created` | The run started — records the trigger, seed variables, and definition version | | `node.entered` / `node.exited` | The instance arrived at / left a step | | `transition.fired` | An edge was taken | | `action.started` / `action.completed` / `action.failed` | An action ran | | `instance.completed` | A terminal node was reached (or `Workflow.End` ran) | | `instance.failed` | The run halted on a failure with no error node to absorb it | | `instance.replayed` | An administrator replayed a failed run | ## Handling failures When an action fails: 1. The instance's variables roll back to their state at node entry 2. The instance routes to the **error node** — the node-level override if set, otherwise the workflow-level one — with two variables describing the failure: `instance._error.action` (which action failed) and `instance._error.message` (why) 3. If no error node is configured, the instance is marked **Failed** A failed instance is never lost: from the instances list you can **retry** it, or **replay** it — optionally correcting variables first when bad data caused the failure. The engine does not retry automatically, so design error nodes for the failures you expect (for example, a task transition asking someone to fix a keyword and try again) and use replay for the rest. ## Environment variables and secrets The **Workflow Variables** page manages the configuration workflows reference: * **Environment variables** (`env.`) — plain-text settings such as an SMTP host, shared by all workflows * **Secrets** (`secrets.`) — encrypted at rest, decrypted only at the moment an action parameter is resolved, never written to history, and audit-logged on access. Secrets cannot be used in guards or trigger filters. Keeping host names and credentials here — rather than hardcoded in definitions — lets the same definition move between environments unchanged. ## Troubleshooting | Symptom | Likely cause | What to do | | ------------------------------------ | -------------------------------------------------------------------------------------------------------------------------------------------- | --------------------------------------------------------------------------------------------------- | | Events happen but no instance starts | The workflow is a draft, the item is outside the scope, the trigger filter doesn't match, or an active instance already exists for that item | Publish the workflow; check scope and filter against the event; look for the existing instance | | An instance sits on a node forever | A non-terminal node with no passing transition — guards are evaluated once on node entry | Check the node's transitions and guards; validation warns about nodes with no outgoing transitions | | An instance is `Failed` | An action failed and no error node was configured | Open the history to find the `action.failed` event, fix the cause, then replay | | A user can't see a task | The task's **Visible to** groups don't include them | Check the transition's visibility groups against the user's group membership | | An email action fails | SMTP settings missing or wrong | Confirm the `env.*` / `secrets.*` values the action references exist on the Workflow Variables page | ## Where to read next Error nodes, validation, and the publish flow these pages monitor. Platform-wide audit, permissions, and observability.