> ## Documentation Index
> Fetch the complete documentation index at: https://www.towbar.dev/llms.txt
> Use this file to discover all available pages before exploring further.

# Operate Towbar

> Find the right observation, incident, or diagnostic when something changes.

Towbar separates a desired configuration, a deployment attempt, and the state observed afterward. Start with the question you need to answer, then check the timestamp and scope of the result. A successful deployment does not guarantee that a workload is healthy now; a missing measurement does not mean zero usage.

| Question                          | Start here                                                                 | What it tells you                                                         |
| --------------------------------- | -------------------------------------------------------------------------- | ------------------------------------------------------------------------- |
| Is Towbar itself ready?           | [System and workload health](/docs/operate/health)                         | Control-plane checks, server checks, and current workload observations    |
| What changed over time?           | [Performance history](/docs/operate/performance)                           | Scout measurements, gaps, deployment markers, and capacity                |
| Did a release cause a regression? | [Deployment comparisons](/docs/deployment-comparisons)                     | Equal observation windows before and after two deployments                |
| Who needs to respond?             | [Alert rules](/docs/scout-alerts) and [incidents](/docs/operate/incidents) | Trigger conditions, active incidents, recovery, and notification outcomes |
| Is an image vulnerable?           | [Vulnerability scanning](/docs/vulnerability-scanning)                     | Findings for a specific immutable app image digest                        |
| Why did an operation fail?        | [Troubleshooting](/docs/troubleshooting)                                   | The first failed stage and its next diagnostic step                       |

## A practical investigation

1. Open the affected server, app, or resource and check its latest state and observation time.
2. Read the most recent deployment or operation result. Open its logs at the failing stage if it did not finish.
3. Compare the same time range on the [Performance](/docs/operate/performance) page. Gaps mean Towbar did not receive measurements for that interval.
4. If a rule fired, open the [incident](/docs/operate/incidents) to see the trigger, graph, and separate delivery results.
5. If the control plane itself is unavailable, run `sudo towbar doctor` on its host and follow [Troubleshooting](/docs/troubleshooting).

The [monitoring concepts](/docs/monitoring) page explains the difference between allocation, measured usage, and health. [Scout Agent](/docs/scout) supplies continuous host and container measurements; deployment and server checks still work without it.