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

# Service and Axottle Doctor

> How to check the panel server, fix typical crashes, and prepare a diagnostic archive for support.

Axottle Doctor helps to check the server on which the panel is running. Run it if the panel does not open, the database is unavailable, there are errors in TLS or before calling for support.

\[!NOTE] requires access to the panel server console to run. Use `sudo axottle`. to restore services.

## Doctor's launch

Open the interactive menu:

```bash theme={null}
sudo axottle
```

Select **Doctor**. The check will start automatically and may take some time: The Doctor consistently checks the dashboard, database, system services, free space, network, DNS, certificates and connected components.

A brief summary stays at the top of the screen, and you can scroll through the detailed results.

* `↑` and `↓` or mouse wheel – scroll the list;
* `Page Up` and `Page Down` – go to the page;
* `Home` and `End`: Beginning and End of Results
* `Esc` is a return to the Doctor menu.

The results use five states:

| State of | What means |
| - | - |
| `OK` | Test passed by |
| `Warning` | Service works, but requires the attention of |
| `Failed` | Found a problem that can interfere with the operation of the panel |
| `Skipped` | Verification does not apply to this installation |
| `Unknown` | There was no access or data to check. |

Find the problem checker and read the lines **Found**, **Why** and **Next**: it indicates what is discovered, why it is important and what can be done.

> [!TIP]
>
> If you only want a text result without an interactive menu, do it. `axottle doctor`Even if the problems are found, the Doctor will complete the diagnosis and show all available results.

## Safe recovery

**Attempt safe recovery** is suitable for typical failures when a database, dashboard, background services or web server has stopped. The Doctor will try to start or restart the desired service, update the already configured certificate if necessary, and then repeat all the checks.

Before starting, consider the following:

* Active connections to the restarted service may be interrupted for a short time.
* Doctor does not change database passwords or edit DNS records.
* If the action fails, it will remain on the list marked `failed` and explained.
* After recovery, the performed actions and the results of the re-check are shown on the screen.

If you do not want Doctor to restart services, use only **Run diagnostics**.

## Place on the disk

A full drive isn't slow, it's a stop: PostgreSQL can't write checkpoint, panics, and goes into a restart cycle that writes even more to the journal, and by this point, the API is no longer available, so you have to sort it out locally.

Screen. **Place on the disk** menu `sudo axottle` It categorizes the space and divides it into two groups: what can be cleaned and what is protected. Database volumes are never touched.

Actions on `Enter`: clear everything, clear what is older than 7 days, older than 30 days, or show where it lies (catalogues and largest files). Before deletion, you show how much will be released and how many files are affected, and the deletion itself requires an explicit `y`. After cleaning, the disk is recalculated.

Logs of working containers are reset in place, not deleted: Docker keeps them open, and deleting the file would not return space before restarting the container.

> [!NOTE]
>
> The installer and the update check the vacant space before the start of work. It's critical for the upgrade: it removes the base dump, and the dump that didn't have enough space is a cropped dump, from which they will roll back. The local backup for the same reason does not start without stock: you need a maximum of 1 GB and double the size of the previous archive.

A background health sample follows the same: an incident starts with a balance of less than 10% and less than 5%, the thresholds are common with Doctor. Log size restrictions are now in every compose file service, in `/etc/docker/daemon.json` and in journald, so Remnawave logs and proxy no longer eat the disk in a week.

## Support bundle

A support bundle is an archive with audit results and logs that can be attached to a support call.

### Choose a period.

In **Log period**, press `Enter` until the appropriate period appears:

* 1 hour;
* 6 hours;
* 24 hours;
* 7 days;
* 30 days;
* All the available history.

Usually, a period that starts a little earlier than the first signs of a problem is enough. For example, if the failure occurred two hours ago, select 6 hours. The longer the period, the longer the archive is collected, and the larger the file will be.

### Collect and verify the archive

1. Select **Collect support bundle**.
2. Wait until the collection is complete. Don't close the terminal while you're creating the archive.
3. Select **Send bundle to support**.
4. Before submitting, select **Show archive contents** to see brief information about the archive.
5. For a full check, select **Save locally**, open the saved archive and browse `manifest.json`.
6. Return to **Send bundle to support** and select **Send to support**.

The archive can include panel logs, databases, web server, background services, agent, Remnawave, Docker, installer, updates and deletions, as well as information about the status of the server. If any source on the server is not available, this will be noted in the archive manifest.

Passwords, keys, tokens, and other common secrets are masked before being saved, and the identifiers used to match the circulation are stored in anonymized form.

> [!WARNING]
>
> Before you transfer the archive outside the organization, still look at its composition. In non-standard logs, you can find data that is specific to your infrastructure.

If you can't send the file, the Doctor will save it to the server and show you the path to the file, and you can attach it manually to the file.

## How to choose an action

| Situation | What to choose | |
| - | - | - |
| Panel is working, you need a preventive check | **Run diagnostics** | |
| Panel or database stopped | **Run diagnostics**, then **Attempt safe recovery** | |
| Need to contact | Select period → **Collect support bundle** → View and submit the | archive |
| | **View last diagnosis** | |
| Sending in support is not available | **Save bundle locally** | |

## Recommended rules

| Frequency | Action | |
| - | - | - |
| Weekly | Launch Doctor and disassemble `Warning` and `Failed` | |
| After | Update Check dashboard, database, ports and TLS | |
| Before appealing in support of | Collect bundle for the period of the | incident |
| Monthly | Check vacant space and old local bundles | |

Backups are serviced separately - see [Backups of the panel](https://docs.axottle.com/en/operations/backups.md). The order of updating is described on the page [Update panel](https://docs.axottle.com/en/operations/updates.md).

## Verification

After the diagnosis, make sure that:

* There are no unexplained `Failed` in the summary.
* Each `Warning` has an understandable following action.
* after safe recovery, the problem services have been re-checked;
* The created support bundle is opened, and its manifesto corresponds to the selected period and the expected sources.

## Rollback

Normal diagnostics don't change anything. The locally saved support bundle can be deleted like a regular file.

Safe Recovery can restart services or update an existing certificate. If the situation worsens after it, do not repeat the action blindly: save the bundle, check the activity log and restore the service according to its standard instructions. For emergency order, use the [Incident Response page](https://docs.axottle.com/en/operations/incident-response.md).

> [accordion] Doctors can't check the database
>
> Make sure the command is running on the panel server. Select **Attempt safe recovery**. If the base does not rise, check the status of the PostgreSQL container or the `postgresql` service, then collect bundle - it will get the available database and installer logs.

> [accordion] Sending the archive failed.
>
> Select **Save bundle locally** and attach the saved file manually, and check also the DNS, outbound connection and support server address.

> [accordion] The results are many Skipped or Unknown.
>
> `Skipped` usually means that the component is not used in this installation.`Unknown` means that the Doctor was unable to retrieve the data - often due to insufficient rights or an inaccessible service.Run `sudo axottle` and repeat the check.

Related pages: [Response to incidents](https://docs.axottle.com/en/operations/incident-response.md), [Escalation](https://docs.axottle.com/en/troubleshooting/escalation.md), [CLI axottle](https://docs.axottle.com/en/reference/cli.md).


This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.