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

# Troubleshooting

> Symptom-first index — find your error message, get the fix.

Organised by where the symptom shows up. If you are not sure which page applies, scan the table below for the message you actually saw.

<Note>
  Before anything else: `make application-ps` shows which containers are up, and `docker compose logs -f <service>` shows why one is not. Most issues resolve from those two commands.
</Note>

## Symptom index

| What you see                                     | Page                               |
| ------------------------------------------------ | ---------------------------------- |
| Stack will not start, containers restart-looping | [Common issues](common-issues)     |
| `.env` not found, missing or empty secrets       | [Common issues](common-issues)     |
| Port already allocated                           | [Common issues](common-issues)     |
| `ModuleNotFoundError: No module named 'apps'`    | [Common issues](common-issues)     |
| Connection refused to FerretDB                   | [Common issues](common-issues)     |
| Call connects but there is no audio              | [Voice and audio](voice-and-audio) |
| Audio in one direction only                      | [Voice and audio](voice-and-audio) |
| Speech is garbled, sped up, or robotic           | [Voice and audio](voice-and-audio) |
| Agent will not stop talking when interrupted     | [Voice and audio](voice-and-audio) |
| Greeting is cut off, or hold messages never play | [Voice and audio](voice-and-audio) |
| No transcript or recording after the call        | [Voice and audio](voice-and-audio) |
| Provider cannot reach `/answer`                  | [Telephony](telephony)             |
| `/answer` returns `400`                          | [Telephony](telephony)             |
| Number will not attach or detach                 | [Telephony](telephony)             |
| Recording never arrives from the provider        | [Telephony](telephony)             |
| Campaign stuck in a state, or paused itself      | [Campaigns](campaigns)             |
| Circuit breaker tripped                          | [Campaigns](campaigns)             |
| Slot acquisition timeout, phone pool exhausted   | [Campaigns](campaigns)             |
| Worker not picking up jobs                       | [Campaigns](campaigns)             |
| WSS fails behind a reverse proxy                 | [Deployment](deployment)           |
| GPU not visible to a container                   | [Deployment](deployment)           |
| Disk fills during a model build                  | [Deployment](deployment)           |
| Volume permission errors                         | [Deployment](deployment)           |

## The pages

* [**Common issues**](common-issues) — startup, environment, ports, imports, health checks.
* [**Voice and audio**](voice-and-audio) — anything you can hear, or cannot.
* [**Telephony**](telephony) — the provider boundary: webhooks, applications, numbers, recordings.
* [**Campaigns**](campaigns) — the orchestrator, ARQ worker, circuit breaker, and concurrency slots.
* [**Deployment**](deployment) — TLS, proxies, GPUs, disk, and volumes.

## Still stuck

Collect these before opening an issue: the failing command, `make application-ps`, the last 100 log lines from the affected container, and your `.env` with **every secret redacted**.

## Related

* [Daily operations](../operator/operations) — health endpoints and which logs matter
* [Ports and defaults](../../developer/reference/ports-and-defaults)