# Curio Market troubleshooting

This page targets the common “deal is stuck” / “indexing is failing” cases seen in support.

Before proceeding, collect the basics:

* [Collect debug info](/troubleshooting/collect-debug-info.md)
* deal UUID + piece CID
* failing task IDs from the UI

***

## 1) Deal stuck at indexing / “CheckIndex” errors

What CheckIndex is:

* Curio runs task-based indexing and health checks; a recurring task checks for missing indexes / announcements and schedules follow-up work.
* Implementation reference: `tasks/indexing/task_check_indexes.go` (task `CheckIndex`).

If you see errors in CheckIndex:

* Determine whether indexing is still progressing (tasks completing) vs stalled.
* Many “indexing failures” are actually DB health/latency problems. Verify Yugabyte health first.

### Common error: `duplicate key value violates unique constraint ... (SQLSTATE 23505)`

What it often means:

* Concurrent retries inserted the same identity row.

What to do:

1. Collect: deal UUID, piece CID, task ID(s), and the full error span.
2. Confirm whether the failing rows keep reappearing or are transient.
3. If stuck, restart only the market/indexing layer process (not the whole cluster).

***

## 2) IPNI failures

Symptoms:

* Indexing seems OK locally, but IPNI publishing is failing.

Checklist:

* Outbound HTTPS works from Curio host.
* The “public URL” for your provider is correct (reachable from the internet).
* DNS + firewall allow inbound connectivity to your HTTP server.

Temporary mitigation:

* If you need onboarding to proceed while IPNI is unstable, you may temporarily disable IPNI publishing (only if supported by your release/config). Document and keep track of the change so it isn’t forgotten.

***

## 3) ParkPiece: `no data URL found for piece_id`

What it means:

* Curio cannot find the configured URL+headers for fetching the piece data.

Most common causes:

* The URL was never registered, or it was registered on a different node than the one running ParkPiece.
* Ingest layer is not enabled on the node you think.

What to collect:

* deal UUID, piece CID
* whether offline/online deal
* the exact command used to add the URL (if online)

***

## 4) Retrieval looks broken, but indexing is fine

Checklist:

* Ensure unsealed data is available on at least one node that can serve retrieval.
* Confirm the Curio HTTP server is enabled and reachable.
* If behind a reverse proxy, confirm TLS delegation mode matches your config.

See also:

* [Curio HTTP Server](/curio-market/curio-http-server.md)


---

# Agent Instructions: Querying This Documentation

If you need additional information that is not directly available in this page, you can query the documentation dynamically by asking a question.

Perform an HTTP GET request on the current page URL with the `ask` query parameter:

```
GET https://docs.curiostorage.org/curio-market/troubleshooting.md?ask=<question>
```

The question should be specific, self-contained, and written in natural language.
The response will contain a direct answer to the question and relevant excerpts and sources from the documentation.

Use this mechanism when the answer is not explicitly present in the current page, you need clarification or additional context, or you want to retrieve related documentation sections.