# DB_HELD_BY_OTHER_PROGRAM

> a program that is not croft has the warehouse open

**Error** · category **coordination** (another process holds the database or the asset) · exit **4** (busy (lock or lease wait exceeded))

A program that is not croft has the warehouse open.

A DuckDB file can be open for writing in one process only. Another program has warehouse.duckdb open: the DuckDB
CLI or UI, DBeaver or another GUI, a notebook, or an app that keeps its own connection. croft waited (off a
terminal at most 90 s) and gave up. The message names the program and its PID (details.holder).

Nothing was changed; retryable is true once the program lets go.

What to do:
- Ask the user to close the program the message names (or the query tab that holds the file), then run the command
  again. Do not kill another program's process yourself.
- So it does not happen again:
  - a GUI should open the read copy, warehouse.read.duckdb, never warehouse.duckdb: "readCopy": true in croft.json
    keeps it current ([`croft docs read-copy`](https://croft.zabaca.com/docs/guides/read-copy.md));
  - an app should read through @zabaca/croft/read, which opens the file only per query, or through croft serve
    ([`croft docs serve`](https://croft.zabaca.com/docs/guides/serve.md)).

The same page is in the CLI: `croft docs DB_HELD_BY_OTHER_PROGRAM`.

Source: https://croft.zabaca.com/docs/reference/errors/db-held-by-other-program/