zylon-ai/private-gpt · error · ConnectionError

One or more database connections failed: \n{errors}

Error message

One or more database connections failed: \n{errors}

What it means

ConnectionError raised by the validate_sql closure after running check_connection() on every DatabaseQueryGenerator concurrently with asyncio.gather(return_exceptions=True). Any connection that returned an error string or raised is collected as '- DB {i} connection error: ...' lines and joined into the message.

Source

Thrown at private_gpt/components/tools/builders/database_query_builder.py:240

                    enable_functions=sql_artifact.enable_functions,
                    enable_procedures=sql_artifact.enable_procedures,
                    description=sql_artifact.description,
                    batch_size=self.settings.database_query.batch_size,
                    timeout_seconds=self.settings.database_query.timeout_seconds,
                    max_mb_result=self.settings.database_query.max_mb_result,
                    cache=self.cache,
                )
                for sql_artifact in sql_artifacts
            ]
            validations = [gen.check_connection() for gen in query_gen]
            results = list(await asyncio.gather(*validations, return_exceptions=True))
            errors = [
                f"- DB {i} connection error: {result!s}\n"
                for i, result in enumerate(results)
                if isinstance(result, str | Exception)
            ]
            if errors:
                raise ConnectionError(
                    "One or more database connections failed: \n" + "; ".join(errors),
                )

        @auto_resolve_media_blocks(blob_visibility=blob_visibility)
        async def execute_sql(query: str) -> list[ResultContentBlockType]:
            additional_context: str | None = await asyncio.to_thread(
                self._get_additional_context,
                chat_history,
            )
            query_gen = []
            try:
                query_gen = [
                    database_query_generator_cls(
                        connection_string=sql_artifact.connection_string,
                        ssl=sql_artifact.ssl,
                        schemas=sql_artifact.schemas,
                        enable_tables=sql_artifact.enable_tables,
                        enable_views=sql_artifact.enable_views,

View on GitHub (pinned to 4a030776a3)

Solutions

1. Read the per-DB lines in the message to identify which index failed and why (index order matches sql_artifacts order).
2. Test connectivity with the same connection string using a plain client (psql/sqlcmd) or a small asyncpg/sqlalchemy script from the same host.
3. Verify credentials, host, port, and ssl flag in each artifact's connection_string; install the driver extra for every engine used (e.g. uv sync --inexact --extra database-postgres).
4. If failures are transient (network blip), retry after fixing network/DNS.

import asyncio

async def probe_connections(sql_artifacts):
    results = await asyncio.gather(*(gen.check_connection() for gen in gens), return_exceptions=True)
    return [r for r in results if isinstance(r, (str, Exception)) and r not in (None, True)]

try:
    await validate_sql()
except ConnectionError as e:
    errors = str(e).splitlines()
    # map '- DB {i} connection error' lines back to sql_artifacts[i] and report which DB failed
    raise

When it happens

Trigger: One or more sql_artifacts have unreachable or misconfigured connection strings: wrong host/port, bad credentials, missing SSL config, or an engine whose driver was not installed; check_connection either raises or returns an error string.

Common situations: Database behind a firewall/VPC not reachable from the app; expired or wrong password; ssl=True against a server with no TLS; using a postgres connection string in an environment where only database-mysql extra was installed.

Related errors

• Failed to inspect the database schema.
• Failed to inspect the database schema.
• Failed to inspect the database schema.
• DB2 database query dependencies are not installed. Install w
• Default LLM model '{model_id}' could not be initialized: {e}

AI-assisted analysis of zylon-ai/private-gpt@4a030776a3 (2026-08-15). Data as JSON: /api/errors/dbc0ab7c6119c632. Report an issue: GitHub.