[bug] Git config --local fails on bare repositories causing complete repository indexing failure in v4.7.0

[shaharkedar]

Describe the bug

Sourcebot v4.7.0 seems to have a critical bug that prevents all repository indexing from working. The issue occurs in the repository cloning and configuration process where:

1. Repositories are correctly cloned as bare repositories using git clone --bare
2. Later in the process, the code attempts to read git configuration using GitConfigScope.local (which translates to git config --local)
3. The --local flag is incompatible with bare repositories and fails with: fatal: --local can only be used inside a git repository
4. This causes all repository indexing jobs to fail after 3 retry attempts
5. No repositories can be indexed, making search functionality completely non-functional

Expected Behavior
Repositories should be successfully cloned and indexed, with logs showing:
• Cloned [repository] in X.XXs
• Indexing [repository]...
• Indexed [repository] in X.XXs
• Repo index job for repo [repository] completed

Actual Behavior
Repository indexing fails with the following error pattern:

[backend] | 2025-09-19T10:32:56.025Z error: [repo-manager] Failed to sync repository github.com/sourcebot-dev/sourcebot (id: 1), attempt 1/3. Sleeping for 60s... Error: Error: Failed to set git config /data/.sourcebot/repos/1: fatal: --local can only be used inside a git repository

To reproduce

1. Create a basic config.json file:

{
    "$schema": "https://raw.githubusercontent.com/sourcebot-dev/sourcebot/main/schemas/v3/index.json",
    "connections": {
        "test-connection": {
            "type": "github",
            "repos": [
                "sourcebot-dev/sourcebot"
            ]
        }
    }
}

2. Create a docker-compose.yml with Sourcebot v4.7.0:

services:
  postgres:
    image: postgres:16
    environment:
      POSTGRES_DB: sourcebot
      POSTGRES_USER: postgres
      POSTGRES_PASSWORD: postgres
      POSTGRES_HOST_AUTH_METHOD: trust
    volumes:
      - postgres_data:/var/lib/postgresql/data
    healthcheck:
      test: ["CMD-SHELL", "pg_isready -U postgres"]
      interval: 5s
      timeout: 5s
      retries: 5

  redis:
    image: redis:7-alpine
    volumes:
      - redis_data:/data
    healthcheck:
      test: ["CMD", "redis-cli", "ping"]
      interval: 5s
      timeout: 5s
      retries: 5

  sourcebot:
    image: ghcr.io/sourcebot-dev/sourcebot:latest  # v4.7.0
    depends_on:
      postgres:
        condition: service_healthy
      redis:
        condition: service_healthy
    ports:
      - "3000:3000"
    volumes:
      - .:/data
    environment:
      - CONFIG_PATH=/data/config.json
      - DATABASE_URL=postgresql://postgres:postgres@postgres:5432/sourcebot
      - REDIS_URL=redis://redis:6379
      - DATA_CACHE_DIR=/data/.sourcebot
      - DATABASE_EMBEDDED=false
      - SOURCEBOT_LOG_LEVEL=debug

volumes:
  postgres_data:
  redis_data:

3. Start the services and monitor the logs.

Sourcebot deployment information

Environment
• Host OS: macOS (also affects Linux deployments)
• Docker: Docker Compose v2+
• Deployment Method: Docker containers with separate PostgreSQL and Redis services

Affected Versions
• Sourcebot: v4.7.0 (latest as of 2025-09-19)
• Status: Critical regression bug

Working Versions
• Sourcebot: v4.6.0 and earlier versions work correctly

Additional information

No response

[msukkari]

hey @shaharkedar thanks for reporting!

I followed your repro steps and was unable to repro this issue. Does this happen to you even on a fresh cache?

[brendan-kellam]

@shaharkedar Are you hitting this on a fresh Sourcebot install against v4.7.0? Or only when upgrading from v4.6.0 -> 4.7.0

[igloo15]

I am also getting this error on a fresh install of ghcr.io/sourcebot-dev/sourcebot:latest as of today

[brendan-kellam]

Hey @igloo15 I haven't tried since but was unable to reproduce this locally - I'll try again today. Do you mind sharing a bit about your deployment and the environment you are deploying in?

[igloo15]

Sure I am running docker inside WSL - Ubuntu on Windows 11. I was able to solve the issue I think by creating /root/.gitconfig with the following contents

[safe]
  directory = *

which I found from a separate issue that was some what related.

[brendan-kellam]

Hmm interesting, yea I see #168 - I'm assuming this is some Docker on WSL specific issue potentially.

@shaharkedar could you try the same .gitconfig workaround and see if it works?

[igloo15]

I will say that even with this change I got a lot of failures and even with debug output on I get no good errors. I am not sure if its the number of repos I am syncing 998 or something to do with the file system. I have moved to testing this entirely on an Amazon EC2 instance of Ubuntu so no longer testing on WSL.

Typically the first sync works great but subsequent syncs fail 95% of the time with no good error message.

[p856c ](<url>/#/tasks/p856cty0cilr1f6dfo10ngl4t?log=1)[backend] | 2025-10-06T13:37:35.922Z �[32minfo�[39m: [repo-manager] Repo index job for repo <repo-name> (id: 937, jobId: 2955) failed with error: Error: Failed to fetch repository: /data/.sourcebot/repos/937. Reason: fatal: not a git repository (or any parent up to mount point /)
[p856c ](<url>/#/tasks/p856cty0cilr1f6dfo10ngl4t?log=1)[backend] | Stopping at filesystem boundary (GIT_DISCOVERY_ACROSS_FILESYSTEM not set).
[p856c ](<url>/#/tasks/p856cty0cilr1f6dfo10ngl4t?log=1)[backend] |

[brendan-kellam]

@igloo15 this second issue with subsequent sync failures: are you hitting this on the EC2 instance or on WSL? Do you have that same .gitconfig set in the container?

[igloo15]

I have the same .gitconfig set in the container and I am running it all in a docker container inside an EC2 instance of Ubuntu. I am also using a separate postgres and redis not the ones in the container.

[brendan-kellam]

Gotcha thanks. I'm looking into this, will have more tomorrow

[igloo15]

If this helps at all here is the the docker compose I am using

services:
  sourcebot:
      image: "ghcr.io/sourcebot-dev/sourcebot:latest"
      container_name: sourcebot
      volumes:
        - ./data/:/data
        - ./.gitconfig:/root/.gitconfig
      environment:
        - GITLAB_TOKEN=${GITLAB_READ_TOKEN}
        - CONFIG_PATH=/data/config.json
        - DATABASE_URL=postgres://${POSTGRES_ADMIN_USR}:${POSTGRES_ADMIN_PWD}@<db-url>:5432/sourcebot?sslmode=disable
        - REDIS_URL=redis://<redis-url>:6379
        - DATABASE_EMBEDDED=false
        - SOURCEBOT_LOG_LEVEL=debug
        - FORCE_ENABLE_ANONYMOUS_ACCESS=true

and my config.json

{
    "$schema": "https://raw.githubusercontent.com/sourcebot-dev/sourcebot/main/schemas/v3/index.json",
    "connections": {
        // comments are supported 
        "gitlab-connection": {
            "type": "gitlab",
            "url": "<local-gitlab-url>",
            "token": {
                "env": "GITLAB_TOKEN"
            },
            "all": true,
            "exclude": {
                "forks": true,
                "archived": true,
                "userOwnedProjects": true
            },
            "revisions": {
                "branches": [
                    "main",
                    "master",
                    "develop"
                ]
            }
        }
    },
    "models": [
        {
            "provider": "openai-compatible",
            "baseUrl": "<local-litellm-url>",
            "model": "claude-3-5-sonnet",
            "displayName": "Claude Sonnet 3.5",
            "token": {
                "env": "LLM_API_KEY"
            }
        }
    ],
    "settings": {
        //12 hours
        "reindexIntervalMs": 43200000
    }
}