Skip to content

Monitoring Configuration

This guide walks you through configuring LangSmith monitoring for DAIV. LangSmith provides comprehensive observability for your AI agents, including tracing, logging, and performance monitoring.

Prerequisites

Before configuring monitoring, ensure you have:

  • DAIV installed and running - Follow the installation guide first
  • LangSmith account - Create a free account at smith.langchain.com
  • LangSmith API key - Generated from your LangSmith dashboard

Step 1: Create LangSmith API Key

  1. Sign in to LangSmith:

  2. Go to smith.langchain.com

  3. Sign in with your account or create a new one

  4. Generate API Key:

  5. Navigate to Settings → API Keys

  6. Click Create API Key
  7. Name: DAIV Integration
  8. Description: API key for DAIV monitoring

  9. Click Create

  10. Copy the API Key:

  11. Important: Copy and save the API key immediately - you won't see it again
  12. The key format looks like: lsv2_pt_xxxxxxxxxxxxxxxxxxxxxxxx_yyyyyyyyyyyy

API Key Security

Store your API key securely. Never commit it to version control or share it publicly.

Step 2: Configure Environment Variables

Add your LangSmith configuration to DAIV's environment settings.

For Docker Compose Setup

Edit your docker-compose.yml file:

YAML
1
2
3
4
5
6
7
x-app-defaults: &x_app_default
  # ...
  environment:
    LANGSMITH_TRACING: true
    LANGSMITH_PROJECT: daiv-default
    LANGSMITH_API_KEY: lsv2_pt_xxxxxxxxxxxxxxxxxxxxxxxx_yyyyyyyyyyyy
  # ...

For Docker Swarm Setup

Environment configuration:

Bash
1
2
3
4
# LangSmith Monitoring
LANGSMITH_TRACING=true
LANGSMITH_PROJECT=daiv-production
LANGSMITH_API_KEY_FILE=/run/secrets/langsmith_api_key

Create Docker secret:

Bash
1
2
# Create secret for LangSmith API key
echo "lsv2_pt_xxxxxxxxxxxxxxxxxxxxxxxx_yyyyyyyyyyyy" | docker secret create langsmith_api_key -

Using EU Endpoint

If you're in Europe, you may want to use the EU endpoint (default is US):

Bash
1
LANGSMITH_ENDPOINT=https://eu.api.smith.langchain.com

Step 3: Configure Project Settings

Customize your LangSmith project settings for better organization.

Project Names

Use descriptive project names to organize your traces:

Bash
1
2
3
4
# For different environments
LANGSMITH_PROJECT=daiv-production    # Production environment
LANGSMITH_PROJECT=daiv-staging       # Staging environment
LANGSMITH_PROJECT=daiv-development   # Development environment

Step 4: Restart DAIV Services

Apply the new monitoring configuration by restarting DAIV.

For Docker Compose

Bash
1
2
3
4
5
# Restart all services
docker compose restart

# Or restart specific services
docker compose restart app worker

For Docker Swarm

Bash
1
2
# Update the stack with new configuration
docker stack deploy -c stack.yml daiv

Step 5: Verify Monitoring Setup

Test that LangSmith monitoring is working correctly.

  1. Generate Some Activity:

  2. Create a test issue in your repository with the daiv label

  3. Wait for DAIV to process the issue

  4. Or trigger any AI agent activity

  5. Check LangSmith Dashboard:

  6. Go to smith.langchain.com

  7. Navigate to your project (e.g., daiv-default)

  8. You should see traces appearing for agent executions

  9. Verify Trace Details:

  10. Click on any trace to see detailed execution steps

  11. Check for proper agent names, model calls, and timing information

Step 6: Dashboard and Analytics

Set up monitoring dashboards and alerts for your DAIV deployment.

Agent Metadata and Tags

Each DAIV agent automatically includes standardized metadata and tags for LangSmith tracing, making it easy to create dashboards and analyze performance:

Standard Tags

All agents include these tags in their traces:

Tag Description Example Values
Agent Name The specific agent type IssueAddressor, PipelineFixer, ReviewAddressor, CodebaseChat, PullRequestDescriber, CodeDescriber, SnippetReplacer, ImageURLExtractor, CodebaseSearch, PlanAndExecute
Client Slug The repository client identifier gitlab, github

Agent-Specific Metadata

Different agents include additional context-specific metadata:

Issue Addressor (IssueAddressor):

JSON
1
2
3
4
5
6
7
8
9
{
  "author": "username",
  "thread_id": "unique-thread-id",
  "project_id": 123,
  "source_repo_id": "group/repo",
  "source_ref": "main",
  "issue_id": 456,
  "repo_client": "gitlab"
}

Pipeline Fixer (PipelineFixer):

JSON
1
2
3
4
5
6
7
8
{
  "merge_request_id": 789,
  "job_id": 101112,
  "thread_id": "unique-thread-id",
  "source_repo_id": "group/repo",
  "source_ref": "feature-branch",
  "job_name": "build_and_test"
}

Review Addressor (ReviewAddressor):

JSON
1
2
3
4
5
6
7
8
{
  "merge_request_id": 789,
  "discussion_id": "abc123",
  "author": "reviewer-username",
  "thread_id": "unique-thread-id",
  "source_repo_id": "group/repo",
  "source_ref": "feature-branch"
}

Codebase Chat (CodebaseChat):

JSON
1
2
3
4
{
  "model_id": "DAIV",
  "chat_stream": true
}

Pull Request Describer (PullRequestDescriber):

JSON
1
2
3
{
  "thread_id": "unique-thread-id"
}

Creating Custom Dashboards

Use these tags and metadata to create focused dashboards:

By Agent Type: - Filter by tag: IssueAddressor to see all issue processing activity - Filter by tag: PipelineFixer to monitor CI/CD troubleshooting - Filter by tag: ReviewAddressor to track code review interactions

By Repository: - Filter by metadata: source_repo_id = "your-org/your-repo" - Group by repo_client to compare GitLab vs GitHub activity

By User Activity: - Filter by metadata: author = "username" to see user-specific interactions - Group by author to identify most active users

By Performance: - Monitor execution time by agent type - Track token usage patterns across different agents - Analyze success/failure rates by agent and repository

Setting Up Alerts

Configure alerts in LangSmith for: - High error rates (> 5%) - Slow response times (> 30 seconds) - Excessive token usage - Failed agent executions

Troubleshooting

Common Issues

No traces appearing in LangSmith: - Verify API key is correct and has proper permissions - Check that LANGSMITH_TRACING=true is set - Ensure network connectivity to LangSmith endpoints - Review application logs for authentication errors

Incomplete or missing trace data: - Verify project name matches in all configurations - Check that all required environment variables are set - Ensure Docker secrets are properly mounted (for Swarm deployments)

High costs or token usage: - Review trace filtering settings - Consider disabling tracing for development environments - Monitor token consumption patterns in LangSmith dashboard

Advanced Configuration

Sampling Configuration

Configure trace sampling to reduce costs while maintaining visibility:

Bash
1
2
# Sample 50% of traces (default: 100%)
LANGCHAIN_TRACING_SAMPLE_RATE=0.5

⏭️ Next Steps

For more detailed information about LangSmith features, visit the LangSmith documentation.