Troubleshooting

Common issues and their solutions when using LLM Client.

Installation Issues

Rust Installation Fails

Problem: Can't install Rust or cargo commands not found

Solutions:

1. Ensure you have a C compiler installed:

   • Linux: sudo apt install build-essential
   • macOS: xcode-select --install
   • Windows: Install Visual Studio Build Tools

2. Try the official installer:

   curl --proto '=https' --tlsv1.2 -sSf https://sh.rustup.rs | sh

3. Add cargo to PATH:

   source $HOME/.cargo/env

Build Errors

Problem: cargo build --release fails

Solutions:

1. Update Rust:

   rustup update

2. Clean and rebuild:

   cargo clean
   cargo build --release

3. Check for missing dependencies in error messages

Provider Issues

"No providers configured"

Problem: Getting error when trying to use lc

Solution:

# Add a provider first
lc providers add openai https://api.openai.com/v1

# Verify it's added
lc providers list

"Provider not found"

Problem: Provider name not recognized

Solutions:

1. Check exact spelling:

   lc providers list

2. Ensure provider is added:

   lc providers add <name> <url>

"Invalid endpoint URL"

Problem: Provider endpoint rejected

Solutions:

1. Include protocol: https:// or http://
2. Don't include trailing paths unless needed
3. For custom endpoints, use -m and -c flags

API Key Issues

"No API key found"

Problem: Provider requires API key

Solution:

lc keys add <provider>
# Enter key when prompted

"Authentication failed"

Problem: API key rejected

Solutions:

1. Verify key is correct:

   lc keys remove <provider>
   lc keys add <provider>

2. Check if provider needs custom headers:

   # For Claude
   lc providers headers claude add x-api-key <key>
   lc providers headers claude add anthropic-version 2023-06-01

3. Ensure you have API credits/quota

Model Issues

"Model not found"

Problem: Specified model doesn't exist

Solutions:

1. List available models:

   lc providers models <provider>
   # or
   lc models

2. Use exact model name from the list

3. For HF router, use format: model:provider

"Context length exceeded"

Problem: Input too long for model

Solutions:

1. Use a model with larger context:

   lc models --ctx 128k

2. Reduce input length

3. Split into multiple prompts

Vector Database Issues

"Database not found"

Problem: Vector database doesn't exist

Solutions:

1. Check available databases:

   lc vectors list

2. Create database by adding content:

   lc embed -m text-embedding-3-small -v <name> "content"

"Dimension mismatch"

Problem: Embedding dimensions don't match

Solutions:

1. Check database model:

   lc vectors info <database>

2. Use the same model for all operations

3. Delete and recreate with consistent model:

   lc vectors delete <database>
   lc embed -m <model> -v <database> "content"

"No similar content found"

Problem: Similarity search returns nothing

Solutions:

1. Verify database has content:

   lc vectors info <database>

2. Try different search terms

3. Check if content is relevant

Chat Issues

"Session not found"

Problem: Can't continue chat session

Solutions:

1. List recent sessions:

   lc logs recent

2. Use correct session ID:

   lc chat -m <model> --cid <session-id>

"Chat history lost"

Problem: Previous messages not remembered

Solutions:

1. Ensure you're in chat mode:

   lc chat -m <model>

2. Don't use /clear unless you want to reset

3. Check logs database:

   lc logs stats

Performance Issues

Slow Response Times

Solutions:

1. Use a faster model:

   lc -m gpt-3.5-turbo "prompt"

2. Check network connection

3. Try a different provider

High Token Usage

Solutions:

1. Use concise prompts
2. Set up system prompts for consistency
3. Use smaller models when appropriate

Sync Issues

"Sync failed"

Problem: Can't sync configuration

Solutions:

1. Check provider configuration:

   lc sync configure s3 show

2. Verify credentials:

   lc sync configure s3 setup

3. Check network/firewall settings

"Decryption failed"

Problem: Can't decrypt synced files

Solution: Use the same password used for encryption

Proxy Issues

Proxy -h Conflict

Problem: Using -h flag with proxy command doesn't work as expected

Cause: The -h flag conflicts between --host and --help options

Solutions:

1. Use full flag names to avoid ambiguity:

   # Instead of: lc proxy -h 0.0.0.0
   lc proxy --host 0.0.0.0

2. Use --help instead of -h for help:

   lc proxy --help

Reference: See Proxy Command Documentation for all available flags

Web Chat Proxy Port Issues

Problem: "Port already in use" or "Address already in use" errors

Solutions:

1. Check what's using the port:

   netstat -tlnp | grep :8080
   # or on macOS
   lsof -i :8080

2. Use a different port:

   lc web-chat-proxy start anthropic --port 3000

3. Stop existing proxy servers:

   lc web-chat-proxy list
   lc web-chat-proxy stop anthropic

4. Kill process using the port (if needed):

   # Replace PID with actual process ID from netstat/lsof
   kill -9 <PID>

Reference: See Web Chat Proxy Documentation for more details

Sync Provider Authentication Errors

Problem: "Authentication failed" when syncing with cloud providers

Solutions by Provider:

AWS:

1. Check AWS credentials:

   cat ~/.aws/credentials

2. Set environment variables:

   export AWS_ACCESS_KEY_ID=your-key
   export AWS_SECRET_ACCESS_KEY=your-secret

3. Verify IAM permissions for S3 bucket access

Azure:

1. Login to Azure CLI:

   az login

2. Check current account:

   az account show

GCP:

1. Authenticate with service account:

   gcloud auth activate-service-account --key-file=key.json

2. Or login interactively:

   gcloud auth login

General Steps:

1. Reconfigure the provider:

   lc sync configure <provider>

2. Test connectivity:

   lc sync providers

3. Check network/firewall settings

Reference: See Sync Command Documentation for provider-specific setup

Debug Mode

For detailed error information:

# Enable debug logging
export RUST_LOG=debug
lc -m gpt-4 "test prompt"

Getting Help

If you're still having issues:

1. Check the FAQ
2. Search GitHub Issues
3. Create a new issue with:
   • Error message
   • Steps to reproduce
   • System information
   • Debug logs

Common Error Messages

Error	Meaning	Solution
"No providers configured"	No providers added	Add a provider
"API request failed"	Network or API error	Check connection and API key
"Model not found"	Invalid model name	Use exact model name
"Rate limit exceeded"	Too many requests	Wait and retry
"Invalid API key"	Wrong or expired key	Update API key
"Context length exceeded"	Input too long	Use shorter input or larger model