Troubleshooting

Common issues and solutions for Kaisar AI Ops.

Overview

This section covers:

• FAQ - Frequently asked questions

• Known Issues - Current limitations

• Support - How to get help

Quick Troubleshooting

Cannot Log In

Symptoms: Login page shows error or redirects back

Solutions:

1. Clear browser cache and cookies

2. Try incognito/private mode

3. Verify credentials with admin

4. Check if MFA is required

5. Try password reset

Experiment Won't Start

Symptoms: Experiment stuck in "pending" status

Solutions:

1. Check resource quotas

2. Verify compute resources are available

3. Review experiment configuration

4. Check cluster capacity

5. View experiment logs for errors

Slow Dashboard Loading

Symptoms: Dashboard takes long to load

Solutions:

1. Check internet connection

2. Clear browser cache

3. Reduce number of displayed items

4. Check system status page

5. Try different browser

API Requests Failing

Symptoms: 401, 403, or 500 errors

Solutions:

1. Verify API token is valid

2. Check token permissions

3. Review rate limits

4. Check API endpoint URL

5. Verify request format

Common Error Messages

"Quota Exceeded"

Cause: Resource limit reached

Solution:

1. Check current usage

2. Clean up unused resources

3. Request quota increase

4. Optimize resource allocation

"Permission Denied"

Cause: Insufficient permissions

Solution:

1. Check your role

2. Request access from admin

3. Verify resource sharing settings

4. Check organization membership

"Resource Not Found"

Cause: Invalid ID or deleted resource

Solution:

1. Verify resource ID

2. Check if resource was deleted

3. Ensure you have access

4. Try listing resources first

Performance Issues

Slow Experiment Training

Possible Causes:

• Inefficient data loading

• Suboptimal batch size

• CPU bottleneck

• Network I/O issues

Solutions:

1. Profile your code

2. Optimize data pipeline

3. Increase batch size

4. Use data caching

5. Check GPU utilization

High Memory Usage

Possible Causes:

• Large batch size

• Memory leaks

• Inefficient model architecture

Solutions:

1. Reduce batch size

2. Use gradient accumulation

3. Enable mixed precision training

4. Profile memory usage

5. Clear unused variables

Integration Issues

Authentication Service Failing

Solutions:

1. Verify Authentication Service is running

2. Check client configuration

3. Review realm settings

4. Verify redirect URIs

5. Check SSL certificates

Storage Connection Failed

Solutions:

1. Verify credentials

2. Check bucket/container exists

3. Review IAM permissions

4. Test network connectivity

5. Verify endpoint URL

Getting Help

Self-Service Resources

• FAQ - Common questions

• User Guide - Feature documentation

• API Reference - API documentation

Contact Support

• Support Portal - Submit tickets

• Email: [email protected]

• Slack: #kaisar-support

Community

• GitHub Discussions

• Stack Overflow (tag: kaisar-ai-ops)

• Community Forum

Diagnostic Tools

Health Check

Check system health:

curl https://ai.kaisar.io/health

API Verification

Verify API access:

curl -X GET https://ai.kaisar.io/api/v1/auth/verify \
  -H "Authorization: Bearer YOUR_TOKEN"

Network Test

Test connectivity:

ping ai.kaisar.io
traceroute ai.kaisar.io

Best Practices

• ✅ Check system status before reporting issues

• ✅ Collect error messages and logs

• ✅ Try basic troubleshooting first

• ✅ Document steps to reproduce

• ✅ Include relevant screenshots

• ✅ Provide system information

Next Steps

• Review FAQ for common questions

• Check Known Issues

• Contact Support if needed