Troubleshooting

Resolve common issues when onboarding and operating models

Use this guide to diagnose and resolve frequent model onboarding and readiness issues.

Troubleshooting flow

Common issues

Model onboarding fails

Possible causes: invalid URI/path, expired credentials, unsupported source details.

Resolution: revalidate source configuration, confirm access, and resubmit.

Model not visible in list

Possible causes: filters are too restrictive, indexing delay, or insufficient permissions.

Resolution: clear filters, search by exact name, and verify role access.

Verification or scan remains incomplete

Possible causes: ingestion still running or scan pipeline interrupted.

Resolution: wait for ingestion completion, then trigger scan again and review logs.

Poor benchmark results

Possible causes: unsuitable infrastructure, incorrect configuration, or noisy baseline.

Resolution: rerun with controlled settings, compare against baseline models, and adjust target cluster.

Fast diagnostic checklist

• Confirm model source details are valid.
• Validate tags/metadata for search discoverability.
• Review verification and scan status.
• Compare latest benchmark run with previous stable run.
• Escalate with run identifiers and timestamps for support.

Need more help?

• Revisit Quickstart for baseline setup.
• Use Model Security and Verification for trust controls.