Omni Analytics Outage History

Omni Analytics experienced a 3.3-hour outage where users encountered 403 "Invalid bearer token" errors when using MCP integrations through callbacks.omniapp.co (Claude, ChatGPT, etc.), affecting the EastUsa and Ireland regions. The issue was caused by a regression in OAuth token handling for MCP tool calls that was introduced in a recent deployment, preventing users from performing actions like listing models or running queries despite successful initial connections. The service was restored after deploying a patch to fix the authentication token handling.

Omni Analytics experienced a major outage for 1.5 hours where pages returned "Application DB Error" messages, affecting the EastUsa and Ireland regions. The issue was caused by a scheduled update to an internal database table that resulted in full page errors. The team implemented a fix to safely complete the database update and put measures in place to prevent similar errors from future updates.

Some users experienced difficulty accessing their Omni Analytics instance due to DNS resolution issues affecting certain ISPs, while Omni's infrastructure remained healthy. The issue was isolated to specific internet service providers and did not originate from Omni's systems. The incident was resolved after 3.8 hours when the primary DNS provider fixed their underlying issue, with users who switched ISPs reporting immediate resolution.

The Omni Analytics service experienced significant slowness and error messages affecting users in the EastUsa and Ireland regions for 41 minutes. The issue was caused by a recent deployment change that degraded performance. The team resolved the incident by rolling back to a previous deployment, restoring normal performance levels.