Amplitude Outage History

Amplitude experienced slower than normal processing in their data ingestion pipelines for 4.5 hours. The issue affected the Data Processing component, causing delays in realtime data processing. The incident was resolved after implementing a fix and monitoring the results.

Amplitude experienced an ingestion lag in their batch pipeline affecting one partition of their data processing system for 33 minutes. The incident caused delays in data ingestion but no data was lost during the event. The issue was resolved after implementing a fix and monitoring the results.

Amplitude's data processing systems experienced delays for 6.3 hours due to an AWS infrastructure issue, causing delayed metrics, higher latencies for the experiment evaluation API, and slower cohort syncs to third-party destinations. Data ingestion continued normally with no data loss during the incident. The issue was resolved through collaboration between AWS and Amplitude teams.

The Amplitude web application in the US region was inaccessible for 38 minutes due to a misconfigured feature flag, while the EU region remained operational. Users experienced errors when trying to load Amplitude web pages, though data ingestion continued normally without any data loss. The issue was resolved by reverting the problematic configuration change.

Amplitude experienced high latency in Event Streaming starting at 12:20 PT on April 15th, causing delayed event data delivery to third-party streaming destinations for all customers over a 3-hour period. While ingestion and data processing systems continued working normally with no data loss, customers experienced delays in receiving event data at their downstream destinations. The incident was resolved after 2.1 hours, with streaming largely recovering and fewer than 1% of events still experiencing delays before full resolution.

Amplitude's Guides and Surveys feature incorrectly added a large number of event properties at 8:45 AM PDT, causing projects that were pushed over the 2000 event property limit to be unable to add new event properties or use them for filtering and creating cohorts. The team resolved the incident after 5.3 hours by deleting the erroneous event properties and restoring normal functionality, with post-processing scheduled to recover any legitimate event properties that were blocked during the outage period.

Amplitude's EU data collection endpoint (api.eu.amplitude.com) experienced service disruptions for 2.5 hours, causing customers to encounter errors when sending event data and making data ingestion temporarily unavailable. The team implemented auto-scaling measures and applied a fix to restore normal operations. No permanent data loss occurred for events with proper SDK retry logic configured.

Amplitude experienced a 5-hour authentication issue that prevented users from signing in with external tools including Ampli CLI, MCP Server, and Notion/Atlassian chart previews. A fix was implemented and the incident was resolved, though some users needed to re-authenticate these components to restore full functionality.

Amplitude experienced a site-wide outage lasting 2 hours that affected Web Reporting and Data Processing components, with users encountering 503 error messages when accessing the app. The issue was identified and fixed with services gradually restored, while analytics event collection continued operating normally throughout the incident. The incident was fully resolved with no data loss and all systems returned to operational status.

Amplitude experienced a major issue where exposure charts failed to load for some feature flags and experiment analysis pages displayed "user property not available" and "Invalid user property gp:" errors. The problem affected users' ability to view experiment data and analytics for 1.4 hours. The issue was identified, fixed, and rolled out to restore full functionality to the affected charts and analysis pages.