LaunchDarkly Outage History

LaunchDarkly experienced a major incident where the flag dashboard failed to load, affecting the web interface for 56 minutes. The issue was identified and a fix was implemented, with monitoring confirming successful resolution. Users were unable to access the flag dashboard during this period.

LaunchDarkly experienced issues with Slack notifications for segment approvals, while email and in-app notifications continued to function normally. The service implemented a fix and confirmed Slack notifications were restored after 1.1 hours of disruption.

LaunchDarkly experienced a 6-minute incident where event ingestion delays impacted Experimentation results and Guarded Rollouts functionality. The root cause was quickly identified and a fix was deployed to restore normal operations. No data was lost during the incident, and all results and events remained intact throughout the disruption.

LaunchDarkly experienced a 1.2-hour delay in processing Session Replay and Error event data, causing a temporary lag in data appearing in the UI while incoming data was queued. The issue was identified and resolved with all queued events successfully processed. No data was lost during the incident.

LaunchDarkly experienced elevated streaming initialization error rates in the APAC region for 30 minutes, caused by an infrastructure issue. Customers using streaming server-side SDKs experienced longer than normal initialization times or timeouts when connecting to the service. The issue was resolved by implementing a fix that restored normal streaming SDK connectivity.

LaunchDarkly experienced delays in experiment and multi-armed bandit results, with data delayed by 1-2 hours during the incident. The service identified and implemented a fix to resolve the processing backlog. All experiment data was fully caught up within 41 minutes with no data loss.

LaunchDarkly experienced elevated streaming errors in the Europe region for 6.2 hours, caused by an infrastructure issue that affected the Streaming API. Customers using streaming server-side SDKs experienced longer than normal initialization times or timeouts when connecting to the service. The issue was resolved after implementing a fix that restored normal streaming SDK connectivity.

LaunchDarkly experienced a 6.4-hour incident that caused delays in experimentation results processing and data warehouse export functionality. The service identified the root cause and implemented a fix, then worked through a backlog of delayed data processing. The issue was fully resolved when all experimentation results and data warehouse exports were caught up.

LaunchDarkly experienced a 45-minute delay in streaming data export delivery caused by latency issues in their Pub/Sub export system. Customers using streaming data export experienced multi-minute lags in receiving their data, though no data was lost during the incident. The issue was resolved by addressing the Pub/Sub latency and clearing the accumulated export backlog.

LaunchDarkly experienced an issue where editing flag variations between experiment iterations caused sample ratio mismatches (SRMs) or empty experiment results for some customers. The service implemented a fix to prevent the issue in new experiments and then validated a complete resolution for all affected experiments. The incident lasted 27.9 hours and impacted only a small subset of experimentation customers.