We were experiencing connectivity issues on our feature flags nodes due to resource exhaustion on those nodes. After identifying this bottleneck, we applied changes that resolved this resource exhaustion and the service recovered.
Posted Oct 30, 2025 - 11:10 UTC
Identified
We've rolled back some recent caching changes and have observed latency dropping to more appropriate levels but continue to look at improving this further.
Posted Oct 30, 2025 - 06:10 UTC
Investigating
We're looking into increased error rates and latency on the feature flags service in US. We rolled back some hardware changes for the service and the errors have mostly subsided. We're still looking into increased latencies currently.