Latency and error rates across all systems have returned to normal levels. The major API services - Maps, Geocoding, and Directions - failed over to redundant capacity quickly and impact was minimal. Mapbox.com and Upload API processing, which were impacted more, are now fully restored. We plan to share information shortly regarding preventative maintenance that will bypass future issues with our database provider and permanently resolve this issue.
Posted Sep 23, 2015 - 19:22 UTC
Update
Upload processing is now resumed and previously delayed jobs are being processed.
Posted Sep 23, 2015 - 16:43 UTC
Update
Failover is complete for the majority of requests. We are continuing to monitor as remaining traffic fails over. Users may continue to experience errors when editing maps on Mapbox.com and edits may take longer than usual to propagate globally. Upload processing is currently paused and will continue working when service is restored.
Posted Sep 23, 2015 - 14:15 UTC
Identified
Our upstream database provider is experiencing a large scale availability event similar to the event that took place on September 20th. We are redirecting traffic away from the problem.
Posted Sep 23, 2015 - 13:52 UTC
Investigating
We're experiencing elevated latency on our Maps API and are currently looking into the issue.