From 18:35 to 20:35 UTC, customers connected to the us-east-2 region may have experienced HTTP 502 errors. This was caused by a latent memory leak that caused some hosts to be terminated for being out of memory. The team began investigating at 18:53 UTC and deployed mitigations by adding capacity in the region. We deployed a fix for the underlying root cause by 20:35 UTC. We have already taken steps to add monitoring on the cause of the memory leak to enable us to react proactively if there is a future recurrence.
Posted Sep 09, 2024 - 22:45 UTC
This incident affected: Oso Cloud (Authorization APIs).