Affected
While upgrading kubernetes nodes, we had a blip of 30 seconds where the load balancer was not responding. This could have caused a few network timeouts on the client side that should have been retried by the SDKs.
