We've noticed an interesting edge case where PubSub encountered a io.grpc.StatusRuntimeException for a messages in a minute long window. What I'm guessing happened here is that PubSub had a hiccup for a moment, however the messages that were attempted to be sent in that window were dropped.
We depend on both Logstash and Pubsub for an ingestion pipeline with guaranteed at least one delivery, so this edge case is quite concerning for us.
From what I can tell however, this failure is retry-able, so similarly to how the retry mechanism works on the Kafka output, can we get one for PubSub which will handle these edge cases?
We've noticed an interesting edge case where PubSub encountered a
io.grpc.StatusRuntimeExceptionfor a messages in a minute long window. What I'm guessing happened here is that PubSub had a hiccup for a moment, however the messages that were attempted to be sent in that window were dropped.We depend on both Logstash and Pubsub for an ingestion pipeline with guaranteed at least one delivery, so this edge case is quite concerning for us.
From what I can tell however, this failure is retry-able, so similarly to how the retry mechanism works on the Kafka output, can we get one for PubSub which will handle these edge cases?