ES version: 126.96.36.199
.NET ES client version: 5.0.1
We have a setup of 3-node ES cluster running in Azure.
We have a catch-up subscription on the $all stream. We are projecting events to a document database and also keeping the checkpoint there.
When the ES node the subscriber is connected to crashes, the subscription cannot recover. After connecting to another ES node, we get subscription dropped event immediately after starting it, with reason CatchUpError and the exception listed below. The only option to move the subscription forward is to move the checkpoint to the end of the $all stream. What may be the reason for this behavior?
EventStore.ClientAPI.Exceptions.ServerErrorException: Unexpected error on server: Log record at actual pos 144459231 has too large length: 1852143173 bytes, while limit is 16777216 bytes. In chunk #1488-1488 (chunk-001488.000000).
at EventStore.ClientAPI.Internal.EventStoreNodeConnection.ReadAllEventsForwardAsync(Position position, Int32 maxCount, Boolean resolveLinkTos, UserCredentials userCredentials)
at EventStore.ClientAPI.EventStoreAllCatchUpSubscription.ReadEventsInternalAsync(IEventStoreConnection connection, Boolean resolveLinkTos, UserCredentials userCredentials, Nullable`1 lastCommitPosition)