RememberEntitiesShardStore$UpdateDone is leaking over ExtractorAdapter

[mehmetsalgar]

Hi,

I am using

HashCodeNoEnvelopeMessageExtractor

which is only expected my sealed Events and crashes by receiving

RememberEntitiesShardStore$UpdateDone

with Pekko 1.6.0

java.lang.ClassCastException: class org.apache.pekko.cluster.sharding.internal.RememberEntitiesShardStore$UpdateDone cannot be cast to class xxx.TravelSM$TravelSMEvent (org.apache.pekko.cluster.sharding.internal.RememberEntitiesShardStore$UpdateDone and xxx.TravelSM$TravelSMEvent are in unnamed module of loader 'app')
at xxx.TravelSMGuardian$$anon$2.entityId(TravelSMGuardian.scala:364)
at org.apache.pekko.cluster.sharding.typed.internal.ExtractorAdapter.entityId(ClusterShardingImpl.scala:67)
at org.apache.pekko.cluster.sharding.typed.internal.ClusterShardingImpl$$anonfun$1.isDefinedAt(ClusterShardingImpl.scala:177)
at org.apache.pekko.cluster.sharding.Shard$$anonfun$idle$1.applyOrElse(Shard.scala:637)

I think the problem lies in

ExtractorAdapter

override def entityId(message: Any): String = {
message match {
case ShardingEnvelope(entityId, _) => entityId
case ClassicStartEntity(entityId) => entityId
case msg => delegate.entityId(msg.asInstanceOf[E])
}
}

does not forsee that it will receivce events from

RememberEntitiesShardStore

definetely HashCodeNoEnvelopeMessageExtractor does also not.

Is this event leaking to ExtractorAdapter or it is wanted the ExtractorAdapter receive this...

with this configuration

pekko {
....

cluster {
shutdown-after-unsuccessful-join-seed-nodes = 60s
sharding {
remember-entities = on
...
}
....
}

[pjfanning]

I created #3411

[Philippus]

Looks like these unsolved (>3 years old) issues in akka: akka/akka-core#31100 (also reported by @mehmetsalgar) and akka/akka-core#30641.

[mehmetsalgar]

I totally unregistered akka/akka-core#31100 but I have a vague memory that problem was consequence was somebody refactoring another part of Akka Persistence and because of that this Event leaking to this layer and it was fixed in Akka Persistence....

But it is from 2022 and I have a very vague memory of it, that was the reason I was asking, this Event is leaking or is it reasonable to deal with this event at this layer?

As I can see from that issue, I could not test the fix in Snapshot version and most probably real life interfered by following that issue.....

Ok, this the PR fixing leaking Issue for another Event to the message extractor

johanandren/akka@ff4295d

As I said, I am not able to say, it is a wanted behaviour that 'UpdateDone' is landing in MessageExtractor or it is a leakage

Isn't it more approriate that this Event dealt here

`EventSourcedBehavior
.receiveSignal {
case (state, org.apache.pekko.persistence.typed.RecoveryCompleted) =>
context.log.debug("Recovery completed, current state: {}", state)

              case _ =>
                context.log.debug("Another Event")
            }`