j5ik2o / akka-persistence-s3 Goto Github PK

Similar to https://doc.akka.io/docs/akka-persistence-cassandra/current/cleanup.html

Hi!

First of all let me tell I think it's great idea to store snapshots in S3! I've tried using it but found a few rough edges, so there'll be more issues created ;) I'd be happy to create PRs for that myself, just let me know.

So, why I'm really here:
I think the resources.conf should be called reference.conf to be automatically included.

Or at least there should be a reference.conf including defaults like

j5ik2o {
  s3-snapshot-store {
    class = "com.github.j5ik2o.akka.persistence.s3.snapshot.S3SnapshotStore"
    plugin-dispatcher = "akka.actor.default-dispatcher"
  }
}

Akka 2.6.16 introduced a new Durable State Store. It's still marked as "may change" and documentation around the implementation is non-existent. Nevertheless it seems it's mostly about implementing those traits. Testkit already exists too.

Seems like a good fit for this plugin.

In private def snapshotMetadatas (https://github.com/j5ik2o/akka-persistence-s3/blob/master/snapshot/src/main/scala/com/github/j5ik2o/akka/persistence/s3/snapshot/S3SnapshotStore.scala#L270) the bucketNameResolver is used directly instead of calling private def resolveBucketName (https://github.com/j5ik2o/akka-persistence-s3/blob/master/snapshot/src/main/scala/com/github/j5ik2o/akka/persistence/s3/snapshot/S3SnapshotStore.scala#L82).

This leads to override def loadAsync (https://github.com/j5ik2o/akka-persistence-s3/blob/master/snapshot/src/main/scala/com/github/j5ik2o/akka/persistence/s3/snapshot/S3SnapshotStore.scala#L134) always using the bucketNameResolver instead of the config provided bucketName