You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
For Akka.Cluster.Sharding 1.3.2-beta54, I've noticed that if a non-shard node or proxy node is selected to become the singleton shard coordinator due to failure, it will not start until that new node is restarted and the coordinator is moved back to a node with a shard region started. The two cases I can reproduce are:
Movement to a node with no shard region or proxy started. This fails with an error of "Error [Cannot find serializer with id [13]", which matches ClusterShardingMessageSerializer in the code. This would make sense, since it's probably not loaded in the first place on that node.
Movement to a node with a shard region proxy started. This shows as the coordinator being moved to the shard, but no persistence recovery is attempted and no new registrations can be completed. No exceptions seem to be thrown, however.
As an example, I have two nodes with shard regions started on them, and a single node that either has a shard proxy or none. If crash the shard nodes and bring them back up, the coordinator is moved to the single node, and the behavior occurs as above.
Is this a valid use case I'm attempting?
The text was updated successfully, but these errors were encountered:
@marrick66 When specifying settings on start of cluster sharding region, you may specify the role, that all nodes containing regions of that type, should have:
I had roles set for the cluster, but not the shards. Adding them to the shard configuration prevents the non-shard nodes from attempting to become the coordinator. Thanks for the help, I appreciate it.
For Akka.Cluster.Sharding 1.3.2-beta54, I've noticed that if a non-shard node or proxy node is selected to become the singleton shard coordinator due to failure, it will not start until that new node is restarted and the coordinator is moved back to a node with a shard region started. The two cases I can reproduce are:
Movement to a node with no shard region or proxy started. This fails with an error of "Error [Cannot find serializer with id [13]", which matches ClusterShardingMessageSerializer in the code. This would make sense, since it's probably not loaded in the first place on that node.
Movement to a node with a shard region proxy started. This shows as the coordinator being moved to the shard, but no persistence recovery is attempted and no new registrations can be completed. No exceptions seem to be thrown, however.
As an example, I have two nodes with shard regions started on them, and a single node that either has a shard proxy or none. If crash the shard nodes and bring them back up, the coordinator is moved to the single node, and the behavior occurs as above.
Is this a valid use case I'm attempting?
The text was updated successfully, but these errors were encountered: