Hey,

Using SE.Redis v2.1.0.

We added a new shard to our redis cluster, and mostly things were uneventful, except for our backend worker machines receiving EXECABORT Transaction discarded because of previous errors for a certain code path.

This code path uses transactions, and was working fine with 2 shards prior. After adding the 3rd shard, these errors seemed to be constant for a code path.

A restart of our worker application immediately fixed the issue.

I'm thinking perhaps the transaction code did not detect/know to refresh the cluster slots map?