PostgreSQL14 & Citus11 using workers for 2 different clusters doesn't work

Hi,
I use PostgreSQL 14.6 and Citus 11.1.5
I have 2 cluster named cluster1 and cluster2, and 3 citus worker node.

I want to distribute my tables in cluster1 and cluster2 to the same citus workers.

The nodes in both clusters are as follows:
postgres=# select * from pg_dist_node;
nodeid | groupid | nodename | nodeport | noderack | hasmetadata | isactive | noderole | nodecluster | metadatasynced | shouldhaveshards
--------+---------+----------------+----------+----------+-------------+----------+----------+-------------+----------------+------------------
87 | 86 | citus1-primary | 5433 | default | t | t | primary | default | t | t
88 | 87 | citus2-primary | 5433 | default | t | t | primary | default | t | t
89 | 88 | citus3-primary | 5433 | default | t | t | primary | default | t | t
(3 rows)

But I got an error. Here is my scenerio:

1. I created distribute table in cluster1 named cluster1_t1,cluster1_t2
2. I created distribute table in cluster2 named cluster2_t1,cluster2_t2
3. I created distribute table in cluster1 named cluster1_t3 and I got en error:

SELECT create_distributed_table('cluster1_t3', 'c1');
ERROR: Node with group id 86 for shard placement 102880 does not exist
CONTEXT: while executing command on citus1-primary:5433

I am using this scenerio in PostgreSQL 12 & Citus10 and PostgreSQL 14 & Citus10. But I can't use PostgreSQL 14 & Citus 11

Is this a bug or a new feature?

Thanks.

Start by reproducing the reported setup with PostgreSQL 14, Citus 11.1.5, two clusters, and the three workers, then inspect create_distributed_table and the pg_dist_node state around the failing shard placement. Compare the behavior with the PostgreSQL 12/Citus 10 and PostgreSQL 14/Citus 10 scenarios; done means determining whether this is a regression, an unsupported configuration, or a documented feature limitation.

Written by the indexing model from the issue text.