citus_rebalance_start() ERROR

[diogoaraujodtx]

Hi guys!

I am using a citus cluster with 2 workers. At this point, I want to add a new worker and redistribute the tables among the 3 workers. I used citus_rebalance_start() function but it returns an error (ERROR: connection to the remote node localhost:5432 failed with the following error: fe_sendauth: no password supplied). I have already tried to change pg_hba.conf file and still doesn't work. Can someone help me?

Thanks

[JelteF]

what does your pg_hba.conf look like now?

[JelteF]

The student table on the worker is not one of the shards. If you want to see the shards you can set:

SET citus.show_shards_for_app_name_prefixes = '*';

And if you then look at the existing tables you can see the underlying shard tables. These are named something like student_123456

[JelteF]

Also could you share how you solved your problem?

[diogoaraujodtx]

right, but if table student is distributed, how can count of rows in a worker return the initial table size (100)? Should't be a value less than 100 resulting by the sum of rows of tables student_shardid in that worker?

[diogoaraujodtx]

sure, I solve my problem changing line host all all 127.0.0.1/32 scram-sha-256 to host all all 127.0.0.1/32 trust in the pg_ha.conf file

[JelteF]

The result is 100 on all workers, because when you query student you don't query the shards directly and the worker will act like a coordinator in the sense that it will send queries to all workers. If you query student_123456 you will get the number of entries in that shard.

[JelteF]

See here for more details: https://www.citusdata.com/blog/2022/06/17/citus-11-goes-fully-open-source/#any-node