I can't remove the nodes in the cluster

Anyone know what is going on, what happens is that I want to remove some nodes that really are no longer part of the cluster but in admintools and in the MC it considers them as part of the cluster. When I try to remove them from admintools I get this error

[dbadmin@AZURE ~]$ /opt/vertica/bin/adminTools -t db_remove_node -p 'password' -d DB_name -s 172.18.220.111

Error removing node(s) from database. ['Unable to mark node \'v_db_name_node0002\' ephemeral: Severity: ROLLBACK, Message: Node "v_db_name_node0002" does not exist, Sqlstate: 42704, Routine: AlterNode, Fi le: /data/jenkins/workspace/RE-ReleaseBuilds/RE-Miner/server/vertica/Commands/DDL.cpp, Line: 7813, Error Code: 4136, SQL: VALTER NODE v_db_name_node0002 IS EPHEMERAL;\' ']

[dbadmin@AZURE ~]$ /opt/vertica/bin/adminTools -t list_allnodes
Node | Host | State | Version | DB
---------------------+----------------+-------+------------------+----------
v_db_name_node0001 | 172.18.220.135 | UP | vertica-24.2.0.1 | DB_name
v_db_name_node0002 | 172.18.220.111 | DOWN | vertica-24.2.0.1 | DB_name
v_db_name_node0003 | 172.18.220.86 | DOWN | vertica-24.2.0.1 | DB_name

as can be seen, the two nodes to be removed are in down status, the version used is 24.2.0-1

querying from vsql actually only returns a single node, which is correct

select * from nodes;

node_name | node_id | node_state | is_primary | is_readonly | node_address | node_address_family | export_address | export_address_family | catalog_path | node_type | is_ephemeral | standing_in_for | subcluster_name | sandbox | last_msg_from_node_at | node_down_since | build_info
---------------------+-------------------+------------+------------+-------------+----------------+---------------------+----------------+-----------------------+--------------------------------------------------------------+-----------+--------------+-----------------+--------------------+---------+-------------------------------+-----------------+----------------------------------------------------
v_db_name_node0001 | 45035996273704992 | UP | t | f | 172.18.220.135 | ipv4 | 172.18.220.135 | ipv4 | /data01/catalog/ModeloBI/v_db_name_node0001_catalog/Catalog | PERMANENT | f | | default_subcluster | | 2024-08-14 18:41:29.005455-05 | | v24.2.0-1-d52dc2eff25ea7e221904aded25063b05974ebab
(1 row)

what is wrong and how to fix it, it is worth mentioning that it is an Eon mode deployed in Azure.

Thanks for your answers.

Answers

  • SruthiASruthiA Administrator

    Vertica should be able access node via SSH so that it can be removed. If instance no longer exists, it is tough. Please open a support case as it requires many manual steps to remove nodes.

This discussion has been closed.