[pgpool-general: 8283] Re: Problems taking node offline

[Jon SCHEWE]

I've setup a development cluster using pgpool 4.3.2 and postgresql 13. I've upgraded the pg_recovery extension to 1.4
and I'm able to get all backend nodes online using pcp_recovery_node.

Now I'm trying to test taking down a backend node.

template1=# show pool_nodes;
 node_id |         hostname         | port | status | pg_status | lb_weight |  role   | pg_role | select_cnt | load_balance_node | replication_delay | replication_state | replication_sync_state | last_status_change  
---------+--------------------------+------+--------+-----------+-----------+---------+---------+------------+-------------------+-------------------+-------------------+------------------------+---------------------
 0       | psql-dev-01.mgmt.bbn.com | 5432 | up     | up        | 0.333333  | standby | standby | 0          | true              | 0                 | streaming         | sync                   | 2022-07-01 16:34:16
 1       | psql-dev-02.mgmt.bbn.com | 5432 | up     | up        | 0.333333  | standby | standby | 0          | false             | 0                 | streaming         | potential              | 2022-07-01 16:37:19
 2       | psql-dev-03.mgmt.bbn.com | 5432 | up     | up        | 0.333333  | primary | primary | 6          | false             | 0                 |                   |                        | 2022-07-01 12:41:51
(3 rows)

Trying to take down the primary node:
[jschewe-adm at psql-dev-03 ~]$ pcp_detach_node -h psql.mgmt.bbn.com -p 9897 -U pgpool -g -n 2
Password: 
ERROR:  invalid degenerate backend request, node id : 2 status: [3] is not valid for failover

Why didn't this work?

Trying to take down a standby node:
[jschewe-adm at psql-dev-03 ~]$ pcp_detach_node -h psql.mgmt.bbn.com -p 9897 -U pgpool -g -n 1
pcp_detach_node -- Command Successful


template1=# show pool_nodes;
 node_id |         hostname         | port | status | pg_status | lb_weight |  role   | pg_role | select_cnt | load_balance_node | replication_delay | replication_state | replication_sync_state | last_status_change  
---------+--------------------------+------+--------+-----------+-----------+---------+---------+------------+-------------------+-------------------+-------------------+------------------------+---------------------
 0       | psql-dev-01.mgmt.bbn.com | 5432 | up     | up        | 0.333333  | standby | standby | 0          | false             | 0                 | streaming         | sync                   | 2022-07-01 16:34:16
 1       | psql-dev-02.mgmt.bbn.com | 5432 | up     | up        | 0.333333  | standby | standby | 0          | false             | 0                 | streaming         | potential              | 2022-07-01 16:37:19
 2       | psql-dev-03.mgmt.bbn.com | 5432 | up     | up        | 0.333333  | primary | primary | 6          | true              | 0                 |                   |                        | 2022-07-01 12:41:51
(3 rows)

Node is still up.

Nothing in the pgpool log other than a bunch of messages that say
2022-07-01 16:59:04.966: watchdog pid 311085: LOG:  new IPC connection received


Jon Schewe