[pgpool-general: 8128] Re: PGPool version 4.2.3 3 nodes cluster: all nodes marked as standby after restart, but system seems to work as expected

[Bo Peng]

Hello,

> Hi all,
> we have a 3 nodes cluster of PG12 with PGPool 4.2.3 on CentOS 7.
> 
> After a cold restart of all the nodes, we noticed that all nodes are marked
> as standby, but the real PG servers are correctly running as 1 Primary and
> 2 Standby and recovery is working fine.

At startup, Pgpool-II will try to find the primary node.
Could you share your pgpool logs?

> "Show pool nodes" output:
>  node_id |    hostname    | port | status | lb_weight |  role   |
> select_cnt | load_balance_node | replication_delay | replication_state |
> replication_sync_state | last_status_change
> ---------+----------------+------+--------+-----------+---------+------------+-------------------+-------------------+-------------------+------------------------+---------------------
>  0       | server1             | 5433 | up     | 0.333333  | standby |
> 220909254  | false             | 0                 |                   |
>                      | 2022-03-31 08:33:02
>  1       | server2             | 5433 | up     | 0.333333  | standby |
> 6026186    | false             | 0                 |                   |
>                      | 2022-03-31 08:33:02
>  2       | server3             | 5433 | up     | 0.333333  | standby |
> 5798758    | true              | 0                 |                   |
>                      | 2022-03-31 08:33:02
> (3 rows)
> 
> server1: is NOT in recovery
> server2: is in recovery and streaming is working
> server3: is in recovery  and streaming is working
> 
> Nevertheless it seems that pgpool is working since clients are connecting
> to leader.
> 
> Why pgpool is not detecting correctly primary? How can we solve this
> situation?
> 
> Thanks in advance.
> Regards
> Luca


-- 
Bo Peng <pengbo at sraoss.co.jp>
SRA OSS, Inc. Japan
http://www.sraoss.co.jp/