View Issue Details
| ID | Project | Category | View Status | Date Submitted | Last Update |
|---|---|---|---|---|---|
| 0000417 | Pgpool-II | Bug | public | 2018-07-26 20:47 | 2018-08-27 10:31 |
| Reporter | akash | Assigned To | |||
| Priority | normal | Severity | major | Reproducibility | always |
| Status | closed | Resolution | open | ||
| Product Version | 3.6.2 | ||||
| Summary | 0000417: show pool_nodes is showing inconsistent roles(primary, standby) of postgresql nodes | ||||
| Description | We have following set up: 2 PostgreSQL VMs with version 9.6.7 3 Pgpool VMs with version 3.6.2 mode: master-slave streaming replication. node_id0: primary node node_id1: slave node Show pgpool_nodes output: node_id | hostname | port | status | lb_weight | role | select_cnt | load_balance_node | replication_delay ---------+--------------+------+--------+-----------+---------+------------+-------------------+------------------- 0 | 10.11.32.208 | 5432 | up | 0.500000 | primary | 0 | true | 0 1 | 10.11.32.209 | 5432 | up | 0.500000 | standby | 0 | false | 0 We have a bi-weekly update during which we update postgresql vms. While updating if master postgresql VM goes down i.e. node_id0 then we promote slave node node_id1, as new master. This step is completely handled by postgresql vms and pgpool has no involvement. Now, when we run select pg_is_in_recovery(), we get following output: node_id0 - t node_id1 - f That means node_id1 is master and node_id0 is slave as expected. However when we run show pool_nodes, we get following output: node_id | hostname | port | status | lb_weight | role | select_cnt | load_balance_node | replication_delay ---------+--------------+------+--------+-----------+---------+------------+-------------------+------------------- 0 | 10.11.32.208 | 5432 | up | 0.500000 | primary | 0 | true | 0 1 | 10.11.32.209 | 5432 | up | 0.500000 | standby | 0 | false | 0 Pgpool is not able to detect that role is changed. It does not update the state and remain in the same state unless we restart all 3 Pgpool-II. | ||||
| Steps To Reproduce | 1. Set node_id0 as primary and node_id1 as slave 2. Run "select pg_is_in_recovery()" on Postgresql nodes On node_id0: postgres=# SELECT pg_is_in_recovery(); pg_is_in_recovery ------------------- f (1 row) On node_id1: postgres=# SELECT pg_is_in_recovery(); pg_is_in_recovery ------------------- t (1 row) On pgpool VMs postgres=# SHOW pool_nodes; node_id | hostname | port | status | lb_weight | role | select_cnt | load_balance_node | replication_delay ---------+--------------+------+--------+-----------+---------+------------+-------------------+------------------- 0 | 10.11.32.208 | 5432 | up | 0.500000 | primary | 0 | true | 0 1 | 10.11.32.209 | 5432 | up | 0.500000 | standby | 0 | false | 0 3. Promote node_id1 as master from node_id0 and restart node_id0 as slave 4. Now Run "select pg_is_in_recovery()" on Postgresql nodes and run "show pool_nodes;" on Pgpool-II nodes On node_id0: postgres=# SELECT pg_is_in_recovery(); pg_is_in_recovery ------------------- t (1 row) On node_id1: postgres=# SELECT pg_is_in_recovery(); pg_is_in_recovery ------------------- f (1 row) On pgpool VMs postgres=# SHOW pool_nodes; node_id | hostname | port | status | lb_weight | role | select_cnt | load_balance_node | replication_delay ---------+--------------+------+--------+-----------+---------+------------+-------------------+------------------- 0 | 10.11.32.208 | 5432 | up | 0.500000 | primary | 0 | true | 0 1 | 10.11.32.209 | 5432 | up | 0.500000 | standby | 0 | false | 0 | ||||
| Tags | No tags attached. | ||||
|
|
If you directly execute promote and recovery, not using Pgpool-II's failover and online recovery, Pgpool-II will not detect the change of status. Restart Pgpool-II to detect the correct status of backend nodes. Or promote and recovery backend nodes by using Pgpool-II's failover and online recovery functions. |
|
|
If issue resolved, may I close this issue? |
| Date Modified | Username | Field | Change |
|---|---|---|---|
| 2018-07-26 20:47 | akash | New Issue | |
| 2018-07-27 10:11 | pengbo | Note Added: 0002126 | |
| 2018-08-13 09:42 | pengbo | Note Added: 0002149 | |
| 2018-08-27 10:31 | administrator | Status | new => closed |
