View Issue Details
| ID | Project | Category | View Status | Date Submitted | Last Update |
|---|---|---|---|---|---|
| 0000484 | Pgpool-II | Bug | public | 2019-03-26 21:51 | 2019-04-02 09:10 |
| Reporter | staslg | Assigned To | pengbo | ||
| Priority | high | Severity | major | Reproducibility | always |
| Status | closed | Resolution | open | ||
| Product Version | 4.0.1 | ||||
| Summary | 0000484: Change node status withount change role type | ||||
| Description | Hello pgpool --version pgpool-II version 4.1devel (karasukiboshi) Configuretion type [ raw mode ] I have a problem with role status chages: After failover: In pgpool: node_id | hostname | port | status | lb_weight | role | select_cnt | load_balance_node | replication_delay | last_status_change ---------+---------------+------+--------+-----------+--------+------------+-------------------+-------------------+--------------------- 0 | db-n1 | 5432 | down | 0.500000 | slave | 0 | false | 0 | 2019-03-26 15:13:54 1 | db-n2 | 5432 | up | 0.500000 | master | 594 | true | 0 | 2019-03-26 15:13:28 In repmgr : ID | Name | Role | Status | Upstream | Location | Connection string ----+-------+---------+-----------+----------+----------+-------------------------------------------------------- 1 | node1 | standby | running | node2 | default | host=db-n1 port=5432 user=repmgr dbname=repmgr 2 | node2 | primary | * running | | default | host=db-n2 port=5432 user=repmgr dbname=repmgr If I do change (pcp_attach_node 0 ), pgpool(master<->slave) update role status and begins to think that node 0 is master now, although it is not true, because in repmgr node1 in slave, node2 primary node_id | hostname | port | status | lb_weight | role | select_cnt | load_balance_node | replication_delay | last_status_change ---------+---------------+------+--------+-----------+--------+------------+-------------------+-------------------+--------------------- 0 | db-n1 | 5432 | up | 0.500000 | master | 0 | true | 0 | 2019-03-26 15:40:40 1 | db-n2 | 5432 | up | 0.500000 | slave | 3745 | false | 0 | 2019-03-26 15:13:28 Why node 0 cannot be a slave role in up status ? How to fix it or configure ? | ||||
| Tags | No tags attached. | ||||
|
|
Is node0 is a slave of node1? Did you recovery node0 from node1? Connect to node0, try the command below: ------------ select pg_is_in_recovery(); ------------ If the result is "f", Pgpool-II will consider this node0 as primary. So you have to recovery this node0 from node1 first by using "pcp_recovery_node". |
|
|
Is node0 is a slave of node1? -Yes Did you recovery node0 from node1? - Yes You can saw it in repmgr out: node1 | standby | running | node2 ID | Name | Role | Status | Upstream | Location | Connection string ----+-------+---------+-----------+----------+----------+-------------------------------------------------------- 1 | node1 | standby | running | node2 | default | host=db-n1 port=5432 user=repmgr dbname=repmgr 2 | node2 | primary | * running | | default | host=db-n2 port=5432 user=repmgr dbname=repmgr node_id | hostname | port | status | lb_weight | role | select_cnt | load_balance_node | replication_delay | last_status_change ---------+---------------+------+--------+-----------+--------+------------+-------------------+-------------------+--------------------- 0 | db-n1 | 5432 | down | 0.500000 | slave | 0 | false | 0 | 2019-03-26 15:40:50 1 | db-n2 | 5432 | up | 0.500000 | master | 107229 | true | 0 | 2019-03-26 15:13:28 db-n1$ postgres=# select pg_is_in_recovery(); pg_is_in_recovery ------------------- t ------------------------------------------------------------------ db-n2$ postgres=# select pg_is_in_recovery(); pg_is_in_recovery ------------------- f pcp_recovery_node option for replication and streaming replication modes, but we use raw mode |
|
|
You are using streaming replication between node0 and node1. Why are you using raw mode. I think you should use streaming replication mode. |
|
|
Role status works correctly only in streaming replication mode ? |
|
|
> Role status works correctly only in streaming replication mode ? Yes. I think you may misunderstand raw mode. In raw mode, pgpool will consider the alive node with smallest node number as master node. |
|
|
Ok, I've got it, thanks You may close the case. |
| Date Modified | Username | Field | Change |
|---|---|---|---|
| 2019-03-26 21:51 | staslg | New Issue | |
| 2019-03-27 09:20 | pengbo | Note Added: 0002466 | |
| 2019-03-27 09:20 | pengbo | Assigned To | => pengbo |
| 2019-03-27 09:20 | pengbo | Status | new => assigned |
| 2019-03-27 10:22 | staslg | Note Added: 0002468 | |
| 2019-03-28 19:30 | pengbo | Note Added: 0002479 | |
| 2019-04-01 13:43 | staslg | Note Added: 0002490 | |
| 2019-04-01 16:44 | pengbo | Note Added: 0002493 | |
| 2019-04-01 18:13 | staslg | Note Added: 0002496 | |
| 2019-04-02 09:10 | pengbo | Status | assigned => closed |
