View Issue Details
| ID | Project | Category | View Status | Date Submitted | Last Update |
|---|---|---|---|---|---|
| 0000752 | Pgpool-II | New feature | public | 2022-04-07 21:56 | 2022-06-28 10:53 |
| Reporter | garg1982@gmail.com | Assigned To | pengbo | ||
| Priority | high | Severity | major | Reproducibility | always |
| Status | closed | Resolution | open | ||
| OS | RedHat | OS Version | 7.5 | ||
| Product Version | 4.2.3 | ||||
| Summary | 0000752: Connection to database goes to hand due to auto failover | ||||
| Description | Hi, Just to mentioned that I observed that whenever connection between pgpool and database is break due to some network transient failure, new connections going to halt through pgpool until search_primary_node_timeout reached. It seems this auto failover is newly introduced from version 4.1 onward. Earlier it immediately downgrade the node status to standby from primary. No wait there. is this my understanding correct And I also preformed one testing, where I bring down the primary database subsequently pgpool detect it and went into failover mode but before it reach to search_primary_node_timeout, I started the database but still it is not reorganized that the database service up and after reaching above timeout, it degrade its status from primary to standby. As per my understanding it is a problem because during failover time it isn't checking that master is healthy. Please help to understand is it normal behavior? how we can disable auto failover or increase the time to initiate failover to cover the the transient failure? because we are using pgpool only for connection pooling. How we can re-attach the node primary if its status changed to standby from primary without restarting the pgpool ? is any auto way or other way ? Thanks & Regards Gaurav Garg | ||||
| Tags | No tags attached. | ||||
|
|
> I observed that whenever connection between pgpool and database is break due to some network transient failure, > new connections going to halt through pgpool until search_primary_node_timeout reached. It seems this auto failover is newly introduced from version 4.1 onward. > Earlier it immediately downgrade the node status to standby from primary. No wait there. is this my understanding correct We have enhanced failover in the latest version. Could you try the latest version of 4.2.x? > And I also preformed one testing, where I bring down the primary database subsequently > pgpool detect it and went into failover mode but before it reach to search_primary_node_timeout, > I started the database but still it is not reorganized that the database service up and after reaching above timeout, it degrade its status from primary to standby. The primary couldn't automatically failback. If pgpool detects the failure of primary, it will promote any standby to primary immediately. |
|
|
Have you resolved this issue? May I close this issue? |
|
|
No response over two weeks. Close issue. |
| Date Modified | Username | Field | Change |
|---|---|---|---|
| 2022-04-07 21:56 | garg1982@gmail.com | New Issue | |
| 2022-04-26 15:09 | pengbo | Note Added: 0004027 | |
| 2022-04-26 15:09 | pengbo | Assigned To | => pengbo |
| 2022-04-26 15:09 | pengbo | Status | new => feedback |
| 2022-06-02 13:50 | administrator | Note Added: 0004050 | |
| 2022-06-28 10:53 | pengbo | Note Added: 0004065 | |
| 2022-06-28 10:53 | pengbo | Status | feedback => closed |
