|
|
| Reporter | thanos.cyrax | Assigned To | | |
|---|
| Priority | normal | Severity | major | Reproducibility | always |
|---|
| Status | closed | Resolution | open | |
|---|
| Platform | PgPool II implementation in K8s | OS | RHEL | OS Version | 7.8 |
|---|
| Product Version | 4.2.5 | |
|---|
|
|
| Summary | 0000734: PgPool is not working in manual repmgr switch-over |
|---|
| Description | We have implemented PgPool in Kubernetes which connects to three external Postgres'es in Active Passive using repmgr.
The implementation was made from an initial Bitnami Helm Chart which was stripped down to fit our needs, so it cannot be considered bitnami at this point in time but custom.
Pgpool is isolated and Postgres A/P Cluster is remote.
When we test failover by bringing down the master or the slaves Postgres'es everything works as expected.
When we try to do a manual switch-over, PgPool is not recognizing the new Postgresql servers' states.
It continues looping without acquiring the status for the switched-over master database. |
|---|
| Steps To Reproduce | - Create a cluster with 3 to 4 PostgreSQL databases, 1 master, 2 slaves, 1 to act as Witness. The witness might not be needed but I am sharing it with you either way.
- Configure PgPool to see the 3 databases (1 master, 2 slaves).
- In a slave Postgres execute: repmgr standby switchover —siblings-follow
- PgPool is not recognizing the new Postgres cluster states. It continues looping indefinitely with errors. Unfortunately I do not have the logs right now, will try to reproduce and share it with you.
- At this point if we restart PgPool, the cluster state is recognized and all database operations run as expected. |
|---|
| Tags | No tags attached. |
|---|
|
|
