Issues

Title: ETCD full causing PGSQL HA unavailable

Level: Critical, Please arrange fix ASAP

Scope: Pigsty v2.0.0 - v2.5.1, fixed in Pigsty v2.6.0

Description:

etcd has a default 2GB database capacity limit. If your etcd database exceeds this limit, etcd will reject write requests, which may cause PostgreSQL high availability mechanisms that depend on etcd to fail. At the same time, etcd’s data model creates a new version with each write, so even if you have only a few keys, frequent writes to your etcd cluster can cause the etcd database size to continuously grow and eventually fail when it reaches the capacity limit.

Solution:

Update Pigsty to v2.6.0 or above, or update the code in the roles/etcd section and re-execute ./etcd.yml to force reset the etcd cluster to implement the fix.

Key configuration update: roles/etcd/templates/etcd.conf.j2