Version 5.0 v5
Enhancements, bug fixes, and other changes in EFM 5.0 include:
| Type | Description | Addresses |
|---|---|---|
| Enhancement | The startup log is no longer used by Failover Manager agents. Agent startup information now goes to systemd or standard out, depending on how the agent was started. | |
| Enhancement | New CLI command efm create-standby added to create standby databases from an existing primary using pg_basebackup. | |
| Enhancement | Failover Manager now simplifies handling cluster size for quorum purposes. Failed nodes/agents are still part of a cluster until removed. | |
| Enhancement | Added new backup.wal property to control whether or not standby WAL is backed up during reconfiguration. | 1277862, 91506, 101026 |
| Enhancement | The auto.resume.period property has been split into two new properties, one for agent startup and one for database failures. | |
| Enhancement | Added new check.num.sync.period property to control how often a primary database is checked for synchronous standbys. | 1133047, 91506 |
| Enhancement | Removed unnecessary check of database operating system user in sudoers configuration file. | 42594, 99957 |
| Enhancement | Removed unnecessary network status check that was added for older operating systems. | |
| Enhancement | Table information in efm cluster-status output now resizes to handle long host names. | 101878 |
| Enhancement | Updated "xlog" references in efm cluster-status-json output to use "lsn" instead. | |
| Enhancement | Changed the default detach.on.agent.failure behavior to match the suggested value. | |
| Enhancement | Added more cluster size checks during failures to improve split brain safety during network outages that result in gradual node isolation. | |
| Enhancement | Removed local database checks after an agent cannot connect that produced confusing logging. | |
| Enhancement | Added more information in some cases when an agent cannot connect to a database. | |
| Enhancement | Failover Manager was upgraded to use JGroups version 5.4.4. | |
| Enhancement | Failover Manager was upgraded to use PostgreSQL JDBC Driver version 42.7.3. | |
| Enhancement | Failover Manager was upgraded to use Log4J version 2.24.3. | |
| Enhancement | Failover Manager was upgraded to use Jakarta Mail version 2.0.1. | |
| Enhancement | Failover Manager was upgraded to use Json-Simple version 1.1.1. | |
| Bug Fix | Fixed agent shutdown code that could cause a systemctl stop call to finish before the agent completes shutdown, resulting in the agent process not exiting gracefully. | 37270 |
| Bug Fix | Handle a case where an agent on a failed/rebooted standby node attempts to join the cluster before the previous agent from the same host has timed out. | 43162 |
| Bug Fix | Removed efm upgrade-conf support for passwords in Failover Manager versions 4.0 and earlier. In rare cases, this support was causing upgrade failures. | |
| Bug Fix | Fixed an issue where a failed user fencing script could cause a temporary internal state mismatch. | |
| Bug Fix | Added missing descriptive output about node statuses in some cases when returned to an agent collecting cluster status information. | |
| Bug Fix | Added jdbc.properties property to properties template to avoid losing it during a configuration upgrade. |