gpinitstandby or gpupgrade fails to synchronize extensions/packages to the standby coordinator host in Greenplum 7.x

Products

• VMware Tanzu Greenplum
• VMware Tanzu Greenplum / Gemfire

Issue/Introduction

Symptoms

After completing a Greenplum upgrade using gpupgrade or initializing a new standby coordinator using gpinitstandby, the system reports a successful completion. However, when performing package-related operations (such as installing Greenplum Command Center metrics_collector or running gppkg query), the utility fails with an inconsistency error.

Error Message:

[gpadmin@coordinator ~]$ gppkg queryDetecting network topology:    [==============================================================] [OK]2 coordinators and 32 segment instances are detected on 6 unique hosts.
Error: InConsistentClusterStatus: server 'gm02' missing package: 'pljava', 'plr', 'DataSciencePython3.11', 'madlib', 'DataScienceR'

In the gpinitstandby logs, the utility may claim to be syncing extensions, but the physical binaries are not actually copied to the standby host: 20260607:11:59:07:2956633 gpinitstandby:data14:gpadmin-[INFO]:-Syncing Greenplum Database extensions to standby

Cause

This issue is caused by a regression in Greenplum 7.x (specifically following refactoring to remove internal gppkg dependencies) where the gpinitstandby utility fails to physically synchronize installed extension and package binaries to newly initialized standby coordinator hosts in multi-host cluster configurations.

 

Resolution

This issue is scheduled to be fixed in Greenplum 7.8.3. Users are encouraged to upgrade to this version or later once it is released to resolve the synchronization logic.