Greenplum Disaster Recovery (GPDR) with Direct Replication reports "ERROR: [056]: unable to find primary cluster - cannot proceed"

Products

• VMware Tanzu Data
• VMware Tanzu Greenplum
• VMware Tanzu Greenplum / Gemfire

Issue/Introduction

When GPDR is configured with Direct Replication and a full backup is started on the DR cluster to initiate the replication an error like below is reported:

2026-03-10 17:17:56.461 P00   INFO: backup command begin 2.48+GPDR.6: --config=/usr/local/gpdr/configs/pgbackrest-seg-1.conf --db-timeout=360000 --exclude=pg_stat_tmp/ --exclude=pg_replslot/ --exclude=pg_dynshmem/ --exclude=pg_notify/ --exclude=pg_serial/ --exclude=pg_snapshots/ --exclude=pg_subtrans/ --exclude=backups --exclude=postgresql.auto.conf.tmp --exclude=postmaster.pid --exclude=postmaster.opts --exclude=log/ --exclude=current_logfiles.tmp --exclude=pg_internal.init --exec-id=3419847-394a38f7 --no-expire-auto --generate-backup-label-only --log-level-console=info --log-path=/usr/local/gpdr/logs --pg1-host=cdw --pg1-host-cmd="source /usr/local/greenplum-db-7.3.3/greenplum_path.sh && /usr/local/gpdr/bin/pgbackrest" --pg1-host-user=gpadmin --pg1-path=/data/coordinator/gpseg-1 --pg1-port=5442 --pg1-socket-path=/tmp --pg1-user=gpadmin --repo1-path=/data/coordinator/stanza-1 --repo1-type=posix --stanza=gpdb-seg-1 --type=full
WARN: option 'repo1-retention-full' is not set for 'repo1-retention-full-type=count', the repository may run out of space
      HINT: to retain full backups indefinitely (without warning), set option 'repo1-retention-full' to the maximum.
WARN: unable to check pg1: [FileMissingError] raised from remote-0 ssh protocol on 'cdw': unable to open missing file '/data/coordinator/gpseg-1/global/pg_control' for read
ERROR: [056]: unable to find primary cluster - cannot proceed
       HINT: are all available clusters in recovery?
2026-03-10 17:17:56.661 P00   INFO: backup command end: aborted with exception [056]

Cause

If the coordinator on both primary and DR clusters have the same "hostname", then this error may occur.

The "hostname" and "address" entries in the gp_segment_configuration table of the primary cluster are the same as the "hostname" on the DR cluster.

It is important that the hostnames of all the hosts in the DR cluster are different to the hostnames of the hosts in the primary cluster.

Resolution

Change the hostnames and the /etc/hosts entries on the DR cluster so that the hosts in the DR cluster have different names to the hosts in the primary cluster.

For example add "_dr" to the hostnames if they are the same as the primary cluster.