Troubleshooting pg_upgrade v17

These troubleshooting tips address problems you might encounter when using pg_upgrade.

Upgrade Error - There seems to be a postmaster servicing the cluster

If pg_upgrade reports that a postmaster is servicing the cluster, stop all EDB Postgres Advanced Server services and try the upgrade again.

Upgrade Error - fe_sendauth: no password supplied

If pg_upgrade reports an authentication error that references a missing password, modify the pg_hba.conf files in the old and new cluster to enable trust authentication, or configure the system to use a pgpass.conf file.

Upgrade Error - New cluster is not empty; exiting

If pg_upgrade reports that the new cluster isn't empty, empty the new cluster. The target cluster might not contain any user-defined databases.

Upgrade Error - Failed to load library

If the original EDB Postgres Advanced Server cluster included libraries that aren't included in the EDB Postgres Advanced Server cluster, pg_upgrade alerts you to the missing component during the consistency check by writing an entry to the loadable_libraries.txt file in the directory from which you invoked pg_upgrade. Generally, for missing libraries that aren't part of a major component upgrade:

  1. Restart the EDB Postgres Advanced Server service.

  2. Use StackBuilder Plus to download and install the missing module.

  3. Stop the EDB Postgres Advanced Server service.

  4. Resume the upgrade process. Invoke pg_upgrade to perform consistency checking.

  5. After you resolve any remaining problems noted in the consistency checks, invoke pg_upgrade to perform the data migration from the old cluster to the new cluster.