On Tue, Dec 24, 2024 at 4:28 AM Matthias Leisi <matthias@xxxxxxxxx> wrote:
I installed Postgres 16/17 from apt.postgresql.org (with the intent of using version 17) on Debian 12/Bookworm. Both have the same version of repmgr:postgres@dev-pgcl01:~$ /usr/lib/postgresql/17/bin/repmgr --versionrepmgr 5.5.0postgres@dev-pgcl01:~$ /usr/lib/postgresql/16/bin/repmgr --versionrepmgr 5.5.0root@dev-pgcl01:~# apt-cache show postgresql-17-repmgr | grep "Filename"Filename: pool/main/r/repmgr/postgresql-17-repmgr_5.5.0-2.pgdg120+1_arm64.debIn postgresql.conf, I explicitly load the version 17 .so (just to be sure it’s not loading something potentially different):shared_preload_libraries = '/usr/lib/postgresql/17/lib/repmgr.so'repmgr.conf is minimalistic:node_id=1node_name=node1conninfo='host=10.20.70.26 user=repmgr dbname=repmgr connect_timeout=2'data_directory='/var/lib/postgresql/17/'failover=automaticuse_replication_slots=truepromote_command='/usr/bin/repmgr standby promote -f /etc/postgresql/17/main/repmgr.conf --log-to-file'follow_command='/usr/bin/repmgr standby follow -f /etc/postgresql/17/main/repmgr.conf --log-to-file --upstream-node-id=%n‘(plus a few configs about the logs)I created the cluster with/usr/lib/postgresql/17/bin/repmgr -f /etc/postgresql/17/main/repmgr.conf primary registerAnd it shows what I would expect:postgres@dev-pgcl01:~$ /usr/lib/postgresql/17/bin/repmgr -f /etc/postgresql/17/main/repmgr.conf cluster showDEBUG: connecting to: "user=repmgr connect_timeout=2 dbname=repmgr host=10.20.70.26 fallback_application_name=repmgr options=-csearch_path="DEBUG: connecting to: "user=repmgr connect_timeout=2 dbname=repmgr host=10.20.70.26 fallback_application_name=repmgr options=-csearch_path="ID | Name | Role | Status | Upstream | Location | Priority | Timeline | Connection string----+-------+---------+-----------+----------+----------+----------+----------+----------------------------------------------------------------------------------------1 | node1 | primary | * running | | default | 100 | 1 | host=10.20.70.26 user=repmgr dbname=repmgr connect_timeout=2I attempt to start repmgrd withpostgres@dev-pgcl01:~$ /usr/lib/postgresql/17/bin/repmgrd -v -f /etc/postgresql/17/main/repmgr.conf --verboseWhich gives the following log output:[2024-12-24 09:21:25] [NOTICE] repmgrd (repmgrd 5.5.0) starting up[2024-12-24 09:21:25] [INFO] connecting to database "host=10.20.70.26 user=repmgr dbname=repmgr connect_timeout=2"[2024-12-24 09:21:25] [DEBUG] connecting to: "user=repmgr connect_timeout=2 dbname=repmgr host=10.20.70.26 fallback_application_name=repmgr options=-csearch_path="[2024-12-24 09:21:25] [DEBUG] set_config():SET synchronous_commit TO 'local'[2024-12-24 09:21:25] [DEBUG] expected extension version: 50500; extension version: 50400[2024-12-24 09:21:25] [ERROR] this "repmgr" version is newer than the installed "repmgr" extension version[2024-12-24 09:21:25] [DETAIL] "repmgr" version 5.5.0 providing extension version 5.5.0 is installed but extension is version 5.4[2024-12-24 09:21:25] [HINT] update the installed extension version by executing "ALTER EXTENSION repmgr UPDATE" in the repmgr databaseSELECT * FROM pg_catalog.pg_extension WHERE extname = 'repmgr'shows version 5.4The suggested ALTER EXTENSION from the log hint does not do anything I could see.The same effect if I try the same using version 16.How to continue from here? What am I missing?
This [HINT] says that YOU are supposed to run the ALTER EXTENSION command:
[quote]
update the installed extension version by executing "ALTER EXTENSION repmgr UPDATE"
[/quote]
Death to <Redacted>, and butter sauce.
Don't boil me, I'm still alive.
<Redacted> lobster!