Help for: Upgrading PostgreSQL to a new major version (v15 to v17)

[JGLord]

Hello, I'm trying to follow this simple guide (https://docs.teslamate.org/docs/maintenance/upgrading_postgres) to upgrading from Postgres v15 to v17, but I get an error when I try to delete the volume:

1. Backup
   docker-compose exec -T database pg_dump -U teslamate teslamate > ./teslamate.bck

2. Stop containers
   docker-compose down

3. Delete database volume

docker volume rm "$(basename "$PWD")_teslamate-db"
Error response from daemon: get TeslaMate_teslamate-db: no such volume

Here is my list of volumes

docker volume ls
DRIVER    VOLUME NAME
local     b2218a814cbf2ea13e859b497828a17c327b071d697b213519a0cb3736c4cfd4
local     c83483b66d82ad488aa50ec089310745b90ebc2be173a84baa6aac4d98542de4
local     d05b7c0a5cf9e3cbb70dd33af17755d65c36cf2c4e6a8424b7c108fe0f6ebb7a
local     e88d21f83ba3566ac75cea833465cce50cc42425777e6601fa807bc8b5e1b6fe

What am I doing wrong?

How do I know which container a volume belongs to?

Thank you!

[JGLord]

I don't know if this is the best methode but I destroyed the database container and then the orphan volumes (obviously, after making a backup):

docker rm teslamate-database-1
docker volume prune

Then I was able to update Postgres to version 17. However, Postgres was falling into a stop/start loop (database files are incompatible with server):

docker-compose logs
teslamate-database-1 | 2025-05-04 01:44:01.658 UTC [1] FATAL: database files are incompatible with server
teslamate-database-1 | 2025-05-04 01:44:01.658 UTC [1] DETAIL: The data directory was initialized by PostgreSQL version 15, which is not compatible with this version 17.4 (Debian 17.4-1.pgdg120+2).

I finally destroyed the entire file structure of my database: ./TeslaMate/teslamate-db/

And now everything works...

---

I also found the link below not very useful:
https://docs.teslamate.org/docs/maintenance/upgrading_postgres

Here my steps:

1. Create a backup
   docker exec -t teslamate-database-1 pg_dump -U teslamate teslamate > teslamate_database_$(date +%F_%H-%M-%S).sql

2. Stop your TeslaMate stack
   docker-compose stop

3. Delete the old database container and orphan volumes

docker rm teslamate-database-1
docker volume prune

4. Update the compose file to current version (17). I'm on Synology NAS, so docker = Container Manager.
   Container Manager->Project->Details->YAML Config

database:
	image: postgres:17

5. Recreate the project
   Container Manager->Project->Create
   or i can use the command: docker-compose up -d database

6. Stops the teslamate container to avoid writes during restoration and copies the backup in preparation for loading.

docker stop teslamate-teslamate-1
docker cp ./teslamate_database[_date].sql teslamate-database-1:/tmp/teslamate_backup.sql

7. Destroy the database
   we cant DROP an active DATABASE, so lets drop the SCHEMA to avoid permissions management

docker exec -it teslamate-database-1 psql -U teslamate
DROP SCHEMA public CASCADE;
CREATE SCHEMA public;
CREATE EXTENSION cube WITH SCHEMA public;
CREATE EXTENSION earthdistance WITH SCHEMA public;

8. Restore the backup
   docker exec -i teslamate-database-1 psql -U teslamate -d teslamate -f /tmp/teslamate_backup.sql

9. Restart the container
   docker start teslamate-teslamate-1

I hope it will help others.

[brianmay]

I think, as @JakobLichterfeld said below "You are using docker-compose (v1) which has been deprecated for years. Use only docker compose as stated in the docs."

I should update my PR with instructions to make sure you have the correct version before proceeding.

[JakobLichterfeld]

I should update my PR with instructions to make sure you have the correct version before proceeding.

We do have a note about this in the Backup section: https://docs.teslamate.org/docs/maintenance/backup_restore

Which may be useful in more places, but the deprecation is years old, so the question is when to drop these kind of hints in total.

[MailmanGit]

I think I've succeeded in making the backup. Step one, was to create the backup inside the container with the following command:

docker exec TeslaMate-DB sh -c "pg_dump -U teslamate teslamate > /tmp/teslamate.bck"

I then copied the backup to the host with this command:

docker cp TeslaMate-DB:/tmp/teslamate.bck ./TeslaMate_DB_$(date +%F_%H-%M-%S).sql

That worked and created a backup with bytes of data. I then removed the backup in the container:

docker exec TeslaMate-DB rm /tmp/teslamate.bck

Next step now is to upgrade PostgreSQP and restore...

[MailmanGit]

Updating was a nightmare, but it worked in the end using ChatGPT. For those of you reading this and wanting to do the same. Make sure to use version 17.3, as 17.2 is not supported by Teslamate. At least, that was what I noticed.

[JakobLichterfeld]

Updating was a nightmare, but it worked in the end using ChatGPT. For those of you reading this and wanting to do the same. Make sure to use version 17.3, as 17.2 is not supported by Teslamate. At least, that was what I noticed.

And what is clearly stated in docs, changelog and log message.

[mayishan11]

I have updated teslamate firstly and could not run backup, how could I revise the version of teslamate?

[JGLord]

I don't know how to downgrade TeslaMate, but when you say you can't do the backup, what's your problem? Does the postgres container work, because I don't think the teslamate container is required for the backup.

[jcsp101]

moving to new thread for clarity

[JakobLichterfeld]

What am I doing wrong?

You are using docker-compose (v1) which has been deprecated for years. Use only docker compose as stated in the docs.