fix: use docker compose v2

[ticoombs]

With Ubuntu 22.04 and Debian 12 when using v1 commands this happens:
(Ref: https://askubuntu.com/questions/1508129/docker-compose-giving-containerconfig-errors-after-update-today)

This means your containers end up in a broken state. Updating all commands to be v2 compatible ensures they still work

Edit: Draft. Just realised there are more problems with it than I thought. (you can't > 15_to_16 for example)

[dessalines]

What's the error its giving you with > 15_16 ?

[ticoombs]

sudo docker compose exec -T postgres pg_dumpall -c -U lemmy > 15_16_dump.sql runs as two different users. :

• sudo docker compose exec -T postgres pg_dumpall -c -U lemmy
  • root
• > 15_16_dump.sql
  • Whatever the local user that is using sudo.

If you end up running the script as the regular user, which doesn't have write permissions on the local directory wherever we are piping to, you end up with a failure because you cannot write the output.

You'd get around this by calling the script via sudo scripts/postgres_15_to_16_upgrade.sh as then the "local" user inside the script is still root. But then all sudo commands inside the script are superfluous.

Example Script

#!/bin/bash
echo `id -u`
echo `sudo id -u`

> /bin/bash test.sh
1000
0

> sudo /bin/bash test.sh
0
0

tl;dr, running script via sudo works. running script (as regular user) so that each command inside the script elevates to sudo will fail to write the sql file

[Nothing4You]

people usually use sudo tee to deal with these cases:

sudo docker compose exec -T postgres pg_dumpall -c -U lemmy | sudo tee 15_16_dump.sql > /dev/null

[dessalines]

In that case use the sudo tee method above.

[dessalines]

Might also need to be sudo cat then now.

[dessalines]

I made the fixes above, so this is ready.

[ticoombs]

I'll test in the next day or so to confirm it works on lemmy-ansible deploys