Edit history: Rexctl fails upgrade if external postgresql is used and internal postgres is disabled

Reproduction steps

Start with an existing Fluxer installation
Set postgres_db values to an external postgresql host
Disable local postgres in custom.compose.yml:

services: postgres: profiles: [external-db] api: depends_on: postgres: !reset null worker: depends_on: postgres: !reset null users: depends_on: postgres: !reset null messages: depends_on: postgres: !reset null

start an upgrade via rexctl upgrade

Expected behaviour

After starting the upgrade via rexctl upgrade, rexctl should regognize that the Database is external and skip it in the backup process

Actual behaviour

After starting the upgrade via rexctl upgrade, rexctl downloads the new images and then exists with a message that the instance couldn't be backed up

Platform

Self-hosting

Evidence

zoey@fluxer:~$ rexctl upgrade Fluxer 2026.1025 is available. You're on 2026.1024. Fluxer keeps running while the new version downloads, then stops for about 2 minutes. Upgrade?

Yes

✓ Updated rexctl to 2026.1025. [sudo] password for zoey: ✓ Downloaded Fluxer 2026.1025. ✗ Couldn't back up your instance. Fluxer 2026.1024 is still running. Read /home/zoey/fluxer/.rexctl/upgrade-2026.1025.41605.log to see why, then run ~/rexctl upgrade again. root@fluxer:~# cat /home/zoey/fluxer/.rexctl/upgrade-2026.1025.41605.log rexctl: backup failed: backup: no postgres image to run the copy with