Loading user profile...

ian-elisyan

1thread

4replies

Threads created

Self-Hosted clean Postgres 17 docker instance fails to start due to Analytics

The user is experiencing a failure to start a self-hosted Postgres 17 instance due to an error with the analytics container. Despite changing image versions and cleaning volumes, the issue persists. The error message indicates a dependency failure related to analytics.

Self-hosting

Database

Recent replies

I can confirm that the Nix version of the script successfully executed and performed the upgrade without any issues.

4/14/2026, 9:08:03 PMView on platform

No problem. Happy to help. Will give it a try and report back.

4/14/2026, 12:45:26 PMView on platform

OS: macOS (Mac mini M4 Pro) - latest version Docker: Desktop v4.68 for Mac Apple Silicon Reverse Proxy: Caddy - latest version I had to tweak the docker compose file though as I have 4 separate Supabase instances running on different URL paths. Two instances had limited "pre-set" data so they were rebuilt from scratch using a clean PG17 rebuild and I used sql queries to rebuild the DB. No issues with these ones. The other 2 instances are on the Feb 2026 release, have data and I'm planning to upgrade both over time. I have only activated the Cognito Wrapper extension, everything else is default. Changes to the docker compose files are summarised below: 1. Added distinct networks to each Supabase instance via an env variable - prevents cross communication and isolates each instance. 2. Modified container name(s) with an env variable and replicated this across all container names within the compose file - prevents uniqueness issues within docker. 3. Added Authelia as the forward_auth provider on all instances to protect studio on each instance - disabled Kong's responsibility for this via it's yml. 4. Exposed ports have been modified, via the env, but I saw no specific references to this in the upgrade script. Doing a comparison of the latest repo docker-compose against mine the only differences are items 1 and 2 above. Nothing else has changed, all other configs were done via the env. I did modify the upgrade script's `DB_CONTAINER` variable to point to the actual container name that gets generated in my compose file though. In terms of running the upgrade script, I switched from zsh to bash in macos terminal and ran it as mentioned in the notes in the root folder of the relevant supabase instance. Need anything further?

4/14/2026, 11:40:59 AMView on platform

As requested. [https://gist.github.com/ian-elisyan/a9d61367306d71705e537b2e83fa3a40](url)

4/13/2026, 7:54:51 PMView on platform