Hi Railway team,

I woke up this morning to find that my Chatwoot instance was completely broken after being automatically migrated to Metal overnight without my knowledge.

Everything was working perfectly before this migration, but now the instance fails to redeploy with a Healthcheck failure and vector extension errors related to Postgres. It seems the Metal server doesn't support the pgvector extension that Chatwoot now relies on.

This is a critical issue for me — I rely on this instance for production use and didn't change anything myself. Here's what I need help with:

1. Is it possible to roll back to the previous working deployment (before Metal)?

2. If not, how can I migrate my Postgres to an environment where pgvector is supported (like Neon) and keep everything working with Railway?

3. Why was this migration forced and can it be disabled for future updates?

Please let me know ASAP — this broke my production system and I really need to get it back online.

Thanks in advance!