Postgres healthcheck verifies password; document volume password mismatch (v2.5.11).

Co-authored-by: Cursor <cursoragent@cursor.com>
This commit is contained in:
orohi
2026-07-28 12:06:07 +03:00
co-authored by Cursor
parent 2746f981af
commit 706d2c9c0d
3 changed files with 18 additions and 2 deletions
+12
View File
@@ -206,6 +206,14 @@ The compose file connects the panel to Dokploy's Traefik network (`dokploy-netwo
3. Set env vars in Dokploy: `SECRET_KEY` (long random string), strong `POSTGRES_PASSWORD`.
4. Redeploy after changing domains or env.
**Postgres `password authentication failed`:** the database volume keeps the password from the **first** deploy. If you later change `POSTGRES_PASSWORD` in Dokploy, Postgres ignores the new value but the panel uses it in `DATABASE_URL`. Fix: either restore the original password in Dokploy env, or reset the DB volume (wipes panel DB data):
```bash
docker compose -p home-vpn-g6rfpd down
docker volume rm home-vpn-g6rfpd_amnezia_pgdata
docker compose -p home-vpn-g6rfpd up -d --build
```
If the domain shows **404 page not found** (plain text, Go-style), Traefik has no route to a healthy backend. After redeploying **v2.5.8+**, on the server run:
```bash
@@ -254,8 +262,12 @@ GitHub Actions workflows in `.github/workflows/`:
## 📋 Fix / changelog (this fork)
### v2.5.11
* **Postgres healthcheck** — verifies DB password (surfaces `POSTGRES_PASSWORD` / volume mismatch before panel starts).
### v2.5.10
* **Docker startup** — run `uvicorn` directly in `CMD` (no shell entrypoint); rebuild required after deploy.
* **Postgres healthcheck** — verifies password, not only `pg_isready` (catches `POSTGRES_PASSWORD` mismatch with existing volume).
### v2.5.9
* **Remove move connections between servers** — feature rolled back; fixes startup crash (`ToggleConnectionRequest` was missing after v2.5.4).