14 hours ago
Hello Railway Team,
I'm evaluating Railway as a potential hosting platform for a web application. I would appreciate official guidance on the following PostgreSQL 16 capabilities and operational controls before authorizing deployment. This is a platform compatibility inquiry, not a report of a confirmed defect or a request to audit or debug application code.
- PostgreSQL roles and privileges
Do the administrative credentials supplied with Railway's PostgreSQL 16 service allow custom LOGIN/NOLOGIN and NOINHERIT roles; memberships with SET TRUE, INHERIT FALSE, and ADMIN FALSE; schema and function ownership transfers; restrictions on PUBLIC and default privileges; and column-level and function-level permissions? Can separate runtime, migration, bootstrap, and operational identities be configured?
- pgcrypto
Is pgcrypto 1.3 available for installation in a custom schema? What restrictions apply to ownership and execution privileges for extension functions, including HMAC and digest?
- Logging and instrumentation
Can customers set shared_preload_libraries to an empty value, including disabling pg_stat_statements and pgaudit where present, and preserve that configuration across startup, redeployment, and recovery? Would this require a custom image?
Could SQL text or bound query parameters be captured by Railway's platform instrumentation, tracing, logs, or diagnostic systems? Which controls disable or limit that capture per service, and what exposure remains outside customer control?
- Verified PostgreSQL TLS
What is the recommended approach for full certificate-chain and hostname verification over Railway private networking? Please clarify trusted CA distribution, certificate renewal, whether renewal replaces the CA, restore behavior, certificate coverage for private hostnames, and secure remote administration.
- Secrets and deployment coordination
Which controls support service-specific runtime secrets; staging values before activation; immutable or versioned secret configurations; coordinated deployment across replicas; confirmation that old workers have terminated; prevention of stopped workers automatically restarting; and verification of active configuration before reopening traffic? Are there recommended integration patterns where these controls are not directly provided?
- Backup and recovery
What is the recommended process for restoring PostgreSQL into an isolated environment for verification before connecting application services? What limitations apply to cross-environment restoration, logical exports, and point-in-time recovery?
- Hosting approach
For deployments needing these controls, which approach would you recommend: the standard PostgreSQL 16 template, a custom PostgreSQL 16 Docker image on Railway, or Railway application services connected to an external PostgreSQL provider?
Links to current documentation or official guidance would be greatly appreciated.
Thank you for your time.
2 Replies
Status changed to Awaiting Railway Response Railway • about 14 hours ago
14 hours ago
Roles and privileges: the database runs from our SSL-enabled Postgres image, built on the official Docker Hub Postgres image, and you connect with the supplied PGUSER credentials. Roles, memberships, ownership transfers and grants are standard PostgreSQL that you configure yourself, so separate runtime, migration and operational identities are yours to set up.
pgcrypto and preload libraries: the template ships with no added extensions, and server configuration follows the image. To pin settings like shared_preload_libraries across redeploys, extend the image per the Docker Hub instructions or fork our postgres-ssl repository. Changing the template's image, start command or variables makes the database yours to manage (support scope).
Logging: anything Postgres writes to stdout/stderr appears in your service logs, so your Postgres logging settings decide whether statement text lands there.
TLS: private networking traffic is encrypted and isolated per project and environment. For a CA chain you control with hostname verification, build your own certificates into a forked image.
Secrets and deploys: variables are per service, can be sealed so values are never readable in the UI or API, changes are staged for review before deploying, and a rollback restores a previous deployment's image, settings and variables.
Backup and recovery: volume backups require Pro and restore only within the same project and environment. Point-in-time recovery restores into a new sibling service while the original keeps running, so you can verify the copy before connecting apps.
Hosting approach: the template is maintained by us, a custom image is yours to maintain. For controls beyond what's described here, validate your workflow in a separate test environment first. Compliance reports are at trust.railway.com.
Status changed to Awaiting User Response Railway • about 14 hours ago
14 hours ago
This thread has been opened as a bounty so the community can help solve it.
Status changed to Open Railway • about 14 hours ago
37 minutes ago
Puldi qande yechsa boʻladi