Skip to content
Supported consistent full-volume backup and restore without a local CLI
pedro-sosa-lab
OP

a month ago

Hello Railway team,

What is the supported procedure to obtain a complete, consistent and independently restorable copy of an existing persistent volume containing multiple files while its service may still be writing?

Please clarify:

  1. Are native volume backups application-consistent across files, crash-consistent snapshots, or something else? What must the application do to quiesce writers and flush data, and what evidence confirms that all writers have stopped safely?
  2. Does downloading through Console > Files provide a consistent point-in-time copy? If not, what supported procedure is required, and does it require stopping/restarting a service or redeploying?
  3. Can a complete volume copy, including necessary metadata, be exported to independent private storage and restored into an isolated new volume/service for verification without replacing the original? What are the documented restoration limits and verification steps?
  4. Which options are available on Hobby versus Pro, and what plan requirements, backup/storage/egress charges and retention limits apply?
  5. Can the supported process be completed from the web dashboard or Safari on an iPhone without installing a local CLI or exposing credentials? If not, what exact additional capability is required?

This is a general procedural question. Please do not access or modify any specific project, create a backup, stop services, change a plan or incur charges. No files, secrets or project identifiers are included. Please link the relevant official documentation. Thank you.

Solved

3 Replies

Railway
BOT

a month ago

Native volume backups are point-in-time snapshots of the whole volume, so every file on that volume is captured together. They are crash-consistent, not application-consistent: we don't pause your app, flush its memory, or coordinate transactions, and separate volumes are not snapshotted together, so recovery relies on your application's or database's own crash-recovery and durability settings.

Restoring a backup creates a new volume and stages a change that swaps the service onto it. Deploying that change redeploys the original service, and the previous volume is kept unmounted. Restores stay in the same project and environment. There is no downloadable backup archive, no export to outside storage, and no supported way to attach a restored copy to a separate service for isolated verification while leaving the original untouched. If you discard the staged change, the original stays as it is and the restored copy sits unattached.

Backups can be created, restored and scheduled from the service's Backups tab in the dashboard, with no CLI needed. Scheduled backups are kept for 6 days (daily), 27 days (weekly) or 89 days (monthly). Manual backups are limited to 50% of the volume's size. You are billed only for each backup's incremental size, at the same rate as volume storage. Details are in the Backups docs and on plan pricing.


Status changed to Awaiting User Response Railway • about 1 month ago


Railway

Native volume backups are point-in-time snapshots of the whole volume, so every file on that volume is captured together. They are crash-consistent, not application-consistent: we don't pause your app, flush its memory, or coordinate transactions, and separate volumes are not snapshotted together, so recovery relies on your application's or database's own crash-recovery and durability settings. Restoring a backup creates a new volume and stages a change that swaps the service onto it. Deploying that change redeploys the original service, and the previous volume is kept unmounted. Restores stay in the same project and environment. There is no downloadable backup archive, no export to outside storage, and no supported way to attach a restored copy to a separate service for isolated verification while leaving the original untouched. If you discard the staged change, the original stays as it is and the restored copy sits unattached. Backups can be created, restored and scheduled from the service's Backups tab in the dashboard, with no CLI needed. Scheduled backups are kept for 6 days (daily), 27 days (weekly) or 89 days (monthly). Manual backups are limited to 50% of the volume's size. You are billed only for each backup's incremental size, at the same rate as volume storage. Details are in the [Backups docs](https://docs.railway.com/volumes/backups) and on [plan pricing](https://docs.railway.com/pricing/plans).

pedro-sosa-lab
OP

a month ago

Thank you. I still need help from a human Railway support specialist with the unresolved points in this same question. Please confirm the officially supported method for obtaining a complete, consistent copy of a multi-file persistent volume, exporting it to independent private storage, and verifying restoration in isolation without changing the original service. In particular, please address writer quiescence and how it is verified, whether Console > Files downloads are consistent, and whether this can be done from a browser or iPhone without a local CLI. Please also clarify Hobby versus Pro availability, restrictions and applicable costs. If no supported procedure meets these requirements, please state that explicitly and identify the exact additional capability required. This is a procedural clarification only; no project access, changes, backups, service stops or charges are authorized.


Status changed to Awaiting Railway Response Railway • about 1 month ago


Railway
BOT

a month ago

We don't have a supported procedure that meets all of these requirements together. There's no downloadable backup archive, no export of a volume copy to outside storage, and no supported way to attach a restored copy to a separate service while the original stays untouched.

Native volume backups capture the whole volume at one point in time, so all files on it are taken together. They are crash-consistent, not application-consistent. We don't pause your app, flush memory, coordinate transactions, or confirm that writers have stopped. Quiescing and verifying that is up to your application or database, which recovers using its own crash-recovery and durability settings.

Console Files downloads copy files from the running container as each one is read, so they aren't a point-in-time snapshot of a volume that is still being written to.

A restore creates a new volume in the same project and environment and stages a change moving the original service onto it. Applying that change redeploys the service. If you discard it, the original is unchanged and the restored copy sits unattached, but you can't browse or export it.

Backups are created, scheduled and restored from the service's Backups tab in a browser, with no CLI needed. They are billed only for their incremental size, at the volume storage rate. Scheduled backups are kept 6 days (daily), about 1 month (weekly) or about 3 months (monthly). See Backups and pricing.


Status changed to Awaiting User Response Railway • about 1 month ago


Railway
BOT

24 days ago

This thread has been marked as solved automatically due to a lack of recent activity. Please re-open this thread or create a new one if you require further assistance. Thank you!

Status changed to Solved Railway • 24 days ago


Welcome!

Sign in to your Railway account to join the conversation.

Loading...