CLOUD BACKUP & PHOTO LIBRARIES
The crypt layer encrypts files before they reach another rclone remote. The storage account remains necessary, but its password alone cannot unlock your encrypted collection.
Quick answer
Create a normal remote, wrap a dedicated folder with a crypt remote and use the crypt name for uploads and restores.
Steps to follow
-
Before you start
Create a dedicated empty destination prefix and name the provider and crypt remotes distinctly. Record filename settings, the password and the optional second secret if used. Prepare protected independent configuration recovery before uploading private files.
-
Set up the right scope
In rclone config, configure and test the underlying storage first. Allocate an empty folder or prefix for encrypted files so ordinary files and ciphertext are not mixed.
-
Apply the relevant settings
Create another remote of type crypt pointing to that dedicated location. Choose filename encryption deliberately. Save the crypt password, the optional second secret if used and the configuration securely; a lost encryption secret is not recovered by resetting the provider login.
-
Complete the first run
Upload a small sample with rclone copy /path/sample cryptname:test, then copy it back from cryptname:test to an empty local folder. Access through the underlying remote shows encrypted file contents; the filename appearance depends on the crypt naming options you selected.
Decisions and checks
The limitation that matters
Do not change crypt passwords on an existing collection expecting old files to be re-encrypted automatically. The crypt layer supplies encryption, not historical versions or protection from a sync command’s deletions.
How to verify the result
Open the recovered sample. Where the backend supports suitable checksums, use cryptcheck for an additional integrity comparison; also test access using the saved recovery configuration.
Keep the result reliable
Review the remote and configuration used by scheduled jobs after path or credential changes. Repeat a round trip with saved recovery settings and check independent retention. Keep old crypt access until any key migration has been fully validated.
Separate encrypted storage access from the ability to recover files
Map the ordinary remote and its encryption wrapper
Configure the storage provider first and verify a harmless file transfer to an isolated location. Then create the crypt remote over a dedicated folder or prefix within that storage. These are two different views of the same underlying objects. A transfer addressed to the crypt remote passes through its encryption layer; a direct transfer to the provider remote does not. Use clear names in scripts and recovery notes so the distinction remains visible months later. Keep ordinary files out of the encrypted prefix. Mixing the two makes later inspection confusing and can expose a document that you mistakenly believed was encrypted simply because other objects nearby were unreadable.
Choose filename behavior instead of assuming it
Crypt offers configuration choices for filename and directory-name handling. Record the options you selected alongside the password and any optional second secret used. File contents remain a separate consideration from whether a path reveals recognizable names. If your design requires obscured names, verify that choice by viewing the underlying storage rather than assuming every configuration hides them identically. Include spaces, accented characters and nested folders in the trial. Avoid manually renaming encrypted objects in the provider's console to make them easier to read. The mapping between the stored representation and the names exposed by crypt belongs to the configured encryption format.
Protect the configuration file as sensitive material
Rclone can store secrets in an obscured form, which is not the same as encrypting the whole configuration against someone who can read it. Use the documented configuration-encryption mechanism where appropriate and restrict filesystem access. Keep a protected independent recovery copy, including the exact crypt settings. If a scheduled job needs credentials, use a supported method that does not print secrets into shared logs or shell history. Document how an authorized operator obtains the configuration and its protection password after losing the original computer. Saving that information only inside the encrypted collection creates a circular dependency: you would need access to the archive in order to learn how to access it.
Test the round trip through the correct remote
Copy a small sample to the crypt remote and retrieve it through that same crypt configuration into a new empty directory. Open a text document and image and compare their contents with the source. Inspect the nested paths as well. A direct download through the underlying provider returns the encrypted representation, so inability to open that object in a photo viewer is expected. Do not overwrite the source during the test; it is your reference for comparison. If the backend supports the required checksums, cryptcheck provides another integrity check, but it has its own prerequisites and scope. A practical round trip also verifies that the saved settings produce usable files.
Rotate secrets through a controlled migration
Changing the crypt password field does not automatically re-encrypt existing objects. A collection encrypted with the old settings still needs those settings to be read. Plan a transition that reads the old collection and writes a new separately identified encrypted destination, following the appropriate rclone workflow. Begin with a small sample and verify it using the new configuration before moving further. Keep the old credentials and data until the complete result is validated. Do not write new files with unrelated crypt secrets into the same prefix without a clear supported design. Otherwise a single-looking archive may contain objects that require different configurations, making recovery far harder than the original password change.
Add retention and provider recovery as separate layers
Encryption protects confidentiality; it does not stop an authorized sync operation from deleting destination objects or preserve earlier versions by itself. Decide how previous file states survive mistakes and test that recovery route independently. Provider account access also remains necessary, including its recovery and verification requirements. A correct crypt password cannot retrieve objects from an account you cannot reach, and a recovered provider login cannot decrypt an archive without the crypt secrets. Keep a map of those dependencies. When restoring sensitive files locally, choose an appropriately protected destination and enough free space. Remove temporary working copies deliberately after validation, without deleting the only remaining readable version.
Example: the provider shows a mixture of readable and unusual names
Suppose a scheduled upload produces a normal document name among otherwise encrypted objects. Check the remote used by that job and the configured filename mode before drawing conclusions. If the script bypassed crypt, preserve the local source and assess that specific upload through the provider's normal controls. Correct the job using a harmless sample and inspect both views again. Do not assume that moving the object within the provider's folder will apply encryption. Only the intended crypt operation performs that transformation. Record the corrected command and verify the actual unattended job, because an interactive test may use a different configuration file.
Compare the options
| Situation / component | Purpose / action | What to check |
|---|---|---|
| Provider access works, files remain unreadable | Use the correct crypt configuration | Account recovery does not replace encryption secrets |
| Readable upload appears in encrypted area | Inspect the job remote and filename mode | Direct provider writes bypass crypt |
| Earlier version is required | Use separately configured retention | Encryption alone does not keep history |
Frequently asked questions
Must I configure a second crypt password?
The second secret is optional. If you use one, preserve it with the main password and the other configuration choices needed to reconstruct that crypt remote.
Does a password change update the existing encrypted files?
No. Plan a supported read-and-re-encrypt migration into a separate destination. Preserve the old settings and collection until the new archive has passed recovery checks.
Official documentation
This guide combines current official documentation with practical recovery checks. It does not claim hands-on testing of your installation. Menu labels and supported procedures can vary by release; follow the documentation for your installed version.