Extract files from 0.3 backup without flashing the StartOS 0.3 again

Unfortunately, I had quite a few problems when migrating from the latest version of 0.3 to 0.4 (OTA update triggered by System –> Software Update). I had the latest versions of both OS and the apps.

For my question go directly to point 3.

  1. Bitcoin Core
    “bitcoind daemon crashed”. Will not go into details, but I was in the end able to fix this one without reinstalling and doing the whole IBD again.

  2. electrs
    Refused to connect to Bitcoin Core because there was a conflict of authentication - both RPC auth and auth cookie method was used (at least this was written in the log…). I sshed into the machine but was not able to find any RPC user or password in Bitcoin.conf (never used that), so not sure where the problem was. Anyway, after a lot of fiddling and talking to LLMs, I lost patience and reinstalled the application. So now I’ve been doing blockchain re-indexing for some 12 hours. I hope it works after it is done…

  3. Nextcloud - I have over 200GBs worth of data I cannot access now. I have the 0.3 backup on external drive, but that cannot be loaded into 0.4 - I knew that before, I read the guides and migration instructions. But it still sucks. I was not expecting so many problems in this migration.

I have the backup on external drive and know the encryption password. Is there a way of getting plain-text data out of it? I mean some normal way, using normal backup/encryption tools. I really do not want to flash again 0.3 on my home server just because of this.

I was hoping something like this would do, but not sure about the exact implemented solution in StartOS, especially the encryption part (looks like eCryptfs but who knows):

duplicity restore --no-encryption file:///path/to/backup /restore/dir

For those interested in details, I sshed into the server and tried to manually delete the “offending” files and folders, but to no avail.

The original error in UI was:

Unknown Error: Error: sh failed with exit code 123: find: ‘/var/www/html/data/ivy/files/Ems��files/it/leg/.R_libraries/data.table’: No such file or directory chmod: missing operand after ‘ug+rw,o-rwx’ Try ‘chmod --help’ for more information.

In logs, I found this (note the “Ems—files” path part is correct there, not the “Ems��files” from error message above):

2026-07-26T15:39:12+02:00 chmod migration: processed 191600 directories, current: /var/www/html/data/ivy/files/Ems—files/it/leg/.R_libraries/crosstalk/lib/ionrangeslider/js
2026-07-26T15:39:14+02:00 Stopping container-runtime.service - StartOS Container Runtime…
2026-07-26T15:39:14+02:00 Caught SIGTERM
2026-07-26T15:39:14+02:00 Caught SIGTERM

  1. The cherry on top is that access to StartOS no longer works through TOR network. Interestingly, apps themselves can be accessed through their .onion addresses.

RPC ERROR: Unauthorized

invalid request signature: not valid for any identity this server recognizes (hostname, domain, or IP address); the request was likely signed for a different server identity or by an incompatible client

You technically can, though we don’t offer any support for it. Your 0.3.5.1 backup is an ecryptfs-encrypted folderEmbassyBackups/<server-id>/ — sitting on the backup drive (or CIFS share). Reading it on a plain Linux box is doable, but it’s not a normal mount, because your backup password isn’t the encryption key.

What you need:

  • A Linux machine with the ecryptfs kernel module and ecryptfs-utils installed, and root/sudo (no cryptsetup/LUKS involved — this is ecryptfs).
  • The backup drive attached and mounted normally, so you can see EmbassyBackups/<server-id>/.
  • The backup password.

Providing a LLM with the StartOS v0351 source code should be enough to get you a reasonable guide on how to proceed.

Easier though is to install v0351 on another device or in a VM, so you aren’t recreating all this.

Ok, will probably try the VM-way. Thank you for advice.

Regarding point 4, I suppose it should be fixed by startos update - it is available, but when I try to apply it, I get:

StartOS Update Failed - Update was not successful because of Filesystem I/O Error: No current.rootfs, not safe to prune

The system and all applications seem to be working fine.

To answer my question - restart of the server did not help. Had to re-flash the firmware/startos from a USB drive. Like fresh installation - just selected that app data should be retained, it seems they were.