Backup verification error

[Kelly Dean]

On Qubes 4.0, I did a full backup to an external hard drive using the standard backup utility, which completed successfully. 225GB total, compressed 130GB, took about 12 hours.

Then I tried to verify it (restore with verify-only option), and watched it for a few minutes to make sure it was running ok, then left it alone. Last line in the message window at the time was:
Extracting data: 209.6 GiB to restore

Came back a few hours later and the next line was:
Finished with errors!

And there was a dialog box:

[Dom0] Backup error!
ERROR: failed to decrypt
/var/tmp/restorexnuw0a8p/vm31/private.img.034.enc: b'scrypt:
Decrypting file would take too much CPU time\n'
Partially restored files left in /var/tmp/restore_*, investigate them
and/or clean them up
OK

So now I don't know if I have a good backup. The error message also leaves me doubtful that I'd be able to restore the backup even if it is good.

The indicated file doesn't exist in dom0. Neither does any other /var/tmp/restore* file. Googling the message (Decrypting file would take too much CPU time) finds nothing.

[Tai...@gmx.com]

Considering how long it takes and the chance for errors I also make a
post fsck dd backup of the entire drive/partition and then sha1sum it
just in case, which has saved me a few times.

I wish there was a choice to use more RAM to make it go faster or what not.

[awokd]

Kelly Dean wrote on 10/17/18 6:00 PM:

Looks like that error string comes from scrypt. There is a command line
option (-f) to scrypt that forces it to proceed even if it might take
excessive memory or CPU time, but I'm not sure how to pass it on through
qubes-backup-restore.

[Chris Laprise]

Looking at the scrypt source, this error code results from the estimated
number of operations exceeding a max time limit. Using -f should
override this check.

In restore.py the function 'def launch_scrypt' contains the line
starting with:

> command_line = ['scrypt', action, input_name

This could be changed to:

> command_line = ['scrypt', action, '-f', input_name

Of course, this advice is offered "at your own risk" and a backup copy
of restore.py should be made before doing any modification.

-

BTW, since your backups are rather large, you might be interested in a
new backup tool I'm writing that can perform fast incremental backups
even on large volumes:

https://github.com/tasket/sparsebak

Its still experimental but could be in beta as soon as December.

--

Chris Laprise, tas...@posteo.net
https://github.com/tasket
https://twitter.com/ttaskett
PGP: BEE2 20C5 356E 764A 73EB 4AB3 1DC4 D106 F07F 1886