Restored backup with errors but succesfully finished?

I backed up my VM’s on R4.1 and verified the backup as well. (Though verification was done with the backup still in an AppVM, I moved the backup to an external drive after that).
Now I’ve installed R4.2 and restored my backup. It gave several errors, but after that still popped up “Finished succesfully”.
So now I’m not sure if everything is restored succesfully or not. At least two VM’s seem to be working fine.
If somebody can point me to the location of the log file, I can post the errors.

ERROR: unable to extract files for /home/user/QubesIncoming/backup#restore-tqcjezt8/vm38/root.img.118 tar output: gzip: stdin: unexpected end of file
Error extracting data: unable to read the qubes backup file /run/media/user/deviceID/qubes-backup-file (118): b’Starting Restorecopy\nArguments: /run/media/user/deviceID/qubes-backup-file\nPaths: vm19/ vm47/ vm14/ vm17/ vm46/ vm49/ vm20/ vm53/ vm18/ vm21/ vm41/ vm22/ vm23/ vm43/ vm24/ vm32/ vm26/ vm27/ vm28/ vm29/ vm30/ vm44/ vm12/ vm31/ vm11/ vm6/ vm33/ vm34 vm52/ vm35/ vm36/ vm37/ vm38/ vm13/ dom0-home\nPerforming restore from backup file /run/media/user/deviceID/qubes-backup-file\nCopying /run/media/user/deviceID/qubes-backup-file to STDOUT\ntar2qfile: Fatal error: Copying file vm38/root.img.119.enc: Error reading (error type: Input/output error)n’
→ Done.

So what does this mean? vm38 is the only one with a fatal error, so maybe only vm38 is not restored succesfully? And how do I find which vmID numbers correspond with which VM/Qube?

See (my WIP branch of) the emergency restore instructions, up to and including step 7:


I already fail at step 1 :sweat_smile: backup-header is not found

tar:  backup-header: Not found in archive
tar: Exiting with failure status due to previous errors

You can leave out backup-header in the first step, and then in the next steps use backup-header.verified instead.

Thanks for the quick reply!
If I leave out backup-header, I’m getting backup-header.hmac not found.

Very strange. In the previous post it seemed to extract backup-header.hmac successfully?

Yes, that was my impression as well.
And if I leave out backup-header.hmac too, it says qubes.xml.000.enc not found, even though it seemed to extract that successfully as well.

Did you maybe forget the -i argument to tar this time?

As long as e.g. the unmodified step 1 successfully extracts backup-header.hmac and qubes.xml.000.enc I guess it’s fine to ignore the error for backup-header (but still use backup-header.verified in the next steps).

Thanks Rusty! I did add the -i argument. Tried both on Fedora and Debian templates. It seems the problem was the external drive. I performed your instructions on another copy of the backup downloaded directly in a VM and it went fine.
Fortunately vm38 with the fatal error is an unimportant Windows qube.
Do you happen to know if I can assume the other mentioned VMs are intact?

I’m not really sure. But from looking at the “fatal” tar2qfile error message and peeking at its source code, I’d assume that any vmN data that came after vm38 in the backup file has NOT been restored at all.

Check tar -i -tf <your backup file> to get a file listing, maybe you had a bit of luck and all the other vmN files were ordered before vm38. (vm38 is near the end of the “Paths:” line in the error message - with only vm13 after it - but I don’t know if that necessarily corresponds to the order in the backup file.)

1 Like

Thanks again! There were a few VMs after 38, whose contents were not restored. Fortunately none of major importance.