XSRF Token during restore

Support
1

So, something odd going on. Using Duplicati - 2.0.5.114_canary_2021-03-10. I have 3 backups on same machine to same machine running daily. If I click restore on 2 of them, a second or two later it shows me a list of backups and the directories. On the 3rd backup, if I click restore, it sits a very long time on “getting file versions” (more than 10 minutes) and eventually gives me a missing XSRF token error and shows a list of backup versions. I presume the error is due to how long it takes to get the file versions list. I have deleted and recreated the database. The backup destination is a remote machine via sftp. Backup actually works fine, just the restore. While the file version list is ongoing, I can see UPLOAD link from my site to the remote machine is saturated. To list file versions, I would have thought it would be download (or from local db), but maybe that somehow makes sense. I don’t know what else to look at. Upload can’t possibly go any faster where I live, max is 3-3.5 mbps, and why should a file list be using so much upload bandwidth?

Both other backups list file versions in a couple seconds. I don’t understand why that might be, backup sizes and versions are about the same across the backups. The sqlite database for the one that fails is 1/4th the size of another backup set that works (to restore). It’s the smallest sqlite file of the 3 backup sets.

Additional info: source is 19GB, backup is 30GB, versions is 19.

UPDATE: This is a non issue as was my fault. I found an error in the after backup job that was doing stuff when it shouldn’t be, didn’t check the operation being done.

2

Glad you figured it out! So you were trying to start a restore while a backup was in progress?

3

Sort of. I have an after backup script. It ends up copying the sqlite files and the config off site. Unfortunately, that script is triggered after ANY operation within duplicati. Well, each time I would poke around in duplicati, it would fire off the job. Multiple jobs even. So, becuase of an issue, I had to rebuild the database (there’s one copy after rebuild is done). Then, I poked around (forgot what) and there was 2 scripts running. Then, I tried to restore just to see and I had 2 jobs already copying stuff out (rclone), which meant 8 files at once. Forgot to change rclone to 1 file at a time also.

Instead of this, the script needs to check what operation. So, my script now starts with this (linux):

if [ "$DUPLICATI__OPERATIONNAME" != "Backup" ]; then
  echo $date " Was not a backup so aborting" >>/home/sfatula/Documents/Logs/AfterDuplicatiHome.log
  exit
fi

The echo was just for making sure it worked when not doing a backup, and it does.

4

Yep - common issue with pre-backup scripts. Glad you found the solution!