Update to version 3.9.0 and follwing versions not working

[snorre-k]

There are now two issues related to version 3.9.0, which seems to be broken running in the container: #24, #25

I also tried a test and checked, that the version 3.9.0 has troubles running in the container. But up till now I have no clue, what is happening.
I will try to get this resolved.

Please use version 3.8.0 in the meanwhile:

• snorre0815/idrive-docker:3.8.0
• ghcr.io/snorre-k/idrive-docker:3.8.0

[snorre-k]

The problem seems to be a change in the configuration of the client between version 3.8.0 and 3.9.0.
This change to the user profile (user configuration) seems to be applied by the clients own update mechanism.
To get everything working again, I had to:

• Enter the running container (I did it with running version 3.9.0)
• Do an update of the client: ./idrive --check-update
  • You will get logged out (but settings are kept)
  • Do an update although you are already on the latest version
  • Accept License
  • Confirm to be logged out
• Login again with your username and password
• Check that settings are OK in WEB GUI

I hope this helps to update to the new version 3.9.0.

[mayermatt]

I'm currently digging into what seems to be the same issue.

I upgraded straight to 3.11.0 and started seeing this error when running any idrive command:

json parsing failed in ValidateAndGetJson

When running the base ./idrive command, I get both of these errors:

json parsing failed in ValidateAndGetJson
Compilation failed in require.

I have tried downgrading the container to 3.9.0 and then running ./idrive --check-update. That allowed me to at least start the update process, which is much more progress. But once I accepted the terms of service, I got stuck here:

Enter 'A' to accept and install or 'C' to cancel: A

The installation may take some time. Kindly wait...
Checking installation dependencies... OK
Extracting installer... OK
Checking compatible binary... OK
Running package update...
/root/1757122672_idriveforlinux.bin: line 712: json: command not found

Still trying to figure out next steps other than nuking configuration and starting over; if you happen to have any other recommendations, I'd appreciate it.

Subsequently downgraded to 3.8.0 and followed the same steps as above. Same results.

Then upgraded to latest to see if the failed updates affected behavior. Still got these same errors:

json parsing failed in ValidateAndGetJson
Compilation failed in require.

[snorre-k]

@mayermatt You seem to have a bad configuration (maybe introduced with an update). You should be able to clean the config by deleting the volume (or the bind mounts) and starting over. When you use the same backup location (asked when logging in the first time), your settings including sources and schedules should be restored (at least they did when I tested this with 3.10.0 and 3.11.0).

[mayermatt]

I'm considering my issue resolved -- posting the below for posterity.

I spent some time digging into this, but since the issue was caused by an upgrade, it was tricky to reproduce (and not worth the hassle of clearing all config, downgrading, performing setup, then upgrading). I am running this on two Linux boxes, and both hit the same issue, so it wasn't a fluke. I was using bind mounts instead of a docker volume throughout all of this.

Even after purging all configuration for my container (bringing down the container, removing the host directory that backed my bind mount, rebuilding the idrivecrontab.json file, and bringing things back up), starting with 3.11.0, the container continued to have major stability issues -- nonstop 100% CPU usage, and runaway disk usage (200 GB+ before I shut it down) by filling up the /opt/IDriveForLinux/idriveIt/user_profile/root/<profile name>/CDPDBDUMP directory inside the container, inside one of my bind mounts. Occurred consistently across multiple machines (running different OS's and different server OS upgrade cadence, so very few common factors between the machines) and multiple configuration resets.

I later decided to abandon the bind mounts and go back to docker volumes. And it's been smooth ever since. The benefit I was getting from bind mounts was benign -- I had hoped to use the bind mount so I could back up some of the basic IDrive configuration, such as the backup set / exclude set.

Never did figure out the root cause, but volumes seem like they're likely going to be more stable by nature so I'll stick with them.

Original (broken) docker-compose configuration:

  idrive:
    container_name: idrive
    hostname: <my custom hostname>
    image: snorre0815/idrive-docker:latest
    restart: unless-stopped
    deploy:
      resources:
        reservations:
          memory: 512m
        limits:
          memory: 1g
    # Do not have two different IDrive containers with the same MAC addresses!
    mac_address: <custom MAC address>
    volumes:
      # Host directory to back up (read-only)
      - "<backup directory in host>:<backup directory in container>:ro"
      # Config
      - <config path>/cache:/opt/IDriveForLinux/idriveIt/cache
      - <config path>/user_profile:/opt/IDriveForLinux/idriveIt/user_profile
      - <config path>/idrivecrontab.json:/opt/IDriveForLinux/idriveIt/idrivecrontab.json

Working configuration:

  idrive:
    container_name: idrive
    hostname: <my custom hostname>
    image: snorre0815/idrive-docker:latest
    restart: unless-stopped
    deploy:
      resources:
        reservations:
          memory: 512m
        limits:
          memory: 1g
    # Do not have two different IDrive containers with the same MAC addresses!
    mac_address: <custom MAC address>
    volumes:
      # Host directory to back up (read-only)
      - "<backup directory in host>:<backup directory in container>:ro"
      # Config
      - idrive:/opt/IDriveForLinux/idriveIt

[kingfetty]

FYI, I got it to work with latest version. I had to launch an interactive bash terminal into the container and reconfigure my account.
docker exec -it idrive /bin/bash
idrive -a
Select Yes to reconfigure account.