Joplin-backup fails to start cron

[ilyashupta]

Container fails with this line in logs:
time="2025-01-04T02:16:21+01:00" level=fatal msg="Failed to fork exec: no such file or directory"

It looks like this line throws it:
exec supercronic -passthrough-logs -quiet "${CRON_CONFIG_FILE}"

If I run manual backup like this docker compose run --rm -it joplin-backup backup - it works successfully, just the cron part is not working when container is being launched from a compose file.

My backup.env file looks like this:

RCLONE_REMOTE_NAME=secret
RCLONE_REMOTE_DIR=secret
CRON=@daily
TIMEZONE=secret
ZIP_PASSWORD=secret
BACKUP_KEEP_DAYS=31

[AdrienPoupa]

Good catch, thanks for the report! I was able to piece out what happened.

First, this is an upstream bug in supercronic starting from v0.2.33 that prevents it from executing due to paths mismatch.

My backups stopped working from Dec, 9 and that coincides with an automated build here: https://github.com/AdrienPoupa/rclone-backup/actions/runs/12247358086/job/34165041361

The base image changed from Alpine 3.20 to 3.21 (which was released on Dec, 5: https://alpinelinux.org/posts/Alpine-3.21.0-released.html) and that upgraded supercronic from 0.2.29 to 0.2.33, which has the bug.

In rclone-backup, I've now disabled the scheduled publish job so it will only build the latest tag when there's a commit on master, and not overwrite existing tags. I've also changed the base from alpine:latest to rclone/rclone:1.68.2, that still uses Alpine 3.20 and supercronic 0.2.29, I was able to confirm this fixes the bug.

I've published latest and released a new version, hopefully by the time there's a new Rclone release it will use an updated version of supercronic, but there should not be any new automated builds in the meantime.

Can you pull the images again to make sure it works as expected? (docker compose pull && docker compose up -d)

[ilyashupta]

Pulled images and recreated containers - joplin-backup runs fine again. Thank you, really appreciate your work!