I initially shared a my latest project with you here. A lot have happened in LaManager in the past 2 weeks so I though i would make an update post.
Reminder of what is LaManager: It's a services manager build to use copy on write to reduce downtime when doing offline backups while ensuring that all the data is in a coherent state. It can also create and manage it's own virtual disk images to allow it to work anywhere outside of supported COW filesystems.
First LaManager has now been put in production and as been working without issue since then.
Currently on my homelab it manages : caddy (with anubis), forgejo, jellyfin, jitsi, matrix (including frontend and bridges), nextcloud (including euro-office), pi-hole, qbittorrent, redlib and vaultwarden.
Changes and new features since last time :
- NEW FILESYSTEM SUPPORT: ZFS !
- moved development to my forgejo instance
- shell-completion for bash, elvish, fish, powershell and zsh
- mutithreading of operations applied on multiple services (start, stop, backups, restart, remove)
- new restart command for services
- better error handling with anyhow
- added a lockfile to prevent unmounting during backups or others combinations of incompatible operations
- Licensed under AGPLv3
The forgejo instance is open registration to allow contributions.
With all thoses services the total downtime when doing a backup is under 17 seconds and extremely consistent, even when the remote took more than 20 minutes to sync.
For the entire week since introduction of multithreading daily backups downtime never went bellow 16 seconds or above 18.
As i've seen the discussions around about AI, i can confirm that currently LaManager has been fully created without any use of AI.
Can confirm that I don't see that, probably on your end