My experience is that it works but it will show no progress and in many cases also not when it’s finished. So you always need some hacks to manage it. It would be nice, when this important feature would be more reliable.
I completely agree with scw2wi. There’s no message to track the backup’s progress. A completion message like ‘success’ would be the bare minimum. That’s why I create a clone of my SSD under Windows at crucial moments.
If jobs info is already available to monitor restore progress, it seems like that could be exposed on the restore page with minimal effort until a prettier solution can be designed.
1 Like
True. There is no progress or status.
@mlindstrom
If you have a feature request then those are added here: home-assistant · Discussions · GitHub
In this case the request already likely exists so thumbs up vote it to show support.
Thank you, I’ve added my vote for Add progress bar when restoring a backup · home-assistant · Discussion #1310 · GitHub and created Backup+restore separation for configs vs history · home-assistant · Discussion #2218 · GitHub
2 Likes
Be interested to know exactly how these massive backup sizes actually come about - if it’s a huge DB then that probably needs addressing by
- moving the DB outside of HA - assuming you have something running 24/7, a NAS, another RPi, server etc, install MariaDB and use that for the DB - it won’t be included in the backup then which will speed it up considerably specially if you’re running these mega HA installations on a RPi. (If you want to backup the DB as well, run a separate process on the machine hosting the DB to do this - scheduled task/crontab etc - you can automate uploading it to eg: googledrive or idrive as well) - this will also be more robust specially is using an SD card on a RPi.
- once your using a proper DB (not the HA light version), it should be easy to see what’s causing any DB bloat - work through to eliminate anything you don’t need a history for or reduce the resolution where you do want a history but not, eg 5 second updates (eg: what’s the point in having 5 second updates [28,000 daily records] for energy consumption or import - it adds nothing to analysis of the figures.
For integration based sensors where you can’t amend the resolution, exclude the original sensor from the database and have a trigger based sensor that mirrors the integration sensor, say every 5 minutes rather than 5 seconds - that’ll reduce the DB entries by a factor of 60.
Once all the noise has been filtered out, wait for your retention period to elapse then run an optimise command on the database (suggest with HA turned off) - it’ll remove all the blank space and you’ll be left with a clean and lean structure - it’ll save disk space, reduce disk thrashing and ultimately make any graphs in HA resolve a lot quicker.
I decided to just not backup the DB going forward…I can live without it if I need to restore in the future.