Could we look at making it a requirement for there to be no nodes in an application before it is deleted? Or at least make it much more apparent that you are deleting an application vs a Node. I just deleted an application with 30+ nodes. I know I should have had a backup but we just moved the server and had not got it set up.
I believe this is already there
Could we look at making it a requirement for there to be no nodes in an application before it is deleted?
I’m afraid that if we do this, that somebody else will request a feature to automatically delete all devices when deleting an application because he has 1k devices which is a pain to remove one by one…
brocaar, I think this is a bad argument.
Mass deletion should never be ‘easy’. That’s a very poor model. An accidental deletion can’t be undone without either restoring from backup or manually touching every devices.
Further, when you do delete, Chirp is actually fully deleting those records from postgres. Not marking for deletion, or marking inactive, GONE. Ie, not-recoverable.
When someone deletes an application, it needs to be more ‘aggressive’ in the verification. Ie, type “DELETE APPLICATION AND ALL DEVICES” typed out before you can hit OK. This is a massively destructive operation and it’s far far far too easy right now.
As far as deleting devices, this could be handled much more gracefully. Allow the deletion of an application (really though, mark it inactive…) and then go into devices and have an option to delete unattached devices or assign unattached devices to an application.
Just putting the required typing in is a duct tape fix for both scenarios. The separate delete applications and delete/handle orphaned devices solves both scenarios and improves on them greatly.