Just in case you’re not sick of hearing about it yet, one of the things I liked about CrashPlan was that if a backup didn’t happen over a certain amount of time I’d get an email letting me know about (the lack of) it.
Would it be viable / useful for Dulicati to do something like send a “something’s wrong” email if no backups had finished (or even started) within the last X scheduled startup events?
For example, if I create a backup job to run daily and ALLOW the machine to go to sleep after say 5 min. of non-use it’s possible the initial backup could never complete.
Having “check completed status” step before actually attempting a backup allows for noticing that no backup in the last 7 days (for example) has ever finished.
Obviously if the machine isn’t turned on for a month we would instantly be in this state so it should maybe it should be tempered by something like if the ratio of the number of COMPLETED backups of the number of STARTED backups is below .25 then send an email saying “Hey - only 25% of your scheduled backups are completing! Check you backup and sleep settings for possible issues.”
My plan for this is offering an external service (like the OAuth handler) where backups report to when they have run. This is the only way to reliably report when backups stop running.
Anything that runs on the machine itself is very easy to break (bad network, broken install, firewall, etc).
As much as possible should run on the client, the less infrastructure that relies on me or others the better. But this is not really a thing that can be done that way as we need to look for a trigger that did not occur.
Maybe you saw it, but someone suggests that they will have such a feature:
Monitoring dashboard for the “family sysadmin” looks like a great idea.
In the meantime I’m just going to run a simple PowerShell script once a day on the remote server to make sure there are recent file timestamps in the destination directory. This will ensure that backups are running at all and combined with the client side emails should cover most failure scenarios I can think of.
Not very elegant, but figure it should do the trick. Happy to share the quick PowerShell once I get to it.
