I’ve written before about Azure Backup fails due to Content Testing Screenshots not clearing and the resulting problems with backup size.
That’s not the whole problem though.
Sitecore made some effort to configure some directories to be outside the home folder of the app services, and therefore not subject to backup – but there’s still a lot that shouldn’t be in the backup.
At least on our (9.0 Update 2) system, we’ve got:
- Submit Queue (for some reason)
All of which are under D:\local\, which I presume they’re trying to use as temporary storage.
However… there are still problems:
- \App_Data\logs – And the log files can be quite substantial
- A ‘dumps’ folder? – This is, I presume, the memory dump of the App Service. I’m not sure if this was generated automatically or if someone was debugging, but it was 23 GB in size.
- Device Detection Databases – New ones are downloaded by Sitecore from time to time, and each is about 0.5GB. They’re not removed unless you configure the Cleanup Agent.
- Our Unicorn folder – Used for Unicorn sync, the items in here come from source control. We don’t need to back them up.
- The Content Testing Screenshots mentioned before.
So, your backup can still grow, and still be far too big. There’s no reason a Sitecore CM or CD server should exceed 10GB – but ours did, and kept growing.
There is an alternative – both to including things in the backup, and to cramming them in hard to reach places on disk. You can exclude them from the backup.
All you have to do is create a text file called _backup.filter, put each excluded file or directory path on its own line, and drop it into your wwwroot.
It works, and now our backup is a few 400MB, not over 10GB. Crazy.