Something for the FAQ: How do I stop Kopia using so much local disk space?

I find that kopia (version 0.15.0) uses almost seven gigabytes of disk space - and not in the cloud, but on my system disk. This surprises and indeed somewhat alarms

I, and the FAQ, could do with some information and tips about amelioration.

I did find a post that gave to think that the following command might reduce the local disk usage:

kopia blob gc --advanced-commands=enabled --delete=yes

That gave the following welcome output.

Looking for unreferenced blobs...
Deleted total 32 unreferenced blobs (373.6 MB)

But that leaves more than six gigabytes . . The same post suggested kopia maintenance info. The output suggests that some maintainance had been occuring. But I rn kopia maintenance run --full anyway. So doing freed 87.5 MB.

EDIT: In the documentation (not the FAQ) I found this page. The page mentioned kopia cache clear. Running that command reduced current disk usage by some 5 GB. The same page imparted how to move the cache. I will be moving the cache off my system drive.

It looks like you are backing up to your local disk. Not sure why you expect that it won’t use whatever space it needs.

EDIT - actually I am wrong. It is rather local cache

It is local cache and it makes kopia run much faster by storing some data locally and avoiding repeated downloads from remote storage. Not sure how much data you backup but 7GB cache it nothing unusual. I have repos using 30GB+ of cache.

More info here - Caching | Kopia

If it is a problem for your system you can:

  1. Periodically clear cache:

kopia clear cache

  1. by default there are no hard limits of max space used by cache. To see cache parameters run kopia cache info. You can then change it by using kopia cache set.

Less cache you have slower kopia will be. I would not tinker with it unless you have system with very limited local storage.

kapitainsky: understood. Yet, as I said: an entry for the FAQ?

Yeah good idea.

Feel free to submit PR. Here you are github source to edit

As it happens, I know - just about - how to create a pull request. Yet, evidently, I know little about the topic at issue. So I will not myself be creating the FAQ entry.

No problem. It is open source project - if somebody in community will feel it is needed then will add it. Personally not sure what it would add beyond already documented details.

Personally not sure what it would add beyond already documented details.

It would have a chance of arresting the following chain of events.

  1. Some user has Kopia doing cloud backups.
  2. The user notices large(ish) local disk usage.
  3. The user consequently feels that s/he has a problem and perhaps that the software is malfunctioning.

Admittedly, a moderately close read of the documentation would prevent the move from 2 to 3 (or prevent the user from remaining at 3). So too though, and I think more effectively, would a FAQ entry. Take my own case. I went through steps 1-3, and stayed at 3 - until posting on this forum and receiving answers. Yet, before posting, I had checked the FAQ . .

Not sure - but I might be wrong.

I imagine users in general realise that many software use caches. My Firefox cache is few GB too. The same like kopia it is located in OS recommended location so does not interfere with anything, e.g. OS backup. My cache location is about 50GB (all together, not just kopia). I would be actually surprised if kopia used too little cache - at the end bottleneck is always cloud.

In addition default soft limit for kopia cache is about 10GB. It is not amount most people care about.

Anyway - you think it would be better to clarify things in FAQ? Then please do. Documentation can be always made better.