dotrun uses a large amount of disk space about dotrun HOT 5 CLOSED

Done: #29. Thanks @nottrobin!

from dotrun.

After doing some investigation I've removed some old snap versions and also run dotrun exec yarn cache clean. This has reduced the snap to:
451M /snap/dotrun
1.2G /home/multipass/snap/dotrun

Edit:
After a day of working this is now:
2.2G /home/multipass/snap/dotrun

from dotrun.

I'm confused as to why the /home/multipass/snap/dotrun folder itself would grow in size, since snaps are immutable. The snap is largish, as in it includes a lot of dependencies with it. Overall, though, this should save a lot of disk space over the old ./run solution where we were keeping many copies of large docker images for every project.

The project folders themselves will grow a lot based on yarn and python dependencies. This is unavoidable.

If you discover anything more about any ways you think dotrun is being inefficient with space, please let me know. But I think I'll close this for now as there doesn't seem to be a clear action.

from dotrun.

@nottrobin does the snap keep its own caches for yarn? It seems like it does by me running dotrun exec yarn cache clean and that snap dir reducing in size. I wonder if we need to be able to clean up python/js caches as these will only grow in size as deps update and the old versions live in the cache.

Now that I know to do it I will run dotrun exec yarn cache clean periodically, but it'd be nice to have some kind of dotrun clean-cache or something that cleaned up all the different caches it had.

from dotrun.

Ah right, yeah that's probably what's going on. I realised /home/multipass/snap/dotrun will actually only contain the snap user storage directories, not the snap itself, so that makes sense.

I think in general the existence of that cache is a good thing as it will significantly speed up yarn installs. For most people I imagine they'd happily sacrifice a couple of GB for faster yarn installs. But certainly it makes sense to help people clean it up if needed.

Yeah dotrun clean-cache seems like a reasonable feature request. Do you think you could file that as a separate issue to keep things tidier? Or maybe we could call it purge? Or prune like Docker?

from dotrun.