seagate / cloudfuse Goto Github PK

Problem

I have a relatively large number of files (about 2,000) in the root directory in my container. When I first mount the container, listing the root directory sometimes takes an excessive amount of time. At the moment it's 5s. But it's been over 20s at times.

Expected behavior

Listing 2,000 files should take much less time. I think 1s is ideal, 5s is reasonable, and 10s is not.

How to reproduce

List a directory with a few thousand files right after mounting:

cloudfuse mount ~/mycontainer --config-file=config.yaml
time ls ~/mycontainer
cloudfuse unmount --lazy ~/mycontainer

Logs

cloudfuse_list_2k.log

Deployment information

I'm on Ubuntu 22.04, running against a Seagate Lyve Cloud container, running cloudfuse v0.3.0.

Additional information

No response

Please describe the feature/improvement you would like to see added.

I'm trying to mount a bucket and keep my local cache valid without any timeout.
I tried to set it to 0, thinking it would be a special value meant to mean "unlimited", but cloudfuse takes 0 to mean that all cache items should be deleted as soon as possible.
I'm working around this by setting timeout-sec to a really large number (64 million, which is about 2 years), but that doesn't seem very user friendly. A separate option would be nice.

Please explain why you think adding this feature/improvement is important.

If a persistent local cache is a common use case, it should be easy to configure.

Please describe the feature/improvement you would like to see added.

Mounted buckets are remounted at login on Windows. It would be nice to have the same functionality on Linux.

Please explain why you think adding this feature/improvement is important.

Linux users would be more likely to be able to set up a process to remount their containers on their own, but that's not a guarantee. I see no reason why we should be less friendly to Linux users.