handle sym links in when calculating repo size

[leerspace]

repo/fsrepo: follow symbolic links when calculating RepoSize

Now if the IPFS path is a symbolic it will be followed before
calculating the repo size.

License: MIT
Signed-off-by: John Reed john@re2d.xyz

[leerspace]

Fixes #4056

[leerspace]

I'm confused by the GitCop failures since the commit already has the license and signed-off-by trailers.

[whyrusleeping]

@leerspace sometimes it gets upset because it doesnt perfectly match its regex. If you have the right content, then just ignore it.

[Kubuxu]

If it isn't too much to ask for, a sharness test for it would be great.

[leerspace]

I started working on the sharness test, but now I'm trying to figure out why every time I run ipfs repo stat the RepoSize gets larger by about ~1kB. This is even with no daemon running.

edit: this might be getting outside the scope of this PR a bit, but based on the output of find -type f | xargs md5sum in IPFS_PATH every time I run ipfs repo stat with badger it's updating a badgerds/*.vlog file, creating a new badgerds/*.sst file, and updating the badgerds/MANIFEST file.

With leveldb/flatfs, it creates a new datastore/*.ldb file and updates a datastore/*.log file.

I'm not sure if this is desired behavior (I'm guessing not), but either way I'll have to think of a good way to write the tests when the output of ipfs repo stat is unstable.

[Kubuxu]

@leerspace this is the behaviour of those DBs, every time they are opened they create new logfile and if there are X of them (or other condition) they compact. I would suggest adding few hundreds of KiB to ipfs in sharness and checking if size is larger than X.

[Kubuxu]

Also according to go fmt this code is not formatted correctly.

[whyrusleeping]

@leerspace starting from a fresh repo (as in a test) should yield the same size every run (we have tests that test this). I would just pick a number, run the test, see what it actually was in the test failure output and then adjust the value in your test assertion.

[whyrusleeping]

Though I definitely agree that running ipfs repo stat with no daemon running should not affect the size of the repo.

[leerspace]

I just committed the gofmt fix and a sharness test for this.

[whyrusleeping]

@leerspace looks like you might not have set the executable bit on that test file. the test runners arent very happy trying to run it

[leerspace]

@whyrusleeping sorry about that. I think it should be fixed now. I did this on Windows, but fortunately there's a way to set this on Windows.