Cannot configuring DVC cache file link type as hardlink

[tshu-w]

Steps to reproduce

mkdir dvc && cd dvc
git init && dvc init
dvc config cache.type hardlink
touch test.xml && dvc add test.xml
ls -l

got:

-rw-rw-r-- 1 wangtianshu wangtianshu  0 Mar 28 10:38 test.xml
-rw-rw-r-- 1 wangtianshu wangtianshu 73 Mar 28 10:38 test.xml.dvc

test.xml is copied to the cache instead of being hard-linked, even if the system supports it (verified by ln without -s). If I config cache type as soft link, it works fine.

dvc doctor
DVC version: 2.9.5 (pip)
---------------------------------
Platform: Python 3.10.2 on Linux-4.4.0-189-generic-x86_64-with-glibc2.31
Supports:
	webhdfs (fsspec = 2022.2.0),
	http (aiohttp = 3.8.1, aiohttp-retry = 2.4.6),
	https (aiohttp = 3.8.1, aiohttp-retry = 2.4.6),
	ssh (sshfs = 2021.11.2)
Cache types: hardlink, symlink
Cache directory: ext4 on /dev/sda1
Caches: local
Remotes: None
Workspace directory: ext4 on /dev/sda1
Repo: dvc, git

I can make hard link through ln:

ln test.xml test2.xml && ls -l

got:

total 4
-rw-rw-r-- 2 wangtianshu wangtianshu  0 Mar 28 13:55 test2.xml
-rw-rw-r-- 2 wangtianshu wangtianshu  0 Mar 28 13:55 test.xml
-rw-rw-r-- 1 wangtianshu wangtianshu 73 Mar 28 13:54 test.xml.dvc

[pmrowla]

In this example case it is because you ran dvc add before changing the cache link type to hardlink (so the file was copied on dvc add). DVC will not re-link any existing files automatically when you set cache.type.

If you change the cache link type configuration, you need to run dvc checkout --relink if you want to force relinking. https://dvc.org/doc/command-reference/checkout#options

[skshetry]

@tshu-w, can you please share the output of dvc doctor?

[tshu-w]

@tshu-w, can you please share the output of dvc doctor?

DVC version: 2.9.5 (pip)
---------------------------------
Platform: Python 3.10.2 on Linux-4.4.0-189-generic-x86_64-with-glibc2.31
Supports:
	webhdfs (fsspec = 2022.2.0),
	http (aiohttp = 3.8.1, aiohttp-retry = 2.4.6),
	https (aiohttp = 3.8.1, aiohttp-retry = 2.4.6),
	ssh (sshfs = 2021.11.2)
Cache types: hardlink, symlink
Cache directory: ext4 on /dev/sda1
Caches: local
Remotes: None
Workspace directory: ext4 on /dev/sda1
Repo: dvc, git

[skshetry]

@tshu-w, can you please share the output of dvc doctor?

DVC version: 2.9.5 (pip)
---------------------------------
Platform: Python 3.10.2 on Linux-4.4.0-189-generic-x86_64-with-glibc2.31
Supports:
	webhdfs (fsspec = 2022.2.0),
	http (aiohttp = 3.8.1, aiohttp-retry = 2.4.6),
	https (aiohttp = 3.8.1, aiohttp-retry = 2.4.6),
	ssh (sshfs = 2021.11.2)

This does not have repo information. Do you mind cd-ing into the Repo and then invoking dvc doctor? It has more detailed information. 🙂

[tshu-w]

Sorry, update it.

[dberenbaum]

@pmrowla @skshetry Any thoughts on this one?