dvc: [dvc pull version >= 3.24] ERROR: unexpected error - failed to load directory
Bug Report
pull: ERROR, doesn’t create the cache directory and crash.
Description
Collecting |0.00 [00:00, ?entry/s]
Fetching
ERROR: unexpected error - failed to load directory (‘60’, ‘36c8668869290419aec048f26f8deb.dir’): [Errno 2] No such file or directory: ‘/mnt/data2/users/myuser/project/.dvc/cache/files/md5/60/36c8668869290419aec048f26f8deb.dir’
With DVC version 3.23.0 it works OK! DVC >= 3.24 fails
Environment information
Output of dvc doctor
:
$ dvc doctor
DVC version: 3.24.0 (pip)
-------------------------
Platform: Python 3.10.12 on Linux-5.15.0-84-generic-x86_64-with-glibc2.35
Subprojects:
dvc_data = 2.18.1
dvc_objects = 1.0.1
dvc_render = 0.6.0
dvc_task = 0.3.0
scmrepo = 1.4.0
Supports:
http (aiohttp = 3.8.6, aiohttp-retry = 2.8.3),
https (aiohttp = 3.8.6, aiohttp-retry = 2.8.3)
Config:
Global: /mnt/data2/users/myuser/.config/dvc
System: /etc/xdg/dvc
Cache types: hardlink, symlink
Cache directory: ext4 on /dev/nvme2n1
Caches: local
Remotes: https
Workspace directory: ext4 on /dev/nvme2n1
Repo: dvc, git
Repo.site_cache_dir: /var/tmp/dvc/repo/fb1190580eb02ee00297b6211011d5af
About this issue
- Original URL
- State: closed
- Created 8 months ago
- Reactions: 6
- Comments: 21 (15 by maintainers)
Thanks for the feedback @tonycusackData! Let’s close this then and we can reopen if needed.
Was having the same issues above. Deleting everythin in
Repo.site_cache_dir
didn’t work but this finally fixed it -pip install dvc[gs]==3.38.0
Thank you!
@PythonFZ @themaikelman Mind giving 3.38.0 a try?
@efiop Should we try to bump the path for
site_cache_dir
so it forces a new cache for existing repos?For me, this is also happening in a GitHub CI runner. The
site_cache_dir
should only be created upon starting the runner. The CI was usingurl = https://dagshub.com/PythonFZ/IPS-Examples.dvc
. Furthermore, I’ve seen this on a local machine with a S3 remote, but I can’t excactly tell you how I fixed it then. Now I’ve just pinned the DVC version to 3.23.Hm, we’ve changed index cache key a few times since then, so it shouldn’t pick it up in newer versions. It is possible we are handling this particular case badly, directiories that failed to load should not be marked as loaded thus we should be making an attempt to load them next time. Need to check, maybe there is another bug there somewhere…
@efiop I think it’s an index issue we fixed already (maybe in https://github.com/iterative/dvc/issues/9785?) but for people who have an index that predates that change, they will still run into this issue, so forcing the index to be regenerated should resolve it. WDYT?
I’ve encountered the same issue. I tried to pull data but forgot to set the credentials to the S3 remote in
config.local
. After setting the credentials, I encountered the same error. The suggested fix could resvolve it.EDIT: I only saw this issue and commented on it. This is my
dvc doctor
which is not the newest versionDeleting the directory of
Repo.site_cache_dir
does resolve the problem.Also, could you each try to drop everything in the
Repo.site_cache_dir
location and see if the problem persists?@atekoa Please run
dvc pull -v
and post full log.