Skip to content

DVC repro/pull/checkout hangs on sshfs #4259

@geblanco

Description

@geblanco

Bug Report

Please provide information about your setup

Output of dvc version:

$ dvc version
DVC version: 1.1.11
Python version: 3.8.3
Platform: Linux-5.7.7-arch1-1-x86_64-with-glibc2.2.5
Binary: False
Package: pip
Supported remotes: http, https, s3, ssh
Filesystem type (workspace): ('ext4', '/dev/sda1')

Additional Information (if any):

Setup:
Experiments are run in a server with GPUs, with a local directory as cache, setup with: type = "hardlink,symlink". I run everything with dvc repro, then dvc push. It runs as expected and caches all files.
Afterwards, I run evaluation in my local machine, to do so, the cache directory is setup the same, but the local cache directory is mounted through sshfs against the server (I do it this way because of #3815).

The thing is that it only hangs with a certain project and not the rest. My guess is that it has to do with the number of files involved. In the case of the project working correctly there are only 75 files, whereas the one hanging up has 28005 files (RACE dataset is scattered across multiple files).

When running dvc pull or dvc checkout, dvc hangs indefinitely (or at least for >5minutes) without any prompt (-v option doesn't yield any output either).

Best,

Metadata

Metadata

Assignees

No one assigned

    Labels

    Type

    No type

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions