Clone for large LFS project very slow on windows

Peter Cardwell-Gardner December 1, 2017

Cloning a large repository with ~4gb of LFS files is incredibly slow on windows (sourcetree 2.3.5). There are ~10k files, being pulled from a central repo on the network. 

If I perform `git lfs fetch` from the command line or do a clone in OSX sourcetree it gathers all LFS files in about 10-15 minutes. In comparison souretree on windows took 25 minutes to just get to 15% of files (an order of magnitude slower). Also looking at the output it seems like osx is doing a `git lfs fetch` in the background (can see a summary printed in the command line), whilst windows seems to be doing an operation on every single file. (every single file spawns and logs an entry in the progress window). My understanding is that sourcetree should skip filter smudge when it clones, but it seems this might not be the case on windows? 

From the progress window in sourcetree for windows:

git -c filter.lfs.smudge= -c filter.lfs.required=false -c diff.mnemonicprefix=false -c core.quotepath=false clone --branch master ...



Not quite sure how to interpret if this is correct

1 answer

0 votes
minnsey
Atlassian Team
Atlassian Team members are employees working across the company in a wide variety of roles.
December 5, 2017

Hi

Thanks for the report, I don't have an answer but we can look into it.

Just to clarify, this is from an internal self hosted repository not from an online host like Bitbucket or GitHub?

Peter Cardwell-Gardner December 6, 2017

Yeah, it is internal self hosted (gitlab) on a network LAN. Further fetch operations seem to happen a reasonable speed. It's just the clone that's borked

minnsey
Atlassian Team
Atlassian Team members are employees working across the company in a wide variety of roles.
December 8, 2017

Thanks for the additional information>

Can you post the console window output?

I'm not completely clear how you feel it is different. Sourcetree just issues a single clone command, although the output does include a file by file output.

Suggest an answer

Log in or Sign up to answer
TAGS
AUG Leaders

Atlassian Community Events