Share via

Git integration with ML Azure Jupyterhub is very slow

Giovanni GATTI PINHEIRO 21 Reputation points
2020-08-27T07:33:39.823+00:00

JupyterLab (in ML Azure) default mounts a shared file system for all users in my workspace. The documentation recommends to "clone git repositories into our own user directory". It works, but it is extremely slow. A simple 'git status' can take up to 3 seconds to run. Committing, pushing, pulling, stashing, each can take long time to execute (matter of seconds). In a my local machine it takes the order of milliseconds. This can be very annoying, specially if one does so very frequently.

So here is my question: is there a way to fix it or is it inherently bounded to the slow mounted file system? If it can't be solved, how can I change Jupyter's mounting point (so I can point it out to a local file system)?

Thanks,

Azure Machine Learning
Azure Machine Learning
An Azure machine learning service for building and deploying models.
3,337 questions

2 answers

Sort by: Most helpful
Most helpful Newest Oldest
  1. romungi-MSFT 48,911 Reputation points Microsoft Employee Moderator
    2020-08-31T10:03:21.92+00:00

    @Giovanni GATTI PINHEIRO Our team has confirmed this as a known issue while using Azure File Share and are working on a design change to ensure the performance is improved. We currently do not have an ETA on this implementation, we shall update this thread as soon as we have an update.

    3 people found this answer helpful.
    0 comments
  2. Chamal Gomes 0 Reputation points
    2025-04-15T02:34:20.4033333+00:00

    We noticed a short while ago a change was introduced which actually made git run faster on the compute instances, buy afterwards a rollback was made and now it is slower than even before ? Do you have any ETA as to when a stable fix will be introduced ?

    0 comments

Your answer

Answers can be marked as Accepted Answers by the question author, which helps users to know the answer solved the author's problem.