To manage resources in Jupyterhub, you can set user memory and CPU guarantees/ limits. A resource guarantee means that all users will have at least this resource available at all times, but they may be given more resources if they’re available. A resource limit sets a hard limit on the resources available. You can find more details at - zero-to-jupyterhub.readthedocs.io/en/stable/jupyterhub/customizing/user-resources.html#. This way you can manage the resources in Jupyterhub. However our Python and AI machine learning kit comes with default spawner which does not support setting up the guarantees/limits for resources. A Spawner represents an abstract interface to a process and is responsible for starting each single-user notebook server. You can read more about what is spawner and how it works at - jupyterhub.readthedocs.io/en/stable/reference/spawners.html You can create custom spawner and set CPU via "cpu_guarantee" & "cpu_limit" and memory via "mem_guarantee" and "mem_limit" directive. Please find more details at - jupyterhub.readthedocs.io/en/stable/api/spawner.html
If you are referring to Jupyter resources (notebooks etc) created via Jupyter, then each user has his/her own seperate Jupyter home directory for eg. /home//Desktop/python-machine-learning-kit. Refer second line with the value "c.Spawner.notebook_dir" in /home/anaconda/jupyterhub_config.py file for specific location and to change the same. If you are reffering to the computing resources, then it is shared across the users by the VM.
@@techlatestnet2348 Thanks for the reply. 1. So does it mean that each user will have standalone jupyternotebook running on the shared computing resources? 2. If i am running tljh and want to share the computing resources of the server computer (not with the cloud services like Azure, OCean droplet, etc) to all of the users. Do i need to install with Kubernetes? Your reply is highly appreciated. Thank you in advance~
Yes, each user will have separate Jupyter instance on the vm. These instances are spanned and managed by Jupyter Hub. You don't need to have kubernetes to manage the resources, JupyterHub takes care of it.
Perfect. Thank you for this tutorial. Very helpful.
Thanks for the tutorial.. sir do you have this tutorial for dockerized version?
Can u pls tell me how do I run this on Mac? JupyerHub I need it urgeunely
How can I manage resources ? For example I would let Bob use 30% of 10 GPU, but for Alice 60%?
To manage resources in Jupyterhub, you can set user memory and CPU guarantees/ limits. A resource guarantee means that all users will have at least this resource available at all times, but they may be given more resources if they’re available. A resource limit sets a hard limit on the resources available.
You can find more details at -
zero-to-jupyterhub.readthedocs.io/en/stable/jupyterhub/customizing/user-resources.html#.
This way you can manage the resources in Jupyterhub. However our Python and AI machine learning kit comes with default spawner which does not support setting up the guarantees/limits for resources.
A Spawner represents an abstract interface to a process and is responsible for starting each single-user notebook server. You can read more about what is spawner and how it works at -
jupyterhub.readthedocs.io/en/stable/reference/spawners.html
You can create custom spawner and set CPU via "cpu_guarantee" & "cpu_limit"
and memory via "mem_guarantee" and "mem_limit" directive.
Please find more details at -
jupyterhub.readthedocs.io/en/stable/api/spawner.html
are the users sharing the same resources?
If you are referring to Jupyter resources (notebooks etc) created via Jupyter, then each user has his/her own seperate Jupyter home directory for eg. /home//Desktop/python-machine-learning-kit.
Refer second line with the value "c.Spawner.notebook_dir" in /home/anaconda/jupyterhub_config.py file for specific location and to change the same.
If you are reffering to the computing resources, then it is shared across the users by the VM.
@@techlatestnet2348 Thanks for the reply.
1. So does it mean that each user will have standalone jupyternotebook running on the shared computing resources?
2. If i am running tljh and want to share the computing resources of the server computer (not with the cloud services like Azure, OCean droplet, etc) to all of the users. Do i need to install with Kubernetes?
Your reply is highly appreciated. Thank you in advance~
Yes, each user will have separate Jupyter instance on the vm. These instances are spanned and managed by Jupyter Hub. You don't need to have kubernetes to manage the resources, JupyterHub takes care of it.
@@techlatestnet2348 So meaning users jupyter notebook on jupyterhub will be using the resources on the server computer?
@@tatchan837 Yes
This launches juputerlab and not jupyterhub