I stumbled upon Jupyter a year ago while looking for auto-grader systems. I found one called "nbgrader" which had as a feature "Integrates with Jupyter notebooks". My first question was "What's a Jupyter Notebook?" and down the rabbit hole I went. Awesome stuff. Great video on how to use Google Cloud and thanks a ton for the github setup. I've done it manually in a variety of ways but yea it is sweet to have it scripted by someone who knows it way better than I do.
Video Sections (hot links): What is Jupyter? - 0:08 Google Cloud & Kubernetes - 3:06 Launch JupyterHub - 8:14 io Online - 11:07 JupyterHub Review - 18:11 How to set up Jupyter for small labs, Easier than JupyterHub! ruclips.net/video/eCbelBE9g7Y/видео.html Github: github.com
@Elliot, I have a setver with ubuntu 22, im interested in to install/use io-app to share documents with other jupyterhub users, but your description is based on kubernetes, do you have another alrternativa to use with ubuntu?. Beste Regards
Hello, can you tell me for what version of python your io installer is valid? I am executing your Wellcome to io.ipynb in my jupyterhub deployment, using a python 3.8 kernel, and I get this error: ImportError: cannot import name 'str_to_bytes' from 'IPython.utils.py3compat' ... Thank you
Wow what a time efficient, down to earth, with a quick concrete intro, and practical tutorial! I love it! :D / @consideRatio from the JupyterHub team, maintainer of underlying tech in video.
Thank you for such kind words! It’s wonderful to see the Jupyter team liking these videos. Thanks for your contributions to JupyterHub and zero-to-jupyterhub. That tutorial was a lifesaver when I was a TA for online classes. Through my videos I hope to make Jupyter visible to a wider audience and contribute to the projects mission. Please let me know if there’s anything I can do to help in future Jupyter developments!
Yes, I see why this is useful. I probably won't have time to do a video on this, but here's some really useful pointers. Running a JupyterHub Kubernetes cluster on AWS is more involved. Google created Kubernetes so they're the easiest. There's an awesome tutorial that gives step by step instructions for setting up JupyterHub on AWS (see link). Also, Amazon recently made it possible to run Kubernetes on AWS (see link), so I'd also check that out. If u only need JupyterHub for a small group of people (5-30) u could get away with just running JupyterHub on an ec2 instance (see link). This would be an incredibly easy (albeit inefficient) solution! AWS in JupyterHub: zero-to-jupyterhub.readthedocs.io/en/latest/amazon/step-zero-aws.html AWS Kubernetes: aws.amazon.com/eks/ JupyterHub Docker for ec2: hub.docker.com/r/jupyterhub/jupyterhub
Great stuff! I only have one suggestion. I noticed that you "talked" out your github nice and carefully, while it would have been easier to just run through it and put the link in the description and let us know that. But that's a pretty small thing overall considering how much awesomeness you packed into this. :D
Thanks! Ur right, I could have just shown Github. I also just added a Github link in the description. I'll do a future tutorial where I really dive into how to use Github.
@@elliott3165 Honestly, I would rather see an in depth guide for how you use this to install and use nbgrader. I could not get it to work. It would show that the extensions were installed but it was still not showing up in my notebooks list. (edit: nevermind lol..... after doing some digging and learning more about how it works, kubernetes and nbgrader are apparently not compatible because of the way that the files operate (although I only kept digging because you mentioned nbgrader in the video, so I'm puzzled). I would really (REALLY) appreciate any suggestions or alternatives if you're aware of any. Thanks again, I'm really liking io the more I explore it! )
Thanks for adding such a great video. Unfortunately, at 10:10, my installation failed with "Error: validation failed: namespaces "jhub" not found". I couldn't find the error online and couldn't resolve it, so I had to abandon your tutorial. If you have any idea, what the issue could be, I'd be very thankful. Unitl then, cheers :)
can you help me. I have error in "Welcome to io Online! " NameError Traceback (most recent call last) in ----> 1 display(HTML("")) 2 Git.b.signIn NameError: name 'HTML' is not defined
Hi, is it possible to use your own local machine as the server instead of resorting to cloud services? Can I assign an IP address to it that is reachable via HTTP/HTTPS?
Should the number of nodes be equal to the number of max users at a given time? i.e. if 7 users would always be using the jupyterhub, should the nodes be equal to 7?
Great question! You do not need to have number nodes equal to max users. Nodes can be shared by users. In Jupyterhub's settings (link below) you can actually decide the much computing power each user gets. I think the default it 1/2 a cpu. zero-to-jupyterhub.readthedocs.io/en/latest/user-resources.html
Great question! The "private" section is when u want to back up ur files on github but don't want others to see them. I'm currently updating io online application. When I'm done I'll do a tutorial on how this works. Hang tight!
What about just making a docker image? one command and it can literally run everything.. well almost one command.. still gotta install docker but thats not hard..
Great Question! So you can run jupyterhub inside a docker image (see link), however this has a major limitation. A docker image is a prepackaged OS that runs on ONE machine. U can set this docker image up on Compute Engine, but you have to decide how many cpus you want to use up front and the cpus don't have the "scaling" feature that Kubernetes has (see 5:09). This ends up being super inefficient, and if u have a group of 20 or more users u waste tons of money, it's like 10x more expensive. Right now the closest thing there is to a "Kubernetes Docker" is Google's Tiller software, which is still rather complex to use. I tried automating it as best I could, but yes there's still definitely room for improvement! JupyterHub Docker: hub.docker.com/r/jupyterhub/jupyterhub
great. what about installing jupyterhub on a local server accessible only to the workgroup LAN ? it has no Kubernetes, just one VM with decent capacity to run jobs for a bunch of colleagues
:) For that I think it's actually WAY easier. U can use a docker image of jupyterhub which u can then spin up on ur local computer (JupyterHub image attached below). hub.docker.com/r/jupyterhub/jupyterhub
So if you want to change the look of Jupyter, I can think of 3 ways to do this: 1. check out Jupyter Themes (see link). This'll change the notebooks color/look 2. Launch jupyter lab, go to: "your_jupyter_website"/lab 3. Manually change the look of jupyter by messing with it's html style code. Check out the io_View app for examples of this. Hope that helps!
Depends on what ur referring to mean. Regarding Programming Languages, both the Java and C++ packages do a kinda magical thing where the code is interactive (like python). I don''t know how it's done but it's really cool! (see link to learn more). As for the time it takes to launch JupyterHub, lol, yah I did cheat a little and speed past the installation process. :) github.com/SpencerPark/IJava
😂 Yes, ur right u I can't call cloud computing new tech. That said, new WAYS of using cloud computing are being released rapidly by Google and Amazon. For instance the Kubernetes tech that this video relies on was made with 2014 (lol... still not really new). And some of the latest examples are Google's MemoryStore and Cloud Functions.
I stumbled upon Jupyter a year ago while looking for auto-grader systems. I found one called "nbgrader" which had as a feature "Integrates with Jupyter notebooks". My first question was "What's a Jupyter Notebook?" and down the rabbit hole I went. Awesome stuff.
Great video on how to use Google Cloud and thanks a ton for the github setup. I've done it manually in a variety of ways but yea it is sweet to have it scripted by someone who knows it way better than I do.
Video Sections (hot links):
What is Jupyter? - 0:08
Google Cloud & Kubernetes - 3:06
Launch JupyterHub - 8:14
io Online - 11:07
JupyterHub Review - 18:11
How to set up Jupyter for small labs, Easier than JupyterHub!
ruclips.net/video/eCbelBE9g7Y/видео.html
Github: github.com
The iPad Pro is a must watch! I love your easy explanations and the fact that you follow-up by answering questions. Kudos!
love the way you talk and friendly smile
@Elliot, I have a setver with ubuntu 22, im interested in to install/use io-app to share documents with other jupyterhub users, but your description is based on kubernetes, do you have another alrternativa to use with ubuntu?.
Beste Regards
Hello, can you tell me for what version of python your io installer is valid? I am executing your Wellcome to io.ipynb in my jupyterhub deployment, using a python 3.8 kernel, and I get this error: ImportError: cannot import name 'str_to_bytes' from 'IPython.utils.py3compat' ... Thank you
Wow what a time efficient, down to earth, with a quick concrete intro, and practical tutorial! I love it! :D
/ @consideRatio from the JupyterHub team, maintainer of underlying tech in video.
Thank you for such kind words! It’s wonderful to see the Jupyter team liking these videos.
Thanks for your contributions to JupyterHub and zero-to-jupyterhub. That tutorial was a lifesaver when I was a TA for online classes. Through my videos I hope to make Jupyter visible to a wider audience and contribute to the projects mission. Please let me know if there’s anything I can do to help in future Jupyter developments!
Please do a tutorial to use AWS rather than Google?? Fingers crossed 🤞 awesome channel btw
Yes, I see why this is useful. I probably won't have time to do a video on this, but here's some really useful pointers. Running a JupyterHub Kubernetes cluster on AWS is more involved. Google created Kubernetes so they're the easiest. There's an awesome tutorial that gives step by step instructions for setting up JupyterHub on AWS (see link). Also, Amazon recently made it possible to run Kubernetes on AWS (see link), so I'd also check that out.
If u only need JupyterHub for a small group of people (5-30) u could get away with just running JupyterHub on an ec2 instance (see link). This would be an incredibly easy (albeit inefficient) solution!
AWS in JupyterHub: zero-to-jupyterhub.readthedocs.io/en/latest/amazon/step-zero-aws.html
AWS Kubernetes: aws.amazon.com/eks/
JupyterHub Docker for ec2: hub.docker.com/r/jupyterhub/jupyterhub
No longer works from what I can tell. Once you get to the ./Check_io_Hub.sh, it never receives an IP address.
Thanks for the tutorial, it's very interesting. Do you know if Jupiter hub is compatible with SSO?
The installation doesnt work for me.... it says:Error: failed to download "jupyterhub/jupyterhub" (hint: running `helm repo update` may help)
What are you using for your presentations? iPad Pro but what software to integrate it into the video?
How to set jupyterhub on Ubuntu server ? For website outside access?
I cannot set config file correctly... :(
Great stuff! I only have one suggestion. I noticed that you "talked" out your github nice and carefully, while it would have been easier to just run through it and put the link in the description and let us know that. But that's a pretty small thing overall considering how much awesomeness you packed into this. :D
Thanks! Ur right, I could have just shown Github. I also just added a Github link in the description. I'll do a future tutorial where I really dive into how to use Github.
@@elliott3165 Honestly, I would rather see an in depth guide for how you use this to install and use nbgrader. I could not get it to work. It would show that the extensions were installed but it was still not showing up in my notebooks list. (edit: nevermind lol..... after doing some digging and learning more about how it works, kubernetes and nbgrader are apparently not compatible because of the way that the files operate (although I only kept digging because you mentioned nbgrader in the video, so I'm puzzled). I would really (REALLY) appreciate any suggestions or alternatives if you're aware of any. Thanks again, I'm really liking io the more I explore it! )
On jupyterhub intial web homepage what we will enter credentials??? Is that credentials already setup or its like creating account
Why not use CoCalc platform instead?
Thanks for adding such a great video.
Unfortunately, at 10:10, my installation failed with "Error: validation failed: namespaces "jhub" not found". I couldn't find the error online and couldn't resolve it, so I had to abandon your tutorial. If you have any idea, what the issue could be, I'd be very thankful. Unitl then, cheers :)
I have the same issue and I don't know how to continue :(
This video is great! Thanks a bunch!
Thanks! How do we set up JupyterLab interface? In addition to classical Jupyter Notebook interface I want to be able to use the JupyterLab on my Hub.
How to make users to use host local machine... ? Instead of using Google cloud..
Hi, is this still working? I see no maintenance in the repository. When are the user and password used in 10:40 to create an account?
Can u please upload the whitelist instructions? Its really urguent....
can you help me. I have error in "Welcome to io Online! "
NameError Traceback (most recent call last)
in
----> 1 display(HTML(""))
2 Git.b.signIn
NameError: name 'HTML' is not defined
Hi, is it possible to use your own local machine as the server instead of resorting to cloud services? Can I assign an IP address to it that is reachable via HTTP/HTTPS?
Of course it's possible. There is no reasons how to use it: via some cloud service or via own server.
Should the number of nodes be equal to the number of max users at a given time? i.e. if 7 users would always be using the jupyterhub, should the nodes be equal to 7?
Great question! You do not need to have number nodes equal to max users. Nodes can be shared by users. In Jupyterhub's settings (link below) you can actually decide the much computing power each user gets. I think the default it 1/2 a cpu.
zero-to-jupyterhub.readthedocs.io/en/latest/user-resources.html
Hi do u know how to create the user accounts?
Thanks. Great stuff!
Thanx for the share.
Hi, wanted to know if there's any app for version control in Jupyterhub?
awessome bro i really loved it thanks for ur open contribution
Hi do u know how to create the user accounts?
great video!! I noticed there is a "private" option under "Manage Files" section...can you please explain how can I use it?
Great question! The "private" section is when u want to back up ur files on github but don't want others to see them. I'm currently updating io online application. When I'm done I'll do a tutorial on how this works. Hang tight!
Can you provide the notebook code of the iPad comparison.
how can we limit 1GPU per user ? so that a single user will not consume all the available GPU.
can we run jupyter using a REST api?
How can I integrate nbgrader in this???
What about just making a docker image? one command and it can literally run everything.. well almost one command.. still gotta install docker but thats not hard..
Great Question! So you can run jupyterhub inside a docker image (see link), however this has a major limitation. A docker image is a prepackaged OS that runs on ONE machine. U can set this docker image up on Compute Engine, but you have to decide how many cpus you want to use up front and the cpus don't have the "scaling" feature that Kubernetes has (see 5:09). This ends up being super inefficient, and if u have a group of 20 or more users u waste tons of money, it's like 10x more expensive.
Right now the closest thing there is to a "Kubernetes Docker" is Google's Tiller software, which is still rather complex to use. I tried automating it as best I could, but yes there's still definitely room for improvement!
JupyterHub Docker: hub.docker.com/r/jupyterhub/jupyterhub
Web Application would be a more appropriate term.
great. what about installing jupyterhub on a local server accessible only to the workgroup LAN ? it has no Kubernetes, just one VM with decent capacity to run jobs for a bunch of colleagues
:) For that I think it's actually WAY easier. U can use a docker image of jupyterhub which u can then spin up on ur local computer (JupyterHub image attached below).
hub.docker.com/r/jupyterhub/jupyterhub
I got a question, how to read csv files on my iPad
hi, please explain how can i change theme of jupyter on ipad ?
thanks a lot.
So if you want to change the look of Jupyter, I can think of 3 ways to do this:
1. check out Jupyter Themes (see link). This'll change the notebooks color/look
2. Launch jupyter lab, go to: "your_jupyter_website"/lab
3. Manually change the look of jupyter by messing with it's html style code. Check out the io_View app for examples of this.
Hope that helps!
@Reza, Ur comment just made me think of another video idea! Teaching how to code HTML, by changing io's look! Thanks for the comment! :)
@The iPad Pro wow, thanks a lot 😍
How can i start jupyterhub on minikube
Any luck with enabling HTTPS ?
Please update your video. Since Google cloud platform has changed
And what about compiling
Depends on what ur referring to mean. Regarding Programming Languages, both the Java and C++ packages do a kinda magical thing where the code is interactive (like python). I don''t know how it's done but it's really cool! (see link to learn more). As for the time it takes to launch JupyterHub, lol, yah I did cheat a little and speed past the installation process. :)
github.com/SpencerPark/IJava
Bro new technology cloud computing...really?
😂 Yes, ur right u I can't call cloud computing new tech. That said, new WAYS of using cloud computing are being released rapidly by Google and Amazon. For instance the Kubernetes tech that this video relies on was made with 2014 (lol... still not really new). And some of the latest examples are Google's MemoryStore and Cloud Functions.
@@elliott3165 True, I guess I didn't look at the date. My bad!