National Energy Research Scientific Computing Center (NERSC) NUG Monthly Webinars, 15 Jun 2023

Previous Meeting

Next Meeting

⏯

youtube image

►

From YouTube: NERSC Users Group (NUG) Monthly Webinar: June 15, 2023

Description

The Monthly NUG Meeting is a regular opportunity for our users to show off what they've done, for NERSC to get feedback from users, and for users to exchange ideas.

Topic-of-the-day: Jupyter at NERSC: This presentation will introduce the Jupyter service at NERSC and guide users through the process of starting a notebook as well as more advanced features such as creating and customizing Jupyter kernels.

A

Okay, great: let's go ahead and get started, hi everyone. My name is lipi.

A

um Today, we're going to have um a presentation about jupiter um at nurse, um but before that, let's.

A

Get through a couple different items, um so, as per usual, the plan for today's meeting is to have a.

A

Really interactive session: um we don't have too many people here today either. So please feel free.

A

To raise your hand or just speak up, if you have a question or if you want to share an announcement.

A

um We, you know, want to encourage people to interact with us and interact with each other.

A

During these meetings, um so the plan today is to first go through some announcements um and some.

A

Calls for participation, so this is information that's um available in the weekly email, but we'll.

A

Just go ahead and go through it in case you missed any of it, um then we're going to try something a.

A

Little new, we have some trivia questions. um I personally love trivia and I think other.

A

People tend to enjoy it as well, so you can test your knowledge um in you know today we're doing.

A

A couple sort of jupiter related things or python related, but we might do all you know all kinds of.

A

Different things in the future, then in case anyone has anything they want to share about what they've.

A

Learned recently about the system and then then we will go ahead and go into our talk with kelly.

A

um About jupiter, so, um uh first of all, please, um if you want any more information.

A

About what's coming up, make sure to check the weekly email, but I'm here I'm highlighting.

A

So on july 12th there is going to be an ecp seminar on writing clean scientific software.

A

um Then there will be another spin up training, so this is um about the spin platform at nurse.

A

Which is offered pretty frequently the next one is coming up june 21st um there's going to.

A

Be a really great um session, which is called crash course in super computing. This is for.

A

um Mostly for novice parallel programmers, um but it might be of interest to people who have.

A

Some experience as well, um and so that's going to be on june 22nd um again any information.

A

That is listed here you can find in the um weekly email for like to you can find all the links there.

A

There's going to be a fun training, uh so this is the fortran users of nurse group.

A

They are hosting a modern fortran basics, training which is going to be on the 10th and the 11th.

A

um And then there's going to be a training using uh chapel, uh upc, plus plus and corey.

A

Fortran these are, these are happening on the 26th and 27th, and so remember you can.

A

Go to the weekly email for these links or you can check out the training events page on nurse.gov.

A

um In the next couple months, there's a lot of different uh uh um conferences and workshops.

A

Coming up um so there's uh this one, which is institute infrastructures for enabling extreme.

A

Scale analysis and visualization that deadline is august. 4Th um there's also the us rse association.

A

Conference, the deadline for talks and posters is coming up um on june 19th and then there's.

A

Also workshop deadline, if you want to participate in that which is on june 30th, um there is um the.

A

Sixth, annual parallel applications workshop: this is happening at sc, which is super computing, so.

A

The deadline for that workshop is july, 24th and there's another sc workshop here. A containers.

A

Workshop, which has a deadline august 18th so again, there's a lot of different deadlines.

A

Please make sure to look in the email, there's also registration for confab23. This is the esnet.

A

Community meeting that happens yearly and the registration for that is open as well. I believe.

A

The meeting itself will be in october, so if you're interested, you can register to attend that.

A

um I am able to see chat so in case anyone has any questions. Please feel free to put.

A

Them in chat, I can't see my screen's not big enough to also have the participants open so.

A

If you raise your hand, I might not be able to see um so if someone else wants to call it.

A

Out or just go ahead and unmute and ask if you have any questions, please feel free to do that.

A

um Great okay, so let's get to the part that hopefully everyone is super excited for again.

A

This is something you were trying, so um we want to start doing some little trivia.

A

Sections partially to challenge everybody- uh maybe you'll, spend more time reading through.

A

These docs so you're ready for trivia next time- um or maybe you already know all this so the.

A

The goal is to have little questions about hpc programming, all kinds of different things and if.

A

You want to submit trivia questions you're more than welcome to you can send those questions to.

A

Me or charles, or to our sort of general email at nurse community managers, and maybe one of your.

A

Trivia questions will be featured um so today to trivia. You will not be able to input your answer.

A

Into like a quiz system, you're just gonna have to hold the answer in your head, and then you.

A

Can you can share it in chat at some point, not right away, so don't give the answer away if you.

A

Know it right away just hold it in your head and then you can spam the chat with what you think.

A

Is the right answer and then we'll I'll reveal it after the fact, so we have two questions today. The.

A

First, one is a python programming question. So if you are a python whiz, you may know the answer to.

A

This question, or you may learn something new. It may not be as trivial as it seems.

A

The first question is: what is the output of the following code? So there's a little.

A

Code snippet here in gray and there's a couple different answers. If you know.

A

The answer- please do not put it in the chat immediately just hold on to it for a moment.

A

um And try not to go open up your python interpreter and type it.

A

In unless you really don't know- and you want to learn- and you want to try it out.

A

Think about it for a minute.

A

Okay, on the I'm going to count down from three and then put your answer in the chat.

A

All right, if you know the answer, all right, three two one go.

A

Cool okay, I've seen c a c c, a b c and someone wrote z thanks.

A

That's yeah, that's a really good answer as well.

A

um So uh I guess I don't I maybe I shouldn't reveal the answer immediately. Maybe we should.

A

Discuss it does anyone who uh thinks it's a want to defend their answer.

A

We have a couple people who said a they don't think this is a reasonable thing to ask python for.

A

You can type it in chat if you want, or do you want to shares anyone want to share.

A

B

It's a bad answer, but I think that the point of exercise is that this code is supposed to.

B

Do something completely unintuitive and b was the most unintuitive. Okay.

A

And then we have some c c and b are kind of similar who thinks c.

A

Oh koichi's, rethinking, oh okay, that's my goal is just to stump koichi because koichi knows.

A

Everything so yeah feel the answer see uh yeah. So, okay, so andrea said, I'm guessing.

A

There isn't an if state a statement, so else would throw a syntax error.

A

um So you can go ahead and put this in your um in your python interpreter. I did double check this.

A

um And it does work, so python has a cool thing again. This is something that I I didn't actually.

A

Know about python, you can do a four else statement, so you can have a four loop that does whatever.

A

You want it to do and then, as soon as I is not or you know whatever your for iteration.

A

Or you know, uh statement is as soon as that is over. The else will will um uh take take effect.

A

um And range in python is, um as I think sam has mentioned here, it doesn't include the endpoint so.

A

Because this step is one, it doesn't actually reach three. So it goes one two and then it does the else.

A

um So yeah this I I agree. I think this is a little unintuitive. I didn't realize.

A

You could do a four else in python, but you totally can so maybe this will uh uh change.

A

Your your pythoning from the future! um How is it different from adding code right after the loop.

A

um Yeah I I guess I don't have a good example of why this use case like what exactly you would want.

A

Like maybe you have some kind of unique thing that you're trying to achieve and and under a.

A

Certain circumstance you wanted to take place, um but maybe the answer is that you have this option.

A

If you want it, but maybe you don't need it, I don't know so cool great. Okay, I'm glad to see some.

A

People are learning some things, I'm glad to see. I stumped a couple people. Okay. The next question is.

A

um A jupiter question and kelly feel free to chime in and uh quiz people about this a little bit.

A

um So this is to wet your appetite for today's uh um presentation so which of these jupiter server.

A

Types charges nurse hours, so a is shared cpu node b is exclusive cpu gpu, node and c.

A

Is configurable job, and so the question is which one of these? If you were to launch it and use it.

A

Which one would be charging your nurse hours.

A

So keep keep your question in your head or keep your answer in your head for a moment before.

A

We ask people what they think.

A

Okay, on the count, I'm gonna count down from three then put your answer in the chat.

A

All right three, two one go.

A

Oh, my gosh see almost unanimous.

A

Okay, so people think that it is both exclusive and configurable.

A

And since everyone seems to think it's, let's just check if that's right.

A

It is yay, well done so yeah, so um I maybe kelly will tell us more about this, and koichi is also.

A

Asking koichi doesn't know what a configurable job is, so um maybe kelly will address that.

A

During her presentation, um cool, okay, anything you want to add kelly or uh congrats. Everyone.

C

Everyone's very well prepared, um I don't have a screenshot of the configurable job, but I can.

C

Definitely talk about it when I get to the slide that um you know when I'll be talking about the.

C

Different server types and how it's different from the proceeding too, and how also we kind.

C

Of think on the jupiter team that this page is a little misleading and we're going to be.

C

um Changing these uh server names ever so slightly for a bit of clarity um in terms of the current.

C

State of things, and also some new features, we're bringing in oh great awesome, cool thanks kelly.

A

Wonderful, okay, so before kelly begins her talk, um I did want to give people a chance. We like to hear.

A

What people have been learning about, what they've been doing um and oh sorry um and I I've already.

A

Seen a couple, people learn something today from our python question, so I'm really glad to see that.

A

um But in general, if anybody has anything that they've learned not just today but maybe since.

A

The last nug meeting, if anyone would like to share um this, is your opportunity to share.

A

With with the group um again, most things take practice. I love this. uh This uh cartoon you can.

A

Change it to say anything here, you know how you learn to do x. Well. Well, it just takes practice.

A

um So I think that's the same for using hpc. um Does anyone have anything they'd like to share.

A

Okay, well feel free to do so or just ask more questions in the chat today.

A

um So, let's go ahead and move on to our presentation from kelly, so kelly, I'm gonna, stop sharing and go ahead and take it away, and please also.

A

um Feel free to introduce yourself and tell us a little bit about your role at dirsk.

C

Okay, um cool thanks, uh so my name is kelly. Rowland, I'm in the user engagement group here at nurse.

C

um I've been at nurse for almost five years. I started in the data science engagement group.

C

And have recently moved over to our user engagement group and one of my favorite.

C

Things that I do here is to work on our jupiter deployment, um because I think.

C

Jupiter is really cool. um It's a very human way of accessing high performance, computing resources.

C

um And I think it's just kind of like the way, the future right like classes, huge classes at.

C

Large and small institutions everywhere are training students and future generations in.

C

Jupiter contexts, and so um you know, I think it's important- that we meet users where they.

C

Are we offer features that they want to use and, and jupiter is one of those, so my goal for this.

C

Talk um you'll see I've. I've put the title of the talk is using jupiter at nurse a primer.

C

um If nothing, by the end of this talk, I would like you to know how to get to jupiter at nurse and how.

C

To get help with jupiter at nurse, so I'm not going to go into any extensive details on the back end.

C

um If people are really interested in that, we can give it a different talk about that, but.

C

This is just an overview of various different features um and just getting getting folks a.

C

Little bit acquainted with jupiter and what you can expect when you get there. So what is jupiter.

C

um A nurse we talk about jupiter and it's really in reference to a collection of many things.

C

um Project jupiter itself is a big umbrella of a lot of different efforts.

C

um That includes some things that we do work with, like jupiter, notebooks and uh jupiter hub and.

C

Jupiter lab, but also other things under project jupiter, that we don't necessarily work with so.

C

We talk about jupiter at nurse. We talk about users accessing shareable documents called notebooks via.

C

This jupiter hub deployment that we say um so what do you do with the jupiter notebook once.

C

When you've got it and I'll talk about how you get to a jupiter notebook, but what what is the.

C

Point of these things you can put code to run in your jupyter notebook. You can um put equations.

C

And visualizations and narrative text, so you can use things like markdown for rendering.

C

Equations and writing texts to have a narrative people write books in jupiter notebooks, actually.

C

um And you can put in interactive widgets, so you can see in this bottom right um here is an example.

C

um Kind of one of project- jupiter's favorite, canonical examples- is the lorenz equations.

C

Because they make these really pretty um graphs, so that's um it's a very richly featured interface.

C

And you can use jupyter notebooks for a lot of different applications.

C

um A lot of people use it for data cleaning and data transformation. Looking at simulations.

C

Modeling, um you can load plots and widgets in line right there in the notebook. So it's very popular.

C

For data visualization people use it for running machine learning as well as processing output and.

C

All sorts of workflows and analytics frameworks, um so I've mentioned that it's a very popular service.

C

um We've seen a pretty steady increase historically in running folks running.

C

Jupiter nurse um so right now we're up to about 12 or 1300 unique users a month um which is.

C

You know comparing that to our 3000 user per month, who can act vsh, it's a not insignificant.

C

Proportion of the users of our of our user base, who are actively working in the system so.

C

um It behooves us to provide um this jupiter interface for people to use um okay, great so.

C

How do I use jupiter nurse, uh the shortest answer to that is to go to your browser and.

C

Point it to jupiter.nurse.gov um and what lives at this url so to speak is a jupiter deployment.

C

um That the jupiter council at nurse manages um where uh across group team with a nurse.

C

Who work to support jupiter and provide new features and developments? So what does the.

C

Jupiter hub deployment itself actually do it handles a number of things. um You'll log in.

C

Through federated identity, which is something that you're probably familiar with in terms of logging,.

C

Into iris, or help desk, um the jupiter hub then um spawns a notebook server for you somewhere in the.

C

Nurse system, depending on which server type you choose and I'll talk about that shortly,.

C

um The hub manages communication between what you put into your notebook and the notebook itself.

C

um It manages any processes that could spawn out of the notebook and it keeps.

C

Track of those processes- and there are also other additional services like we.

C

Have cool little menu links to the jupiter related documentation? If you have a question.

C

So the the highest level flow looks like here on the left. You reach the authentication page when.

C

You reach jupiter, there might be a note about such a system is down right now and you can't reach it.

C

Through jupiter, um once you've authenticated you land on the control panel, page, which.

C

um Allows you to choose a server type and then once you're in on your server, you have a notebook.

C

um You can you can work with your notebooks and you can have many different notebooks.

C

Open I've always got a bundle running for my own analyzes that I do for further efforts at nurse.

C

So how do you choose a notebook server to spawn? um We talked about this a little bit.

C

And right now, when you log in you'll, see these four different types of servers on promoter.

C

um So furthest to the left, we have a very common use case. What we right now call the shared cpu.

C

Node server, this launches your notebook on one of the login nodes on chromutter.

C

um It drops you into the same python environment that you would expect on.

C

Login and you can submit jobs with this, um it's called a bash magic command. This.

C

Exclamation mark at the beginning of the command: that's in that's a ipython notebook feature.

C

And since you're on a login, node you're sharing that login node with other users, other processes.

C

Maybe other notebooks there are 40 of them at promoters, so you've got a little bit of leeway.

C

But we generally ask that folks be good citizens um on login nodes, either in jupiter or otherwise.

C

But if you've got you know a big data set to process or you want to use the gpus or um you're.

C

Interested in just a little less noise, you can use one of these other three servers. Server types.

C

um In jupiter come on okay, it's good sorry about that; okay! um So in the middle we have.

C

These exclusive cpu or gpu server types where your notebook gets sent to a job allocation.

C

um Via slurm, and with this, you get one cpu node or one gpu node for your notebook.

C

um And this does, as you all uh are, astutely aware, use nurse hours.

C

um And then further to the right most um on your screen, we have the configurable job and what this.

C

Does is let you so if you click on the um the start, button on configurable, configurable job.

C

um A job does not get submitted automatically, um whereas the exclusive cpu, cpu and gpu know jobs.

C

Just submit a job, the configurable job takes you to a selection menu. Where you can say. Maybe I need.

C

To use four gpu nodes and you can input four for the number of nodes. If you're, if you're running a.

C

Larger gpu based application through the jupiter notebook or, if you'd like to change the time.

C

um On the job, so that's um the configurable job offers a little bit more flexibility in terms.

C

Of your job parameters and you can choose either a gpu node or a cpu, node or multiple.

C

um Gpu nodes or cpu nodes, whereas the exclusive um servers here just.

C

Have this default value of one node for so long and the configurable job itself.

C

um Allows you to put in it's sort of like you. Have you end up with some drop down menus, which.

C

Are akin to writing the s batch headers and then that all gets sent off to slurm to be scheduled.

C

um As a batch job, which you have interactive access to through the jupiter notebook.

C

So we've picked our notebook server, uh then what then we get into the jupiter.

C

Lab interface, so you might see um if you've logged in you might see something like this.

C

This is sort of a most trivial example of a notebook where you can, in the top.

C

What's called cell here, you would do your import statements of other python modules that you need.

C

um You run a couple python code uh commands and then you can render a plot here um right in line.

C

In the notebook and there's a couple fun features you can see down at the bottom, I'm using so much.

C

Memory, so it's good to keep track of that if you're on something like a login node um to.

C

Understand how much memory you're using um you can see in a couple places it says nurse python.

C

um That's what's called the kernel which I'll talk about in a little more detail.

C

um Over on the left is our file browser, so you can use this to get to different places in the file.

C

System open your various notebooks data sets. You can also open files which are not notebook files.

C

In um jupyter lab you can open text files. You can also open use jupyter lab to open a terminal.

C

In the browser, which is quite convenient, um a couple things that we're particularly proud of is.

C

um That jupiter lab features what are called extensions, so people can write code extensions to.

C

The interface and um they can either be standalone, live in a github repost or something like this.

C

Or we can do what's called upstreaming in where the project jupiter says. We think this is a great.

C

Future and we'd like to put it into the sort of base um uh deployment of jupiter lab so that.

C

Everyone who has jupiter lab has this too, and so we've developed a couple extensions at nurse.

C

um With some previous interns that we've hosted and the uh and some of these extensions have been.

C

Upstreamed into jupiter, and so uh one thing- that's a nice convenience- is that um in this.

C

Top left, you have uh your favorites, um so if you um very frequently access somewhere in cfs.

C

um It can be very tedious to type that whole path all the time, but you can just favorite it.

C

um You know kind of not unlike a browser, so this is pre-populated with your home in your scratch.

C

Because we know that people like to access those, but you can add, however many favorites to quickly.

C

Switch between different places in the file system, you can do that with this little star button.

C

um Other cool things, if you click on the file uh menu item, you can open from path. You can click.

C

On this open from path item to jump to anywhere in the file system, so if you do know, you need to.

C

Get to somewhere else, if you're working in scratch and you need to hop over to cfs and you don't want.

C

To click through this whole file browser thing to go all the way back to the root and then get back.

C

To cfs, but you know the path you're looking for um you can open that from path or if you've been.

C

There recently, you can um add that you can just use this recent feature here. Excuse me.

C

So um I mentioned uh kernels, um so the kernel is what actually runs your code, so this is kind of.

C

The extent that I'll talk about what you might think of as like the the back end of jupiter.

C

um I won't go into a lot of detail, but we'll I'm talking about them, because uh this is a.

C

Very useful and popular feature, so you have your user who's using the browser to access jupiter.

C

In their notebooks and the notebooks are each talking to a kernel, um this might be nurse python.

C

um Which is our default? Python module uh nurse also provides other kernels. So if you're.

C

Interested in julia, you can load, one of you can pick a julia kernel and then you can write julia.

C

Code in your notebook and just run that we also provide our kernels if you're an r programmer.

C

And so you can um write your r code in in a notebook. um Our teams uh have several kernels.

C

Which include machine learning packages so pi torch or tensorflow, so you can load on.

C

Those kernels and use those applications right there in the notebook in the jupiter interface.

C

And this is um the reason I introduce this concept of the kernel of how jupiter actually runs your.

C

Code is because the most common question we get from users about jupiter is that they have their.

C

Conda environment in the terminal, which is very useful, very productive. They like working with it.

C

And they want to know how can I use my conda environment in jupiter and the answer to that.

C

There you can do this several ways, but you can create a kernel from that conda environment. So.

C

We have um we'll include these slides and the slide has a link of the documentation which.

C

Includes how to do this, you don't have to you know, take a screenshot of this or anything like that.

C

um But if you activate your environment, your condo environment and install ipy kernel and.

C

Then use ipy kernel, um this ipy kernel command. You can create a kernel spec file which jupiter.

C

Will look for and see? Oh there's a kernel in this path that I'm looking for and then you can pick it.

C

um You can choose this kernel in um in the jupiter interface, um so just to talk about that. A little.

C

Bit more, um you can see if the kernel spec is installed into this uh verbose directory. Please.

C

Command the kernel effect is our way to pass the path of the environment into the command.

C

Good question: there's a question in the chat about the kernel spec, um the the question says when I.

C

Use this command the kernel spec points to the wrong path. Is there a way to pass the path of the.

C

Environment into the command, um and there is a way to set environment variables in your kernel, spec.

C

So if you have um this in in your kernel, spec file in additional customization and additional.

C

Customization that you can do in your kernel, spec file is to set um environment variables, but it.

C

Is the case that, if you're running, um if you run that command from your activated conda environment,.

C

That you should that the path should all line themselves up, not automagically, but correctly.

C

When the command, when the ipi kernel command, is run from the activated content, environment.

C

And so you'll see that this kernel file that gets created um points to the condo python.

C

um It uses this ipi kernel launcher command, and it sets up your um display name that you would expect.

C

To see in jupiter and what language you're working in so python um tends to be a majority usage here.

C

um But you can also do things like set environment variables in your kernel specs. So you can add.

C

um You can set your path, um ld library path, any other environment variables you like to use.

C

um You can actually set these on the command line with the ipi kernel, install command.

C

um Or you can go back into this file and add them um manulate yourself.

C

If you want to get uh really fancy, you can create, what's called a kernel, helper script, where instead.

C

Of so these past two slides. The first argument here has been um the python binary in your cond.

C

Environment, but instead of pointing to that, you might point to a kernel, helper script.

C

In which you can do many things, basically anything that's valid in a shell script,.

C

You can export environment variables, you can load modules, you can do. You can do a whole ton of setup.

C

And um but the most important thing is that your kernel helper script at the end, should.

C

Have this python exact statement um so that um the notebook can connect through appropriately.

C

But this is like really the most flexible approach that you can use in terms of setting up a kernel.

C

In that, um you can load modules with it and export environment variables without using.

C

um Without having to write them out in the arguments here, um a lot of folks who use.

C

Jupiter are interested in using shifter in their environment. You can also do that, um so you can.

C

Create a kernel spec file, which has shifter as an argument which includes the image name.

C

um And the path to the where the python lives in the image. um But then the rest of the.

C

File looks pretty similar to the other one. So if you can create this kernel and then your.

C

Notebook will run in your shifter image that you're interested in so these are sort of.

C

The most flexible custom customization avenues for users to like load up a reach, a richly featured.

C

Environment um which and then use a notebook in that environment to work through their workflows.

C

uh Jupiter issues um it happens, it's okay happens to me. It happens to everyone.

C

We direct a lot of output to a log file in everyone's home directory who uses jupiter.

C

um So if you do something like cat home dot, jupiter.

C

Dash pearlmutter.log you'll, you may see a lot of output that looks like this.

C

Rr and julia kernels, customizable too I so there's the question in the chat is: are r and julia.

C

Kernels customizable too, you could create a custom kernel with r or julia as a language. I believe the.

C

Preset um nurse kernels aren't really editable um they're sort of defaults for people to pick.

C

But [music] um I think if you can create a conda environment for a given language like.

C

R julia, then you could make you should be able to make a kernel spec file out of that.

C

um But that that's a very good question.

C

So [music] if you're running into uh problems with jupiter, um this happens, sometimes you.

C

Get like a 500 error, you might look into your log file and see if there's anything.

C

um Meaningful to you, uh if there's not don't worry, we got your back, but sometimes this is useful.

C

If you've got um some python packages, loaded that are causing conflicts with the way that.

C

um Jupiter lab runs and usually, if we're helping a user debug their.

C

Jupiter issues- this is one of the first places that we look is in their log file.

C

So, um as a summary, uh please visit jupiter.nurse.gov in your browser.

C

Of choice to use jupiter nurse um once you've gotten well seated uh in in that interface, you.

C

Can use kernel specs to use conda environments or shifter images and to run your notebook.

C

um The jupiter cement nurse um works really hard to make jupiter work and work better for.

C

You for all the users, so um a couple of folks have asked about running jobs for jupiter.

C

Notebooks, which only use a single gpu so right now, if you're, using a gpu, an exclusive gpu job.

C

It grants you an entire gpu node on pro mudder, which is four gpus, but some applications.

C

um Can really only leverage one right now, so we're expecting to release that feature.

C

um This month, or next in combination with renaming the servers to make a little more sense in.

C

This context, as well as upgrading jupiter lab to version four right now we're running version. Three.

C

um And jupiter lab four um brings hub performance improvements.

C

um That we're looking forward to it loads, larger notebooks in a way that's more intelligent. It only.

C

Loads, what you're, looking at as opposed to the whole notebook all at once, um we don't have a.

C

Very specific time frame for that, because we're waiting on some of the developers of the jupiter.

C

Extensions that we use to upgrade their extensions for jupiter lab for compatibility. But we hope to.

C

um Get that upgrade in place soon, um so we are always looking for new ways to empower.

C

Jupiter users um to use the jupiter hub and the hpc system through the jupiter interface, so.

C

um Feedback advice, questions even pointers. We always welcome at our help desk at help.nurse.gov.

C

um So thanks everyone for your attention here, um there's some there's some stuff going on.

C

And there's some active debugging going on in the chat. um One tip that in if you're, using.

C

A kernel file in the json that will holds recommends adding the metadata this entry.

C

Metadata debugger true, which enables the use of the integrated python debugger. That's cool,.

C

um Jake asks what is the best way to install specific python packages that you may need.

C

For your code um generally, we recommend that users install packages through conda.

C

um So you would activate, you, would create and activate a conda environment and then.

C

Use conda, or sometimes even pip, to install packages locally for your own use. um You can.

C

Also put these conda environments in somewhere, like cfs. If you want to create an environment.

C

um That your whole team can use, but uh the short answer to that is conda.

C

Sam sam is asking in the chat: how to open a ticket um you can go to help.nurse.gov and that.

C

Should prompt you for an authentication and once you've authenticated, then you.

C

Can open a new ticket? There should be a button that says like open a new ticket.

C

um And then you can fill in um what your uh specific error messages that you're getting are.

A

That's a very nice segue also for uh reminding everyone what our next.

A

uh Presentation is going to be uh before that. Thank you, kelly for your fantastic presentation.

A

um These slides will be available on the nurse website afterwards, as well as the recording.

A

So if you have more questions or you want to reference, these slides you'll have access to them.

A

um And yes, our next presentation in july is going to be about how to submit a good ticket.

A

Basically, how to submit a ticket so that you can get a solution quickly, uh rather than sometimes.

A

What happens is you know? We get a ticket and we just have to ask for a lot of information that.

A

You probably could have given us the first time around, but you know maybe you weren't aware that.

A

That was something that would be helpful, so we want to make sure all of our users know what kinds.

A

Of things we look at, if we're diagnosing, you know different types of problems and how you can help.

A

Us help you get a solution really fast, um so yeah, that's great! So thanks so much kelly. um Why don't.

A

We let people sort of stick around for another couple minutes, I'm going to pause the recording.

A

um And then, let's see, if I can do that, stop recording.