workshop setup for friday including data, ea-environment and a working space

[lwasser]

For our workshop, we will need a similar setup to what a student would need.

1. the students will need a semi persistent (for the time of the workshop) working area where they can export files
2. the students will need access to 1 or more datasets (on figshare) that will be setup for them in a drive on their computer
3. the students will need access to the full earth analytics python environment that we have build

Questions

1. how do students log in to the environment? i'm fine with the way we discussed!!
2. How do i setup the data so it's uniform for them? ie i'll need to know what the path is to the data for them to access it.

@betatim do we have this in place to use for friday? or what do you need to get things going so we are ready? thank you!!

[betatim]

For (1) is it enough if users have a home directory as part of their jupyterhub login?

For (2) is this a dataset that is listed in https://github.com/earthlab/earthpy/blob/master/earthpy/io.py? If it isn't part of earthpy we could add it to the docker image directly.

I started using (3) on https://hub.earthdatascience.org/earthhub yesterday, needs some testing to be sure but should be ready.

[lwasser]

@betatim thank you.

1. I'd like to see how this works but if the student have a home directory that has space to write data to. (write access and space allocation) that should be sufficient!
2. The data are for friday listed in : 'spatial-vector-lidar': ('https://ndownloader.figshare.com/files/12396203', '.', 'zip') <- this is the line in the dictionary. What i don't understand is i want that data to be in their home directory for them to use. i don't understand how we get it there! it may just be me not understanding how this is setup :) will i need to show them how to download the data once they login? AND if they do download it - what happens if they logout and then log back into OR get disconnected? does their workspace persist?

THANK YOU!!

[betatim]

The workspace persists.

I'll create a workshophub just for the workshop tomorrow morning and add this file to the image so the data is there ready to use for the students.

What system should we use for authentication? Do they all have a UC Boulder email address? If yes we can use that for logins.

[lwasser]

ok. ideally for future setup it would be easy for me to add a few datasets to a "hub" space! maybe we can chat more tomorrow?

We will have a mix of students potentially. There may be some without CU authentication as we had two drop in yesterday that are not in the system yet. Could a CSV approach work potentially?

[betatim]

If there is a mixture I'd go for https://github.com/yuvipanda/jupyterhub-firstuseauthenticator with a white list.

Should we also have the material of the course directly available in people's home directory?

[betatim]

• put data into the image [WIP] Add data download command #45
• copy data from the image to the user's home directory on pod start [WIP] Add data download command #45
• authentication HashAuth Switch to HashAuthenticator #47
• switch node type to a bigger node or even use two different node pools
• nbgitpuller with https://github.com/earthlab-education/2018-07-20-spatial-python-workshop [WIP] Add data download command #45
• ~15 students be prepared for up to 25
• add a notebook to admin user that generates passwords
• more grafana monitoring for nice plots

Workshop starts at 9am Boulder time.

[betatim]

Using https://github.com/thedataincubator/jupyterhub-hashauthenticator at the moment which does not seem to support whitelists. Maybe worth fixing at some point.

[betatim]

Added a second nodepool with two machines that each can handle around 14 students. The node pool is allowed to scale up to 4 machines so we should have enough resources even for last minute arrivals.

• turn off woskhop-pool again after the workshop to save money

[betatim]

Collecting to do items and learnings from the workshop:

• kernel died a few times, probably because using too much RAM, current limit was set by guessing a not-crazy number, what is a better way to determine this number?
• data copying was broken (see Make sure target directory exists before rsync'ing #58), next time run through process with completely fresh users to catch problems that only show up on first-use
• right at the start of the workshop (19min in) there was a k8s master update which caused 3minutes of outage. Why did this upgrade happen then?

[lwasser]

NOTE: the kernel is consistently dying on the show_hist function using rasterio! so this is definitely a package issue but the question is what causes it to die? maybe memory hog?

[betatim]

From mid workshop. This shows actual amount of memory and CPU used. Not how much memory/CPU we promised to each pod. This would suggest we can increase the memory limit beyond 2G a bit without having to pay more money for bigger machines. It seems most people, most of the time aren't using all the memory we assign them. Brief peaks for show_hist?

We can definitely increase the CPU limit as people mostly idle.

Before giving away more RAM we need to make sure that all the core services specify how much RAM they need so that they get protected by kubernetes.

[betatim]

@lwasser was their any promise to participants how long their home directories will be available? Otherwise I'd clean (aka delete) them as part of shutting down the workshop cluster.

[lwasser]

No! i told them to download their files as it would all disappear. we can clean it. i'd like to know how to do that. I also have questions post wowrkshop

1. data -- do the data update now ? so if i add a new folder to a dataset, it will update even if the folder already exists?
2. if the folder doesn't exist will it create it ? (ie @joemcglinchy didn't have data i think because his data didn't update

[betatim]

Hub has been turned off via #67 and following documentation in https://earthlab-hub-ops.readthedocs.io/en/latest/day-to-day.html#removing-a-hub

[betatim]

The answer to @lwasser questions is: yes and yes.

If the data in the docker image change that will be reflected in the user's home directory (once they stop&start their server to pickup the new image).

Previously there was a bug that it would not update if some of the directory existed. This should be fixed now.

[betatim]

Closing this as the event is over and most of the action items from lessons learnt have been done.