Where communities thrive


  • Join over 1.5M+ people
  • Join over 100K+ communities
  • Free without limits
  • Create your own community
People
Activity
    Josh Lieberman
    @lieberjosh
    Question: how would one use nebari to "scale" an interactive dashboard for many users?
    Rich Signell
    @rsignell-usgs
    Did anyone get kbatch working? I got this output -- does it tell me what is wrong? https://esip-ogc.nebari.dev/hub/user-redirect/lab/tree/shared/disaster-pilot-2022/rsignell/kbatch.log
    eskild
    @iameskild
    Taking a look now
    Rich Signell
    @rsignell-usgs
    Let's see the intake example
    eskild
    @iameskild
    When you submit your job, this json will be outputted. And this looks correct to me
    You can validate that your job is running by listing your jobs: kbatch job list -o table
    Josh Lieberman
    @lieberjosh
    Thanks Dharhas. Good to identify that use case and possible solutions. Lots of useful indicator publication is limited by concerns about dashboard elasticity.
    Rich Signell
    @rsignell-usgs
    @iameskild , yes, you were right! The "kbatch" did work! It was just taking a while!
    $ aws s3 ls esip-nebari-dev/testing/rsignell/
    2022-07-07 12:21:45        850 ogc-cog-validation.txt
    eskild
    @iameskild
    @rsignell-usgs maybe adding time.sleep was overkill haha
    Rich Signell
    @rsignell-usgs
    Here's an LANDSAT example using STAC (from Element84)
    Rich Signell
    @rsignell-usgs
    I'm testing out building a new environment that includes geogif
    Dharhas Pothina
    @dharhas
    Great. One bug that didn't get fixed in time for this workshop is that only environments created in the git repo are accessible for dashboard sharing. For all other cases, the conda-store UI is fine for creation of environments
    Rich Signell
    @rsignell-usgs
    I'm building it in the "disaster-pilot-2022" namespace with name "ogc". Is it a problem having the same environment name? (e.g. disaster-pilot-2022/ogc and filesystem/ogc)
    eskild
    @iameskild
    I might give it another name to be safe. I'm not sure if that bug has been fully addressed.
    Dharhas Pothina
    @dharhas
    I think it's an issue using the envs from the terminal but not from jupyter but to be safe lets use something else.
    Rich Signell
    @rsignell-usgs
    can I kill the build?
    oh, it's done already!
    Dharhas Pothina
    @dharhas
    well I guess lets see if it works :)
    Rich Signell
    @rsignell-usgs
    too late. I deleted it
    Dharhas Pothina
    @dharhas
    as an fyi it showed in my kernel list without a restart
    Rich Signell
    @rsignell-usgs
    nice
    well, maybe not too late. I clicked the "delete" button on the build (build 8) but it still seem to be there?
    But I'm building again "odc-rps"
    Dharhas Pothina
    @dharhas
    I think 'delete' queues it for delete but it might take a bit of time. Chris would have to chime in on that.
    Rich Signell
    @rsignell-usgs
    I'll be 15 min late
    eskild
    @iameskild
    Hi folks, we'll be starting the "How to deploy Nebari" session in a few minutes :)
    Dharhas Pothina
    @dharhas
    We are starting the next session Deploying and Managing Nebari
    Rich Signell
    @rsignell-usgs
    Have gitlab runners been tested?
    Rich Signell
    @rsignell-usgs
    Just wanted to report that the build of "disaster-pilot-2022/ogc-rps" environment was successful, and the geogif cell now runs in:
    https://esip-ogc.nebari.dev/hub/user-redirect/lab/tree/shared/disaster-pilot-2022/rsignell/geo-notebooks/notebooks/odc-landsat.ipynb
    Dharhas Pothina
    @dharhas
    Have gitlab runners been tested? We have two (I think) deployments that are running for clients on internally hosted gitlabs. The first client was the one that paid for that integration.
    Simon Gagnon
    @Sigagnon
    My organisation is also on a internally hosted Gitlab. @dharhas what these tested runners are doing so far ?
    Dharhas Pothina
    @dharhas
    Right now, mainly just the automated deployments. Nebari uses a GitOps/Infrastructure as Code approach. So the idea is you have a single yaml file that defines all your infra. If you want to manage this manually you can make changes to your yaml and then run nebari deploy. Instead what we usually do is put the code in a GitHub/GitLab repo and Nebari's integration with GitHub Actions/GitLab Runners. Then when you make a change to the config yaml and merge it into main the GitHub Actions/GitLab Runners automatically redeploy the latest changes.
    Dharhas Pothina
    @dharhas
    fyi. lots of good notebooks to try out here: https://gallery.pangeo.io/. You may need to make new conda environments for some of them but many should just work in the ogc or ogc-rps environments. Some of the older notebooks don't work because of changes in cloud datasets or package versions
    Reminder. The conference call starts in 5 mins. Lets have a short kickoff with folks who are around and y'all can explain the things you are working on and then we can work independantly.
    Rich Signell
    @rsignell-usgs
    Although I created one custom environment on our sprint Nebari endpoint successfully about 45 minutes ago, subsequent builds are failing with
    Traceback (most recent call last):
      File "/opt/conda-store-server/conda_store_server/build.py", line 179, in build_conda_environment
        set_build_completed(conda_store, build, output.encode("utf-8"), packages)
      File "/opt/conda-store-server/conda_store_server/build.py", line 45, in set_build_completed
        raise ValueError(
    ValueError: channel url=https://repo.anaconda.com/pkgs/main not recognized in conda-store channel database
    Seems like this might be some conda-forge issue?
    Kim Pevey
    @kcpevey
    That is the same error we saw yesterday with the super simple env demo. I'll ask about this internally
    eskild
    @iameskild
    @Sigagnon here are the docs for deploying nebari (qhub) on Minikube. There are a few differences between linux and wsl deployments which are called out.
    https://docs.qhub.dev/en/latest/source/dev_guide/minikube.html?highlight=wsl#note-for-development-on-windows-subsystem-for-linux-2
    Simon Gagnon
    @Sigagnon
    thanks @iameskild
    Dharhas Pothina
    @dharhas
    jitsi meeting name is esip-ogc
    Kim Pevey
    @kcpevey

    @rsignell-usgs RE: conda error above:
    This issue is documented here: Quansight/conda-store#299
    Chris expects to have this resolved in the next 2 weeks.

    A workaround for this fix is to add https://repo.anaconda.com/pkgs/main to the conda_store_config.py c.CondaStore.conda_allowed_channels.

    Josh Lieberman
    @lieberjosh
    Can you put the full link for the jitsi session here?
    Dharhas Pothina
    @dharhas
    looking for it.
    we are switching back to gotomeeting jitsi is too slow.
    Josh Lieberman
    @lieberjosh
    Anyone have further questions or would like to outline what you're trying out wrt nebari?
    We'll close the GoToMeeting session now. Support via this gitter room will remain available. We'll start up the GTM again at 1pm EDT to review any work with either the sample notebooks or participant's own notebooks that they would like to put forward.
    Josh Lieberman
    @lieberjosh
    The qhub-nebari sprint is concluded, but this room will continue to be available for comments and questions. We'll send out a link to the recordings from Day 1 as well as a link to a short questionnaire to help us understand how to do a better job next time of introducing people to dev ops tools such as nebari.
    eskild
    @iameskild
    Thank you @lieberjosh for organizing and thank you everyone else for participating! As a reminder, we will be shutting down the esip-ogc.nebari.dev shortly so if you have notebooks or files you wish to save, please download them as soon as you can. Thanks again :)