[RELEASE] 2024.11.1 - Hotfix

[viniciusdc]

Release Checklist

Release details

Scheduled release date - 2024/11/01

Release captain responsible - @viniciusdc

Starting point - a new release is out

• Create this issue to track and discuss the upcoming release.
• Use the previous release issue for any final release-specific discussions, then close.
  • This can be a good time to debrief and discuss improvements to the release process.

Looking forward - planning

• Create milestone for next release (if it doesn't already exist) and link it back here.
• Triage bugs to determine what be should included in the release and add it to the milestone.
• What new features, if any, will be included in the release and add it to the milestone.
  • This will be, in large part, determined by the roadmap.
  • Is there a focus for this release (i.e. UX/UI, stabilization, etc.)?

Pre-release process

• Decide on a date for the release.
  • What outstanding issues need to be addressed?
  • Has documentation been updated appropriately?
  • Are there any breaking changes that should be highlighted?
  • Are there any upstream releases we are waiting on?
  • Do we need to update the dask versions in the nebari-dask?
  • Will there be an accompanying blog post?
• Prepare for the release.
  • Update the nebari upgrade for this release
    • Add upgrade messaging including deprecation warnings, version specific warnings and so on.
  • Optionally, announce a merge freeze.
  • Release Candidate (RC) cycle.
    • Is this a hotfix?
      • Create a new branch off of the last version tag.
        • Use this branch to cut the pre-release and the "official" release.
      • git cherry-pick the commits that should be included.
    • Cut RC via GHA release workflow (w/ "This is a pre-release" checked).
    • Perform end-to-end testing. Use the Testing Checklist template.
      • For minor releases, relying on the end-to-end integration tests might suffice.
    • End-user validation.
      • If possible, pull in volunteers to help test.
      • (Repeat steps if necessary)
  • Update RELEASE.md notes.

Cut the official release

If there were changes to the following packages, handle their releases before cutting a new release for Nebari

• Cut PyPI release for nebari-workflow-controller
• Cut PyPI release for argo-jupyter-scheduler

These steps must be actioned in the order they appear in this checklist.

• Tag, build and push docker images
• Update and cut release for nebari-dask meta package on Conda-Forge.
• Update CURRENT_RELEASE (and any other tags) in the constants.py
• Cut PyPI release via GHA release workflow.
  • Avoid appending v to tag.
    • Copy release notes from RELEASE.md.
• Merge automated release PR for nebari on Conda-Forge.
• Merge release branch into main

[viniciusdc]

For the changes made to this release, we have this tracking issue #2798

[viniciusdc]

Tested locally to evaluate fixes, waiting for user feedback

[kenafoster]

Upgrading 2024.7.1 to 2024.11.1rc1 - the prompt took me through the 2024.9.1 prompt before 11.1 - was this expected? I thought it was supposed to skip

(nebari-2024.11.1) (base) kfoster@Mac nebari_2024.11.1 % nebari --version
2024.11.1rc1
(nebari-2024.11.1) (base) kfoster@Mac nebari_2024.11.1 % nebari upgrade -c nebari-config.yaml 

---> Starting upgrade from 2024.7.1 to 2024.9.1

Setting nebari_version to 2024.9.1

Do you want to replace current tag 2024.7.1 with 2024.9.1 for:
default_images.jupyterhub: quay.io/nebari/nebari-jupyterhub? [Y/n]  (Y): y

Do you want to replace current tag 2024.7.1 with 2024.9.1 for:
default_images.jupyterlab: quay.io/nebari/nebari-jupyterlab? [Y/n]  (Y): y

Do you want to replace current tag 2024.7.1 with 2024.9.1 for:
default_images.dask_worker: quay.io/nebari/nebari-dask-worker? [Y/n]  (Y): y



---> Starting upgrade from 2024.9.1 to 2024.11.1

Setting nebari_version to 2024.11.1

Do you want to replace current tag 2024.9.1 with 2024.11.1 for:
default_images.jupyterhub: quay.io/nebari/nebari-jupyterhub? [Y/n]  (Y): y

Do you want to replace current tag 2024.9.1 with 2024.11.1 for:
default_images.jupyterlab: quay.io/nebari/nebari-jupyterlab? [Y/n]  (Y): y

Do you want to replace current tag 2024.9.1 with 2024.11.1 for:
default_images.dask_worker: quay.io/nebari/nebari-dask-worker? [Y/n]  (Y): y

[kenafoster]

Attempting 2024.9.1 -> 2024.11.1rc1 upgrade on an existing AWS cluster, I noticed this:

Would you like Nebari to assign the corresponding role/scopes to all of your current groups automatically? [y/N] (N): y
...
ValueError: Failed to connect to Keycloak server: 401: b'{"error":"invalid_grant","error_description":"Invalid user credentials"}'

The reason for this is that I have changed security.keycloak.initial_root_password from its value when Nebari was first deployed so that the true password isn't stored in config in our CICD repo. I believe this is a common (best?) practice, so relying on that value to connect to keycloak for the upgrade group creation step won't work.

[viniciusdc]

Upgrading 2024.7.1 to 2024.11.1rc1 - the prompt took me through the 2024.9.1 prompt before 11.1 - was this expected? I thought it was supposed to skip

This was expected the sense of skipping was merely in terms of not performing any action on behalf of the user, due to how the upgrade process works, there is not a skip process that we could use, something to enhance.

[viniciusdc]

The reason for this is that I have changed security.keycloak.initial_root_password from its value when Nebari was first deployed so that the true password isn't stored in config in our CICD repo. This is a common (best?) practice, so relying on that value to connect to keycloak for the upgrade group creation step won't work.

Uhm... indeed, that's something we are not considering in many of our tests, and as you commented, it is best practice. There is no easy way to acquire the new password programmatically, though, unless we request the user for it -- in case the attempt with the one in the yaml fails. What do you think?

We can also point the user on how to do this manually in case of an error (sounds like a good idea since it will be difficult to counter all the edge cases)

[dcmcand]

@kenafoster can you open a ticket about that issue? I don't think it is a release blocker, but it is a good callout and we should talk about how to best handle that situation in a separate issue. Thanks for all the work you are doing with testing!

[kenafoster]

Yep, here it is #2833

[viniciusdc]

I have taken care of the issues mentioned above and thoroughly tested them.

• Keycloak API calls now run correctly during nebari upgrade; added the ability to override the keycloak root password by env vars;
• launch_template field has been temporarily disabled due to issues presented [BUG] - JupyterLab Pods Fail to Mount Conda PV When on CUSTOM ami_type (AWS Only) #2832, a full fix will be introduced in 2024.11.2

• Tested upgrade path from 2024.7.1 -> 2024.11.1 and 2024.9.1 -> 2024.11.1;
• Tested fresh deployment and upgrade deployments
• Tested GPU usage since the ami_id logic is coupled with the instance_type classes;
• Tested user creation, groups, and spawning of user instances. All the other resources were left untouched; thus, the previous tests done for 2024.9.1 apply.

Note:
While performing the check, I encountered a weird issue with my cluster where it attempted to re-create all resources, I am assuming it was an issue with my stages folder not being in sync with the storage one, I was able to perform successfully deployments once I replicate the same action with clean stages files

While doing this process, I also took some time to think about how we want to perform the hotfixes in the future and what are some of the policies we should consider in case similar situation like this one appear.