CHANGELOG.md

Change log

Nublado is versioned with semver. Dependencies are updated to the latest available version during each release. Those changes are not noted here explicitly.

Find changes for the upcoming release in the project's changelog.d directory.

13.0.0 (2026-04-27)

Backwards-incompatible changes

• Rename the Rubin menu to Jobs in the standard extensions installed in the JupyterLab base image. Beneath that, rename Query to TAP Query. This menu now separates jobs by dataset, although jobs may be listed under multiple datasets if the extension cannot determine what dataset that job was querying.
• Change the default runtime mount location to /etc/nublado from /opt/lsst/software/jupyterlab. This change should be invisible to users since all code that relies on its location should be using the NUBLADO_RUNTIME_MOUNTS_DIR environment variable to find it.
• Stop creating an .access_token symlink in the user's home directory to the injected Gafaelfawr access token. The lsst.rsp library, which should be used by most payloads, knows how to find the access token, and other languages should use the NUBLADO_TOKEN environment variable.
• Semantic version filters in the dropdown menu are no longer supported for weekly and daily images. Switching between calendar versions and semantic versions for cutoffVersion was potentially confusing.
• Use NUBLADO_ as the prefix for environment variables to configure the nublado purger command-line tool instead of RSP_SCRATCHPURGER_. This brings it in line with the rest of Nublado.

New features

• Add a new command, nublado images list, to list the available Nublado image tags given an image source configuration.
• Add a new command, nublado images prune, to delete some Nublado images from a repository based on a pruning policy.
• Add a new command, nublado images delete, to delete specific images by tag.
• Provide the lab's Gafaelfawr token in the NUBLADO_TOKEN environment variable as well as the ACCESS_TOKEN environment variable. The latter will eventually be deprecated.
• Add a new cutoffDate attribute for dropdown menu filtering, used to restrict display of daily and weekly images to only those after a given date. Filtering of experimental images can now be done with both cutoffVersion and cutoffDate, with the former applying to the semantic version of experimental builds based on releases and release candidates and the latter applying to builds based on weeklies and dailies.
• Add a new cutoffBuild attribute for dropdown menu filtering, used to restrict display of images with build numbers to only those with larger or equal build numbers. Images without build numbers are ignored.

Bug fixes

• Restore lab startup code to increase log output limits. This was accidentally dropped as part of the init container changes.
• Retry Google Artifact Registry requests on internal service errors as well as service unavailable errors. These errors also seem to be transitory problems with the Google API.

Other changes

• Relax the dependency on Safir in rubin.nublado.client to allow Safir 15.0.0.

12.1.0 (2026-03-27)

New features

• Add configuration setting for the timeout for prepulling images. The default of ten minutes may be too low for large images and slow networks.

Bug fixes

• Mount the lab startup volume read-only in the lab container. It should only be written to by the init container and should not be available to the user as a writable directory.
• Always force the lab startup volume to use tmpfs instead of following the configuration for /tmp.
• Update JupyterHub to 5.4.4 (Zero to JupyterHub 4.3.3).

Other changes

• Remove unused packages from the JupyterLab virtualenv in the jupyterlab-base image: hatchling, batch-nodejs-version, firefly-client, ipython-genutils, and nbval.

12.0.2 (2026-03-05)

Bug fixes

• Fix incorrect double-decoding of the body when reporting Kubernetes errors.
• Handle unexpected messages on a Kubernetes watch by logging them and restarting the watch, rather than raising an exception and failing the operation.
• Downgrade Google Artifact Registry retries to warnings so that they're not reported by Sentry.
• Change the element ordering and sort order for RSP versions to list and sort cycle and cycle build number before the RSP build number. The latest cycle build should be preferred regardless of the RSP build. Currently, no tags contain both values, so this should not change existing menu displays.
• Fix the display name of resolved aliases to use the base tag name without any cycle or architecture information.
• Fix the sort order of unknown and alias tags with an architecture to sort architecture-independent tags before architecture-specific tags, matching the behavior of non-alias tags.

12.0.1 (2026-02-19)

Bug fixes

• Always filter out architecture-specific images rather than including them if the non-architecture-specific image with the same tag is not present. This removes images for foreign architectures from the prepull targets and the drop-down menu.
• Retry requests to Google Artifact Registry to retrieve the list of images up to three times with a ten second pause between attempts before reporting a fatal error. This will hopefully reduce log noise from the intermittant Google API failures.

12.0.0 (2026-02-10)

Backwards-incompatible changes

• Move the startup files created by the startup init container from /lab_startup to /etc/nublado/startup.
• Simplify the lab initialization code to merge environment variables with the environment set by the Nublado controller and remove some error handling that should only trigger if the Nublado init containers do not run correctly. This means labs built with the new jupyterlab-base will fail with pre-11.0.0 versions of Nublado by failing to start, not starting in degraded mode.
• In the Nublado startup init container, only put environment variable overrides in /etc/nublado/startup/env.json, not the full environment of the init container, to avoid leaking init-specific environment variables into the lab container.

New features

• Add controller.watch_reconnect_timeout config setting. Sometimes, Kubernetes watch connections can frequently drop with errors, or worse, they can silently stop receiveing events. This new setting specifies the amount of time to keep the connection open before explicitly reconnecting.

11.2.2 (2026-02-03)

Bug fixes

• Fix installation of files into the user's home directory by the startup init container.

11.2.1 (2026-01-21)

Bug fixes

• Remove filtering of the environment variables propagated to the user's lab, which broke Phalanx support for setting arbitrary environment variables.

11.2.0 (2026-01-14)

New features

• Add clear_local_site_packages parameter to notebook execution, which tells the lab to remove all locally-installed Python packages before executing the notebook.

11.1.0 (2026-01-06)

New features

• Switch from TeX to typst for PDF export of notebooks inside the jupyterlab-base container for Jupyter labs.

11.0.0 (2026-01-02)

Backwards-incompatible changes

• Remove configuration of the image and command for the fsadmin pod. This pod will now always use the general Nublado container and a keep-alive loop for the startup command.
• Rename the config.lab.tmpSource setting to config.lab.emptyDirSource. This setting now controls the Kubernetes emptyDir configuration for both /tmp and the new /lab_startup directory.
• Remove configuration for the fileserver image. This will now use the general Nublado container.

New features

• Move the JupyterLab initialization logic from lsst.rsp.startup into the Nublado container and run it as an init container. It communicates with the startup code via a new /lab_startup directory, allowing the lab startup code in the main container to be much simpler.
• Add new configuration option config.lab.standardInithome that, if set to true, adds a standard home directory init container to user pods. This is the same as the previous nublado-inithome container and is a privileged init container that creates the user home directory if it doesn't exist.
• Add new configuration option config.lab.homeVolumeName, used to determine which volume is mounted for the standard inithome init container.
• If RSP_SITE_TYPE is set to science, add an additional init container to user labs that ensures the tutorial landing page is copied into the user's home directory before the lab is started.
• Add the user WebDAV file server implementation to the Nublado repository and container.
• Add new controller.config.lab.namespaceAnnotations setting to add annotations to the per-user namespaces. The value may be a template in which {{uid}} will be replaced by the user's UID and {{gid}} will be replaced by the user's GID.

10.0.0 (2025-12-16)

Backwards-incompatible changes

• Merge the nublado-inithome container with other support containers and the Nublado controller into a single nublado container. Configurations using nublado-inithome as an init container should change the container to nublado and specify nublado inithome as the container command.
• Use the new Gafaelfawr client in the Nublado client mock to manage and verify tokens. Test suites that use the Jupyter mock from the Nublado client now must also mock Gafaelfawr and use the mock Gafaelfawr to generate a token and register corresponding user information before calling the Jupyter mock.
• Drop support for Python 3.12 in the Nublado client.

New features

• Add support for specifying the command to run in lab init containers and the file system admin pod.
• Use Repertoire in the Nublado controller to find the Gafaelfawr API instead of assuming the URL structure of the local Phalanx environment.
• Add the nublado command-line tool, which includes home directory provisioning, file system purging, and other supporting functionality, to the default file administration container.

Bug fixes

• Stop imposing restrictions on usernames and group names in the Nublado controller apart from those already imposed by Gafaelfawr.
• Update JupyterHub to 5.4.3.

9.0.3 (2025-12-03)

Bug fixes

• Use updated Safir so that app metrics won't break the app in rare situations if the underlying Kafka infrastructure is down.

9.0.2 (2025-12-03)

Bug fixes

• Allow user group names where the only alphabetic character is the first one.

9.0.1 (2025-11-19)

New features

• Add support for pretty-printed ADQL queries to jupyterlab-base.

9.0.0 (2025-11-13)

Backwards-incompatible changes

Nublado client API changed significantly

The API of the Nublado client provided as the rubin.nublado.client PyPI module has changed significantly. Users of that API will need updates for this version.

• Export all Nublado client symbols from rubin.nublado.client and make the internal module structure private. Any imports from rubin.nublado.client.exceptions or rubin.nublado.client.models must be changed to instead import from the top-level module.
• Use Repertoire for service discovery in the Nublado client. NubladoClient no longer supports configuration with a base URL; instead, to use a custom service discovery configuration, create a Repertoire DiscoveryClient and pass it as an optional parameter to NubladoClient.
• Remove the GafaelfawrUser class from the Nublado client. Callers should instead pass separate username and token arguments into the constructor.
• Delete JupyterLabSession.run_notebook_via_rsp_extension and replace it with NubladoClient.run_notebook, which uses the same extension but does not require creating a session first. The notebook content to execute must be passed in; providing a path relative to the user's Nublado home directory is not supported.
• Change the field names of the error portion of a notebook execution result to be more general and not exactly echo the nbclient exception attributes.
• Delete JupyterLabSession.run_notebook and JupyterLabSession.run_notebook_cell. Use JupyterLabSession.run_python to run each cell instead.
• Stop exposing username from NubladoClient as a public attribute.
• Rename NubladoClient.open_lab_session to NubladoClient.lab_session to make it more obvious that it returns a context manager, not an open session.
• All the exceptions raised by the Nublado client have changed names. All exceptions inherit from NubladoError, which has a context attribute that can be used to add execution context information in the form of the CodeContext model.
• Reject unknown attributes and adopt an alias of class for image_class in the NubladoImage class provided by the client, allowing it to be used as a child model of services with YAML-based configuration.
• Use service discovery in the Nublado client mock to determine what URLs to mock. There is no longer a way to independently configure the base URL; instead, ensure service discovery is mocked via Repertoire's mock support before mocking Jupyter.
• Simplify setup of the Nublado client mock. Fixtures should now use register_mock_jupyter as an async context manager and yield its result (a MockJupyter object), and no longer need to directly handle WebSocket patching.
• Rewrite the API of the JupyterHub and JupyterLab mock provided as part of the client. Users of the mock should review the new API. Methods have changed names, and all information previously available or changeable via attributes is now done with methods.

Other backwards-incompatible changes

• Enforce camel-case settings in the purger configuration.

New features

• Add new NubladoClient.wait_for_spawn method that can be used if the caller doesn't care about the spawn progress messages.
• Provide MockJupyter.create_mock_token to create a mock Gafaelfawr token usable for authentication to the Jupyter mock provided by the Nublado client.
• Add support for specifying a specific kernel in NubladoClient.run_notebook.
• Provide pretty-printed tooltips for user SQL query history in the base JupyterLab environment.

Bug fixes

• Fix display of architecture-specific lab image tags when architectures are split across pages of a paginated tag list.
• Update JupyterHub to 5.4.2.

8.17.1 (2025-10-17)

Bug fixes

• Enable refresh_pre_stop in the Gafaelfawr authenticator plugin to JupyterHub in the hope that this will fix some rarely-seen problems with long-running labs.
• Update JupyterHub to 5.4.1 (Zero to JupyterHub 4.3.1).

Other changes

• Update to the latest Repertoire client for compatibility with the new service discovery model. This release requires Repertoire 0.6.0 or later.
• Test the Nublado client and JuypterHub plugins with Python 3.14 as well as 3.12 and 3.13.

8.17.0 (2025-10-14)

New features

• Use Repertoire service discovery in the JupyterHub spawner plugin to find the Nublado controller API.
• Enable Sentry and Slack error reporting in the file purger.

Bug fixes

• Support pagination when listing the tags for lab images from a Docker repository in the Nublado controller. This is required when using GitHub Container Registry as an image source if there are a significant number of image tags.
• Send the correct Accept header when requesting container checksums from a Docker registry that may contain multi-architecture images.
• Ignore prepuller updates for nodes that no longer exist instead of raising uncaught exceptions and pausing the prepuller.

8.16.0 (2025-10-08)

New features

• Mount Repertoire discovery information at /etc/nublado/discovery/v1.json in user notebook pods.
• Use Repertoire service discovery to find the logout URL for the JupyterHub Gafaelfawr plugin.
• Export of notebooks to PDF in the jupyterlab-base image now uses TeX instead of Chromium, which shrinks the size of the image by over 1GiB.
• Instrument client exceptions with Sentry metadata.
• Publish multi-platform images that support both linux/amd64 and linux/arm64.

Bug fixes

• Always use user for the username in logging contexts, and include username in more log messages.
• Filter out platform-specific tags from the spawner menu when corresponding generic tag exists.

Other changes

• The jupyterlab-base image is now based on the Debian trixie release.
• Use uv workspaces and uv lock files to manage packages and dependencies.

8.15.0 (2025-09-16)

New features

• Optionally report Nublado controller errors to Sentry.

8.14.0 (2025-09-11)

New features

• Resource requests for user lab pods can (and must) be specified explicitly. Before this change, only resource limits were specified and requests were calculated to be 1/4 of the limits. The controller.config.lab.sizes entries now have a different format.

  Old:

  sizes:
    - size: "small"
      cpu: 1.0      # This was a limit
      memory: "4Gi" # This was a limit

  New:

  sizes:
    - size: "small"
      resources:
        limits:
          cpu: 1.0
          memory: "4Gi"
        requests:
          cpu: 0.25
          memory: "0.75Gi"

Bug fixes

• Push the purger Docker image to the correct Google Artifact Registry rather than overwriting the inithome container.

8.13.1 (2025-08-28)

Other changes

• Adopt updated Safir.

8.13.0 (2025-08-28)

New features

• Added Nublado controller route to control administrative filesystem pod. This allows a user with an admin token to spawn a pod that mounts user file systems with administrative privileges. That pod can then be entered by the administrator with kubectl exec for remedial or forensic purposes. The pod can also be deleted through this route, and the status of the pod can be queried.

Bug fixes

• Install ipython-genutils in jupyterlab-base since ipympl has an implicit dependency on it.

8.12.2 (2025-08-15)

Other changes

• Rebuild jupyterlab-base with newer dependencies.

8.12.1 (2025-08-07)

Bug fixes

• Improve handling of exceptions in the notebook execution extension.

8.12.0 (2025-08-07)

New features

• Include .config in the list of directories moved aside when the user asks to reset their user environment.

8.11.0 (2025-08-01)

New features

• Add a build number as an optional portion of the image tag format.

Bug fixes

• Fix the link to the WebDAV documentation in the HTML page returned after a user's file server was started.

8.10.0 (2025-07-16)

New features

• Add a cron job to purge file systems of old files. This was formerly maintained in the lsst-sqre/rsp-scratchpurger repository.

8.9.2 (2025-07-02)

Bug fixes

• Accept group names from Gafaelfawr that start with digits as long as they contain at least one letter. Group names of that type can arise from per-user groups for users with a username starting with a digit.

8.9.1 (2025-06-28)

Bug fixes

• Rebuild jupyterlab-base with lsst-rsp 0.9.4, which fixes a bug in overquota handling when a welcome page is configured and presents a better overquota landing page.

8.9.0 (2025-06-26)

New features

• Add new controller.config.lab.defaultSize configuration option that determines the default lab size in the spawner menu. If not set, or set to a lab size the user does not have sufficient quota to spawn, the default continues to be the first size the user is allowed to spawn.
• Add controller.config.fileserver.reconcileInterval configuration option to control the frequency with which file server state is reconciled with Kubernetes.
• Add controller.config.lab.reconcileInterval configuration option to control the frequency with which lab state is reconciled with Kubernetes.
• Add controller.config.images.refreshInterval configuration option to control how often the image source is checked for new images and the Kubernetes nodes are checked for the list of cached images.

Bug fixes

• Fix a race condition during file server Kubernetes reconciliation where a file server in the process of being created, but waiting for the ingress to become valid, could be deleted by the background reconciliation job.
• Fix background task timing with very short intervals. This will probably never arise in a production configuration but was triggered by the test suite.

8.8.9 (2025-06-23)

Bug fixes

• Simplify the HTML page shown after the user spawns a file server and point to the RSP documentation.
• Standardize on write:files as the scope for accessing the file server. This was already required to spawn the file server, but exec:notebook was still used to control access to the spawned file server.

8.8.8 (2025-06-16)

Bug fixes

• Update JupyterHub to 5.3.0 (Zero to JupyterHub 4.2.0).

8.8.7 (2025-06-11)

Other changes

• Document how to customize Nublado JupyterLab behavior.

8.8.6 (2025-05-21)

New features

• Update the base Jupyter Lab extensions to support specifying the kernel to use via the /execution endpoint.

Bug fixes

• Fix builds when jupyter labextension list exits with a non-zero status.

8.8.5 (2025-05-16)

Bug fixes

• When building the list of available lab images, only include versions from tags on the lab image name, not other Docker images in the same repository.

8.8.4 (2025-04-28)

New features

• When tagging jupyterlab-base for a new release, also move the latest tag to the new release.

8.8.3 (2025-04-23)

New features

• Add less to the base JupyterLab container image.

Bug fixes

• Fix the install locations of configuration files in the base JupyterLab container image.

8.8.2 (2025-04-03)

Bug fixes

• Push inithome containers to Google as well as GitHub since GitHub has a low rate limit on anonymous API requests.

Other changes

• Update jupyterlab-server in the base JupyterLab container image.

8.8.1 (2025-03-27)

Bug fixes

• Fix uploads of rubin-nublado-client to PyPI.

8.8.0 (2025-03-25)

New features

• Support setting the interval for activity reporting from JupyterLab in the Nubaldo controller configuration.

Bug fixes

• Add a configuration setting for where to redirect the user after logout from JupyterHub. When user subdomains are in use, this needs to point to /logout at the base URL for the Science Platform, not /logout at the current hostname, which may be the JupyterHub hostname and thus create an infinite redirect loop.
• Use the configured delete timeout when deleting invalid labs instead of a hard-coded 30s timeout.

8.7.1 (2025-03-19)

Bug fixes

• Fix the NetworkPolicy created for user labs so that it actually restricts access to only JupyterHub, its proxy, and the lab itself as was intended.

8.7.0 (2025-03-17)

New features

• Add support for per-user subdomains to the Nublado client. If they are enabled, they will be automatically detected and the Nublado client will adjust its HTTP requests accordingly.
• Add support to the Nublado client testing mocks for simulating per-user subdomains.
• Add config.allowOptions to the GafaelfawrIngress for file servers so that the server will work correctly with Gafaelfawr 13.0.0 and later.

Bug fixes

• Set Sec-Fetch-Mode in several places in the Nublado client to suppress harmless but annoying warnings in the JupyterHub logs.

8.6.0 (2025-02-26)

New features

• Add support to the Nublado controller for restricting the list of images shown in the dropdown menu based on age, count within a category, or a version cutoff.

Bug fixes

• Fix an unbound variable error in the Nublado client in one error handling situation.

8.5.0 (2025-02-24)

Other changes

• Raise the upper bound on the Safir in the Nublado client to <11.

8.4.2 (2025-02-19)

Bug fixes

• Fix TypeError exception when trying to establish websocket connections. Version 14 of websockets changed the signature of the connect method.

8.4.1 (2025-02-12)

Bug fixes

• Wait for the default service account of the user's lab namespace to be created before creating the Pod object. Creation of the service account can be slow when Kubernetes is busy, and creation of the Pod object will fail if the service account does not exist.
• Remove the limit on the connection pool used to contact the Nublado controller from the JupyterHub spawner, since otherwise JupyterHub stops being able to do any work once the connections waiting for spawn progress exhaust the connect pool.
• Remove the limit on the Kubernetes client connection pool in the Nublado controller. This will allow the controller to scale to more than 100 (the default) simultaneous watches.
• Remove the limit on the connection pool used to look up users in Gafaelfawr and query a Docker image repository.
• Fix unsafe data structure manipulation when deleting completed labs in the Nublado controller background reconciliation thread.

8.4.0 (2025-02-04)

New features

• Log metrics events in the Nublado controller for lab spawn successes and failures and a count of the number of active labs.

Bug fixes

• Do not query Kubernetes for pod status when responding to status requests, and instead assume the internal state (which is reconciled periodically) is correct. The constant Kubernetes API requests for Pod status seemed to be overwhelming the control plane when there were a lot of pods running. Continue to ask Kubernetes directly immediately before spawn.
• Report Kubernetes operation timeouts properly when they cause a spawn failure rather than throwing an uncaught exception.
• Add tolerations to prepuller pods as well as lab and file server pods so that images can be prepulled to tainted nodes.

8.3.0 (2025-01-24)

New features

• Support the new spawn flag in the user's Gafaelfawr metadata. If set to false, tell the user that spawns aren't allowed rather than presenting a menu and reject spawns in the Nublado controller with a 403 error.

Bug fixes

• Return a spawner form showing only an error rather than an HTTP failure if the user's quota does not allow them to spawn any of the configured notebook sizes.

Other changes

• Drop support for Gafaelfawr groups without GIDs. Gafaelfawr has required all groups have GIDs since Gafaelfawr 11.0.0.
• Add documentation explaining how to use node selection and recommending nodeSelector over affinity.

8.2.0 (2024-12-12)

New features

• Add a service label to GafaelfawrIngress resources created for user file servers for proper Gafaelfawr metrics reporting.

Other changes

• Change the base image for JupyterHub to quay.io/jupyterhub/k8s-hub from jupyterhub/jupyterhub. This means the JupyterHub image now uses Python 3.12.

8.1.0 (2024-12-12)

Bug fixes

• Remove the Token link from the JupyterHub page template, since user tokens for JupyterHub are not supported on the Rubin Science Platform.

Other changes

• Update to JupyterHub 5.2.1.

8.0.3 (2024-11-18)

Bug fixes

• When reporting HTTP errors from the Nublado client, truncate the response body at the start rather than at the end. This makes it more likely that the error message from JupyterHub will appear in the truncated response.
• Drop the XSRF token and cookies before performing a JupyterHub login in the Nublado client. The client previously hung on to the XSRF token indefinitely, which resulted in errors if the token was expired in JupyterHub, such as by user session expiration.

8.0.2 (2024-10-31)

Bug fixes

• Fix broken formatting in error messages reported by the Nublado client.

8.0.1 (2024-10-31)

Bug fixes

• Improve error reporting of exceptions in the Nublado client and sanitize the reported body to remove some security tokens.

8.0.0 (2024-10-22)

Backwards-incompatible changes

• Refactor exception handling in NubladoClient to incorporate optional code context information (used by mobu) and additional exception metadata.

New features

• Provide JupyterLabSession as an exported class from rubin.nublado.client. This class represents an open WebSocket session with a Jupyter lab.
• Add support for artificial Gafaeflawr tokens to the MockJuypter mock of the JupyterHub and JupyterLab API. This allows the mock to extract a username from mock tokens sent by the code under test to the mocked APIs, rather than requiring the test client send an X-Auth-Request-User header.
• Use the most recent Nublado release tag as the default base image for sciplat-lab container builds.

7.2.0 (2024-10-01)

New features

• Add a Docker image build for a jupyterlab-base image, which provides a basic image that can be spawned as a lab container by Nublado and can be used as the basis for more complex images.
• Add the Docker image build for sciplat-lab, an image built on top of jupyterlab-base that provides a JupyterLab kernel that includes the Rubin Science Pipelines Python stack.

Bug fixes

• Revert canonical PyPI module name back to rubin-nublado-client for consistency with other projects. As before, this change should not affect pip install; either form of the name should work.

7.1.2 (2024-09-23)

Bug fixes

• Rename PyPI module for the Nublado client to rubin.nublado.client. Either name should work for pip install.

7.1.1 (2024-09-23)

Bug fixes

• Push the new rubin-nublado-client module to PyPI on release.

7.1.0 (2024-09-23)

New features

• Add the Python module rubin-nublado-client, which provides a client library for interacting with the Nublado-modified JupyterHub and JupyterLab services.

7.0.0 (2024-08-19)

Backwards-incompatible changes

• The /tmp directory in a lab pod now defaults to a tmpfs file system capped at 25% of the pod memory. Add a new configuration option to select between this default and the previous default of uncapped node-local storage.

6.3.0 (2024-08-15)

New features

• All timeout configuration options now support the syntax parsed by Safir's parse_timedelta and therefore support human-friendly durations such as 6h or 5m.

Bug fixes

• Fix crash of the controller during startup if a Kubernetes node reports a cached image with no names.
• Fix bootstrapping of a development environment in an existing virtualenv. Previously, uv was not installed before nox attempted to use it.
• Work around a bug in sphinxcontrib-redoc that prevented building the documentation twice without errors.

6.2.0 (2024-06-13)

New features

• Add configuration settings for the lab launch command and configuration directory.

6.1.0 (2024-06-06)

New features

• Add limits and requests to prepulled pods.

Other changes

• Update the underlying JupyterHub implementation to JupyterHub 5.0.0.
• Switch to uv for package management.

6.0.2 (2024-04-18)

Bug fixes

• Update JupyterHub to 4.1.5, which fixes more issues with XSRF handling.
• Move the .eups directory when performing user environment resets.

6.0.1 (2024-04-01)

Bug fixes

• Update JupyterHub to 4.1.4, which fixes more XSRF cookie issues.

6.0.0 (2024-03-28)

Backwards-incompatible changes

• Move the admin route for deleting a user's file server from /nublado/fileserver/v1/<username> to /nublado/fileserver/v1/users/<username> to better align with other routes and REST semantics.

New features

• Add an admin-authenticated /nublado/fileserver/v1/users/<username> GET route to get the status of a user's running file server (currently only 404 if not running or 200 with trivial content if running).
• Add a /nublado/fileserver/v1/user-status route a user to get the status of their own file server, which similarly returns 404 or 200 with trivial content.

Bug fixes

• Update JupyterHub to 4.1.3, which includes several fixes for XSRF handling.

5.0.0 (2024-03-21)

Backwards-incompatible changes

• Update to JupyterHub 4.1.0. This release has tighter XSRF handling than previous versions. Clients that talk directly to JupyterHub rather than using a browser, such as mobu or Noteburst, will need to be updated to support the stricter XSRF requirements.

Other changes

• Nublado now uses uv to maintain frozen dependencies.

4.0.2 (2024-01-19)

Bug fixes

• Update to Safir 5.2.0, which rewrites the middleware to avoid the Starlette BaseHTTPMiddleware class. This should hopefully produce better error reporting in some cases where exceptions were being mangled and lost by the BaseHTTPMiddleware logic.

4.0.1 (2024-01-12)

Bug fixes

• All user file servers are now protected by a NetworkPolicy that prevents connections except via Gafaelfawr, ensuring that authentication is properly enforced.
• Stop returning the requested lab environment from the lab status API. Nothing uses this information and it may contain secrets that should not be this readily accessible.
• Return the actual running Docker image reference from the lab status API rather than the form parameters the user sent when requesting a lab.
• Return prepuller configuration in the API response with snake-case fields rather than camel-case fields, as was originally intended.
• Restrict access to controller routes via the admin:jupyterlab scope to only those routes that JupyterHub needs to use. Access to other administrative routes is now controlled with exec:admin.
• Fix the response type for the /spawner/v1/labs/{username}/events route in the OpenAPI schema.
• Quietly retry the file server pod watch after 410 errors, since under some circumstances they appear to happen every five minutes. Remove the delay when restarting after a 410 error without a resource version, since this appears to be a normal Kubernetes API response and the delay could miss events.

Other changes

• Change the default Argo CD application for user file servers to nublado-fileservers. Continue to use fileservers as the default namespace, since the nublado-fileservers namespace would conflict with the reserved namespace pattern for user labs.
• Classify the JSON API routes with tags that reflect who has access to that API.
• Add documentation on how to set up Google Artifact Registry as an image source for Nublado, and on why this is the recommended configuration when Nublado is running on Google Kubernetes Engine.

4.0.0 (2024-01-02)

This is the first release of the new merged Nublado release. It contains the Nubaldo controller, a JupyterHub spawner implementation that uses the controller to create user labs, a JupyterHub authenticator implementation to use Gafaelfawr, and the Docker configuration to build a custom JupyterHub image containing that spawner and authenticator.

In previous versions of Nublado, the equivalents of these components were maintained in separate repositories with independent version numbers. As of this release, all of these components are maintained and released together using semver versioning. Further changes will be documented in this unified change log.