- Consider new cate cloud
- Remove services along with deployment on cate launch
- Create new deployment everytime cate is launched
- do not try to create namespaces that do not exist (otherwise an error is raised)
- Added lifecycle handlers for logging
- Set version to 2.1.11.dev1
- Added an env var CATE_STORAGE_MODE (local, pvc) which allows switching off the creation of PVCs per user. This is useful for testing purposes
- Avoid pulling images from dockerhub
- Fixed ingress error when default value is used. Set Ingress port by using class V1IngressServiceBackend
- Removed CATE_USE_DAPR side container config. Not used.
- The RestApi has been extended by a new operation
GET /cubegens/versionthat allows querying versions that will be used when runningcubegensprocesses. This operation is public.
- Added a new configuration for cate. It adds a grace period after the deployment has been created (default: 2s)
- A grace period has been added after a cate pod is started. It can be set via the env variable
CATE_LAUNCH_GRACE_PERIOD - Reverted setting for Restart policy. Deployments only allow restart policy "Always" for containers.
- Set cate timeout default to 86400s (24 hours)
- Updated pydevd-pycharm version to 213.5744.223
- Read
cate_user_rootandcate_user_commandfrom environment - Method
get_pods_count()incate.pyonly counts pods that are part of the cate web application.
- Adapt to change of cate user from
catetoxcube
- Publish geoservice is now raising an error message when a workspace cannot be created
- Cost estimation function is now able to convert scientific notation to an int value for punit limits.
- The above conversion is now raising an
ApiErrorwhen the punit limit obtains a value that cannot be converted to int - Cate: Fixed that usernames get rejected if they contain dots, underscores and @s
- Added additional labels to deployments and jobs allowing to filter for them easier
- Allowed for annotations and labels to be added in create_deployment_object
- The Hub has now an additional options for xcube-gen subscription which is called
processingThis will allow the user to execute BYOA scripts
- xcube-hub is now collecting xcube-gen results for info from a json file provided by
xcube gen2. xcube gen2now returns a result object. This has been added to the hub's responses. The result set has been defined intentionally vaguely as the result can vary.- The Hub can now launch xcube servers
- Added optional dapr logging annotations to job pods.
- Data stores are not served any more by the hub. Hence, the data store validation file is not validated anymore.
- Cate and xcube image versions can now be configured using digest hashes.
- The feature store is now created independently of the workspace
- Improved error messages from the Geoserver (i.e. no empty ones)
- Updated geoserver-rest version to 2.0.0
- Added new operation get_all_collections for geoservices
- Added an operation PUT cubegens/code using content type multipart/form-data. This allows xcube-gen2 to upload the cube configuration as JSON file and user code as a zipped archive.
- The flask server can now be configured to run in debug mode using the en var
- Fixed wrongly names PV and PVC for stores. Used to be called datapools.
- Fixed that _raise_for_no_access would cause a 500 error when the database searched for does not exist
- Fixed issues when using inline user code for generating xcubes. The hub now stores any cube configuration/user codes in a persistent volume shared with the xcube gen2 Jobs. The xcube gen2 command line has been changed accordingly from using pipes to passing the config as file parameter.
- Added operations that allow publishing collections to a geoserver instance
- The web service can now be started as a command line client. It allows specifying host address and port. Was fix before.
- Added an operation that returns all wms services of a database
- The xcube hub is now preventing users to add a geoservice on databases they do not own
- The database name is now configurable for the geoservice
- Removed stores operation. Not used anymore
- Changed the way how tokens are handled when handling subscriptions to avoid random authorization issues
- A proper error is now raised when no collection is find when publishing collections to WMS service
- The hub is now using XCUBE_REPO and XCUBE_TAG allowing to configure xcube's version tag separately from its repo
- Removed stores operation. Not used anymore
- The hub is now using XCUBE_REPO and XCUBE_TAG allowing to configure xcube's version tag
- The database name is now configurable for the geoservice
- Fixed: A proper error is now raised when no collection is find when publishing collections to WMS service
- Added a validation for datapools config files
- Added a services operation to get info about the status of a running service
- Added subscriptions operations to handle service subscriptions
- Added a Cfg class to handle data pools configs
- Added the json schema for data pools
- Added a default value to mayberaise_for_env
- The xcube docker image is now shown in the service info
- When adding Subscription the client secrets can now be overridden
- api Responses do not have a result key anymore
- create cubegens now raises an error if no sufficient punits are available
store_paramsit not required anymore foroutput_configs- CDS credentials are now passed to the xcube-gen container via env vars
/storesis now only returning 'title', 'description', and 'store_id' to protect private information- The hub uses now an openapi flask stub to simplify server configuration
- The OpenApi definition has been overhauled to follow best practices in defining path patterns
- xcube-hub can now issue HS256 tokens via client credentials using an
/oauth/tokenhandler - The
/datastoreshandler has been replaced by a cubegens info handler - The
/cubegens/infohandler provides now much smaller data-pools information for better performance - Changed the CI from travis to appveyor
- The progress information is now returned by the
GET /cubegens/{cubegen_id}handler /callbackshandlers only allow PUT request. Any progress information is now handled byGET /cubgens/{cubgen_id}- The main function has been moved to
__main__.pyto simplify launching the hub inside docker containers - The Docker command has been simplified by using the python executable directly
- The authorization flow allows now to use both xcube-hub as well as auth0 issued tokens
- Added an Auth class that handles the creation of authorization class instances to handle different authorization issuers.
- Ensured that controllers that need the user_id and/or email will get it passed using the token_info function parameter.
- The token_info is now set during the token verification process and contains the user_id and email
- put_callback now receives the user's email address to be used in setting the punits for a user
- Changed
workedtoprogressfor Callbacks in openapi rest definition - The output data_id is now only set to the
job_idif thedata_idis not given - The
/cubegens/infoprocess has changed it now receives any necessary data for the punits calculation from xcube gen2 via a xcube gen2 job
- The cubegen info returns now the limit as an integer instead of a string
- Fixed error when may punits limit is string when coming from env var
- Receiving the users punits is now using email instead of user id
- okteto uses now a xhubehub-base image
- The default cache provider is now inmemory
- Changed parameter user_id to user_namespace in create_if_not_exist
- get_datastores accepts now yaml files
- wsgi will start a debug service if XCUBE_HUB_DEBUG is set to 1
- sizeandcost accepts now a None enddate. Defaults to now()
- Added an okteto service patch to allow external K8s services accessing the hub
- Added a setup.sh to setup dependencies in okteto
- The location of data-pools.yaml is changed to /etc/xcube
- use jobs are now spawned in the xcube-gen namespace instead of a namespace for each user
- The user_info in auth0 is now accepting checking whether the userinfo from the kv store is string
- The redis kv store converts bytes to string in the method 'get'
- Fixed user_info error which was causing the hub to frequently accessing the auth0 user_info api. That caused 429 errors from the auth0 api.
- Added an operation /status which returns a status set by the env vars XCUBEHUB_STATUS and XCUBE_HUB_STATUS_MESSAGE. Helps e.g. when an xchubehub instance shall be suspended.
- Added auth operators to manage users
- data-stores.json location is now configurable via environment variables.
- Set ingress websocket timeouts to 1 day
- Ensured that the grace period foe launching cate services is integer
- Getting a job status does now raise an HTTP error when the job had failed.
- The ApiError and webapi-response returns now outputs and tracebacks when applicable
- Removed poll_job_status. Not used.
- The cate ingress is now patched not created
- Cate instances are now run in namespace cate allowing for NFS permanent volumes in Kubernetes.
- Launch_cate is not returning the URL without https
- Set ingress timeouts to 1 day
- Ensured that the grace period foe launching cate services is integer
- Improved testing for K8s functions
- The launch of cate and xcube-gen-api handlers can now be configured
- Sets now correct xcube-gen-api version in Dockerfile's labels
- The error can now handle errors and tracebacks
- When the xcube webapi is started the service waits ofr a grace period in order to avoid issues with the xcube-gen-ui.json
- A username validation function was added to avoid conflicts with kubernetes naming conventions
- Cate is spawned with JUPYTERHUB_SERVICE_PREFIX. Fixes CORS errors for accessing ws tiles
- xcube gen adds a grace sleep when the cate webapi is started. Fixes issues when starting the cate webapi from the webui
- Supports now spawning of cate containers.
- Also supports Keycloak auth for cate
- Supports now cciodp and cds data stores
- can now handle callbacks using a KeyValue database (either redis or leveldb or inmemory)
- can now serve as a wsgi service
- Can now spawn an xcube webservice and a corresponding xcube viewer
- Includes now a poller for e.g. observing K8s processes
- Added a service mock to simplify service mocks
- An openapi yaml config file was added
- Authentication and authorization processes were strengthened
- handles bucket url config from xcube-gen-ui
- Synced version with xcube-gen-ui
- Updated computation of processing units. Now also "cciodp" datastore is supported.
- Demo version for ESA SRR-F meeting.
- Using environment variables AWS_ACCESS_KEY_ID and AWS_SECRET_ACCESS_KEY to pass AWS credentials to object database.
- Implemented dummy viewer operations
POST /cubes/<user_id>/viewerPOST /mock/cubes/<user_id>/viewer
- create user namespace if doesn't exist
- use dedicated AWS user credentials to manage bucket
eurodatacube - use special scopes for
PUT /users/<user_id>/punits - job output/logs should be separate from job status as this quickly becomes much data
(e.g. xcube-sh may print warnings on every SH request).
Therefore:
GET /jobs/<user_id>/<job_id>/logs - job result dicts are still not consistent, key
job_idvs.jobor not matching{status:'ok', result: any}pattern. service.pyand others: renameuser_nameintouser_id- fix: if we send invalid JSON requests we get error 400
<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 3.2 Final//EN">
<title>400 Bad Request</title>
<h1>Bad Request</h1>
<p>The browser (or proxy) sent a request that this server could not understand.</p>