All notable changes to this project will be documented in this file.
The format is based on Keep a Changelog, and this project adheres to Semantic Versioning.
Not yet released
- Vertex agent now attempts to retry create custom job up to three times to recover from transient errors
Released June 15th, 2023.
prefect_gcp.projects
module. Useprefect_gcp.deployments
instead. - #189pull_project_from_gcs
step. Usepull_from_gcs
instead. - #189push_project_to_gcs
step. Usepush_to_gcs
instead. - #189PullProjectFromGcsOutput
step output. UsePullFromGcsOutput
instead. - #189PushProjectToGcsOutput
step output. UsePushToGcsOutput
instead. - #189
- Bug that
list_folders
method removes dot("."
)s in the middle of paths
Released on May 25th, 2023.
accelerator_count
property forVertexAICustomTrainingJob
- #174result_transformer
parameter to customize the return structure ofbigquery_query
- #176boot_disk_type
andboot_disk_size_gb
properties forVertexAICustomTrainingJob
- #177- Support to stream worker logs for executed flow runs - #183
Released on April 20th, 2023.
CloudRunWorker
for executing Prefect flows via Google Cloud Run - #172
- Fix
CloudRunJob
with VPC Connector usage. - #170
Released on April 6th, 2023.
pull_project_from_gcs
andpush_project_to_gcs
steps - #167
upload_from_dataframe
docstring - #162upload_from_dataframe
file extensions for compressed parquet ('.snappy.parquet', '.gz.parquet') - #166
Released on February 28th, 2023.
upload_from_dataframe
method inGcsBucket
- #140
- Using
GcsBucket
as a deployment storage option - #147 - Breaking: Stop decoding and return a
bytes
type inGcpSecret.read_secret
, as originally annotated - #149 - Resolving paths in
GcsBucket
unintentionally generating an arbitrary UUID when path is an empty string - #150
Released on February 7th, 2023.
get_directory
method inGcsBucket
returns the list of downloaded file paths and supports relative path forlocal_path
- #129.- Reporting the state of the
VertexAICustomTrainingJob
to the Prefect API by passing the base env in job spec - #132.
Released on January 27th, 2023.
- Listing blobs at the root folder - #120
Released on January 20th, 2023.
- Correctly format GCS paths on Windows machines - #117
Released on January 5th, 2023.
- Wrapping type annotations in quotes to prevent them from loading if the object is not found - #105
Released on January 3rd, 2023.
- The
CloudRunJob
timeout parameter is now passed to the GCP TaskSpec. This allows Cloud Run tasks to run for longer than their default of 10min - #99
- Improper imports on the top level - #100
Released on December 23rd, 2022.
- Adds handling for
service_account_info
supplied toGcpCredentials
as a json formatted string - #94
Released on December 22nd, 2022.
list_blobs
,download_object_to_path
,download_object_to_file_object
,download_folder_to_path
,upload_from_path
,upload_from_file_object
,upload_from_folder
methods inGcsBucket
- #85GcpSecret
block withread_secret
,write_secret
, anddelete_secret
methods - #86BigQueryWarehouse
block withget_connection
,fetch_one
,fetch_many
,fetch_all
,execute
,execute_many
, methods - #88
- Made
GcpCredentials.get_access_token
sync compatible - #80 - Breaking: Obfuscated
GcpCredentials.service_account_info
by usingSecretDict
type - #88 GcsBucket
additionally inherits fromObjectStorageBlock
- #85- Expose all blocks available in the collection to top level init - #88
- Inherit
CredentialsBlock
inGcpCredentials
- #92
- Warning stating
Failed to load collection 'prefect_gcp_aiplatform'
- #87
Released on December 5th, 2022.
VertexAICustomTrainingJob
infrastructure block - #75
Released on December 2nd, 2022.
CloudJobRun.kill
method for cancellation support - #76
Released on October 7th, 2022.
- Validation errors for CPU and Memory being raised incorrectly - #64
Released on September 28th, 2022.
- Invoke
google.auth.default
if bothservice_account_info
andservice_account_file
is not specified - #57
- Retrieving the
project_id
from service account orquota_project_id
from gcloud CLI ifproject
is not specified - #57
Released on September 19th, 2022.
CloudRunJob
infrastructure block - #48GcsBucket
block - #41external_config
keyword argument inbigquery_create_table
task - #53content_type
keyword argument incloud_storage_upload_blob_from_file
task - #47**kwargs
for all tasks in the modulecloud_storage.py
- #47
- Made
schema
keyword argument optional inbigquery_create_table
task, thus the position of the keyword changed - #53 - Allowed
~
character to be used in the path for service account file - #38
ValidationError
usingGcpCredentials.service_account_info
inprefect-dbt
- #44
Released on July 22nd, 2022.
- Added setup.py entry point - #35
Released on July 22nd, 2022.
- Updated tests to be compatible with core Prefect library (v2.0b9) and bumped required version - #30
- Converted GcpCredentials into a Block - #31.
Released on July 11th, 2022
- Improved error handle and instruction for extras - #18
Released on March 17th, 2022.
cloud_storage_copy_blob
,cloud_storage_create_bucket
,cloud_storage_download_blob_as_bytes
,cloud_storage_download_blob_to_file
,cloud_storage_upload_blob_from_file
, andcloud_storage_upload_blob_from_string
tasks - #1bigquery_create_table
,bigquery_insert_stream
,bigquery_load_cloud_storage
,bigquery_load_file
, andbigquery_query
, tasks - #2create_secret
,delete_secret
,delete_secret_version
,read_secret
, andupdate_secret
tasks - #3