Lithops

Latest version: v3.4.1

Safety actively analyzes 638720 Python packages for vulnerabilities to keep your Python projects secure.

Scan your dependencies

Page 5 of 15

2.5.0

Added
- [CLI] Add new command in cli to list deployed runtimes
- [Standalone] Add reuse mode that allows to reuse the same VMs for all the maps
- [Config] Allow to configure worker_processes parameter in serverless and standalone sections
- [Localhost] Prevent multiple jobs in the same executor to run at the same time
- [Standalone] Prevent multiple jobs submitted to the same master VM to run at the same time
- [CE] Added COS Direct endpoints for free bandwidth from/to CodeEngine

Changed
- [Core] worker_processes parameter has been moved from map() to FunctionExecutor()
- [CE] Deleted docker naming restrictions in CE and k8s backends
- [CLI] Prevent loading storage config when using 'lithops runtime build' command
- [AWS Lambda] Moved layer build to a lambda, solves OS related errors when compiling libraries
- [AWS Lambda] Adjusted new memory configurations (128 MB minimum and removed 64 MB increments check)
- [AWS Lambda] Add support for Python3.9
- [Standalone] ssh VM password is now a 37 chars random and dynamic password (for create and resue modes)

Fixed
- [CE] Create a new token when it expires after 20 minutes when using the same FunctionExecutor
- [CE] Prevent exception when detecting the docker username in k8s and CE backends
- [Core] Fix minor issue in jobrunner
- [AWS Lambda] Fixed aws config max timeout check
- [Standalone] Fixed issue that prevents to run multiple maps() with the same FunctionExecutor (create mode)

2.4.1

Fixed
- [IBM VPC] Fixed a data inconsistency on consume mode

2.4.0

Added
- [Storage] Added MinIO storage backend
- [Core] Allow to pass function args as part of the invocation payload in FaaS backends
- [Core] Optimize call_async() calls with an internal function caching system
- [AWS Lambda] New invoke method that reduces total invocation time
- [Core] Allow to set the bucket name at storage backend level in config
- [localhost] stop running jobs processes on ctrl-c or exception
- [AWS S3] Added region_name parameter in config

Changed
- [Core] Improved performance and efficiency of the lithops cleaner background process
- [AWS Lambda] Use layer from Klayers API for pre-compiled Amazon Linux numpy binaries
- [Core] Moved invoke_pool_threads param from map and map_reduce calls. Now it must be set at backend level in config

Fixed
- [Localhost] Fixed error when processing localhost objects
- [Localhost] Allow to create a localhost storage instance when a config file exists with a cloud configuration
- [Core] Fixed an unusual inconsistency in configuration between 'backend' and 'mode' parameters
- [Core] Fixed customized_runtime feature
- [Core] Fixed get_result() execution after wait() when throw_except is set to False
- [Core] Fixed internal executions
- [Core] Fixed 'lithops storage list' CLI when a bucket is empty
- [Standalone] Fixed execution

2.3.5

Added
- [Core] Add function chaining pattern in the Futures API
- [Core] ob.data_stream is now also an iterator when using the partitioner
- [AWS Lambda] Add 'account_id' parameter in config (used if present instead of querying STS).
- [k8s] Allow to set the maximum number of parallel workers
- [Localhost] Allow the partitioner to process local directories

Changed
- [Core] Add 'key' and 'bucket' attrs in localhost partitioner for compatibility with OS
- [Serverless] runtime, runtime_memory and runtime_timeout can only be set at backend level

Fixed
- [Standalone] Fix execution
- [Core] Avoid loading the config file twice

Deleted
- [AWS Lambda] Using custom layer runtimes for AWS Lambda due to layer size limitations.

2.3.4

Added
- [Core] Allow to execute a Class as lithops function
- [CE] Allow to to run code engine without kubecfg file
- [CE] Allow private container registries
- [k8s] Allow private container registries
- [knative] Allow private container registries
- [Localhost] Allow the partitioner to process local files
- [joblib] Added `joblib` entry in extras_require for joblib backend dependencies

Changed
- [CE] CPU and memory values must match predefined flavors
- [multiprocessing] Improved nanomsg Pipe implementation
- [joblib] Optimized joblib backend (concurrent args data upload/download)

Fixed
- [Core] Fixed module analyzer
- [Core] Clear only present jobs instead of all after wait() or get_result()
- [multiprocessing] Fix put/get slice to/from mp.Array or mp.RawArray

2.3.3

Fixed
- [Core] Allow to execute class methods as lithops function

Page 5 of 15

© 2024 Safety CLI Cybersecurity Inc. All Rights Reserved.