2.5.0

Added

Changed

• ⚠️ Configuration Breaking data_operation and data_transfer settings are now configurable independently for each cluster.
• ⚠️ Configuration Breaking datatransfer_jobs_directives setting is now under data_operation.

Fixed

• ⚠️ API Breaking Fix transfer directives serialization, now properties names are properly camelcased (see issue: #162).
• Returns an error if the transfer_method chosen for large data transfer is not available.
• Documentation about streamer and wormhole data trasnfer methods
• buffer_limit for /filesystem/<system>/ops/* operation is now adapted to the value of settings.data_operation.max_ops_file_size (it was set to the value by default of 5MB)
• Updated Demo launcher configuration
• Fix error for PBS jobs when no nodes are assigned to it.
• Handle job arrays in PBS. Job IDs will be strings, and not integers anymore in the API responses.
• Remove hardcoded jfrog link from the wormhole download endpoint.
• Customizable Response's headers tracing log

2.4.2

Added

• File target check in S3 file transfer job for external file upload.

• cluster.scheduler.connection_mode setting to configure how the client connects to the scheduler backend (ssh, rest or hybrid)

• Get jobs now allows to specify the account parameter

Changed

• Large file download via s3 no loger appends uuid to file name.

Fixed

• File transfer examples with .NET

2.4.1

Added

• Auto-selection of expected signature algorithm when alg attribute is not set in IdP JWKs endpoint.
• Firecrest streamer now computes checksums to validate data transfer
• Firecrest streamer transfer end-point returns immediet errors if target paths have issues.

Changed

• Firecrest Streamer no longer allows to overwrite existing files.
• Get jobs shows jobs up to one week.
• Firecrest streamer improved error handling, including remote errors.
• SLURM RESTAPI now validates the username claim on the access token from the auth configuration.

Fixed

• Demo Launcher now is adapted correctly to the data_operation setup introduced in 2.4.0
• Old link to the installation documentation
• Timeout value for commands executed via filesystems/ops and status is now configured with command execution timeout setting
• Stdout, stderr path are now fully expanded
• probing configuration is optional now for the clusters settings

2.4.0

Added

• Compress and Extract end-points now support multiple compresion types (none, bz2, gzip, and xz).
• Support for Magic Wormhole data transfer method

Changed

• The System Name path parameter and the corresponding Cluster name configuration are case insensitive.

Fixed

• Fixed Slurm sacct integration and data parsing.

• Docker Compose startup: Added dependency for Slurm to wait for Keycloak health check before starting, preventing JWT certificate download failures.

• Upload and Download transfer endpoints now require to specify transfer directives

• Installation docs:

  • Helm charts: FirecREST settings are all included in values.yaml file
  • Changed documentation name from Deployment to Install

• Improved direct upload endpoints size limit checks

2.3.1

Added

Changed

• New data_operation setting to replace storage. The new setting decouples the max_ops_file_size parameter from the data_transfer settings that are now a child parameter of data_operation. Also, data_transfer can be of multiple types.
• Certificates debug information when SSH connection fails

Fixed

2.3.0

Added

• Add support for the OpenPBS scheduler.
• Add support for the DeiC ssh certificate authority.
• Allows to set the JWT claim that contains the username.

Changed

• /filesystem/cluster-slurm-ssh/ops/view endpoint now accepts size and offset parameters to read an arbitrary chunk of a file

Fixed

2.2.8

Added

• Refactor FastAPI models for Slurm, in order to make it easier to add new schedulers.
• Make cp recursive so that directories can also be copied and add the option to keep symbolic links.

Changed

Fixed

• Fixed Slurm timestamps parsing issues, timezone was not properly handled.

2.2.7

Added

• Query parameter allusers in GET /compute/jobs to show all visible jobs for the user in the scheduler
• Environment variable UVICORN_LOG_CONFIG to enable Uvicorn log configuration file path (analog to --log-config)

Changed

Fixed

• Show nodes from hidden partitions using SLURM CLI
• Fixed reservation start and end datatime parsing.
• Handles instances where no Job exit status is provided.
• Fixed unnecessary user keys retrieval with SSH connection pool.
• Fixed proper SSH process termination on timeout.
• UVICORN_LOG_CONFIG value on helm chart

2.2.6

Added

• account optional parameter to job submission request
• script_path optional parameter for submitting jobs from a remote file
• JupyterHub example
• Documentation for logging architecture
• Workflow orchestrator example
• UI browser app example
• POST and PUT bodies request examples
• Documentation and examples in C# .NET

Changed

• Documentation for logging architecture
• Images for documentation
• Description of API definition

Fixed

2.2.5

Added

• Log for request and command execution tracing

Changed

Fixed

• Fix health check for older versions of Slurm REST API (< v0.0.42)