Pegasus 5.1.2 Released

by on February 3, 2026 with No Comments
We are happy to announce the release of Pegasus 5.1.2. It is a minor release in the 5.1 branch. We invite our users to give it a try.

Pegasus 5.1.2 Release Image
Pegasus 5.1.2 Release Image

The release can be downloaded from: https://pegasus.isi.edu/downloads   

Highlights of the Release

  1. Move to Condor File IO for OSDF transfers.OSDF transfers are now always delagated to HTCondor to manage using HTCondor file IO, especially when turning on Bypass Input File Staging. This applicable both for condorio andnonsharedfs data configurations.More details can be found in the documentation.
  2. Support for Flux This release of Pegasus has support for running workflows on HPC resources managed by Flux resource manager. This support relies on changes to HTCondor which will be made available in an upcoming 25.7.0 release scheduled for March 2026.Details of mapping Pegasus resource profiles to flux parameters can be found here. This repository has useful scripts that a user can use to deploy pegasus in user mode on Flux Systems. #2143
  3. Modified Pegasus versioning scheme to use Semantic Versioning Scheme v2 #2126
  4. Support for Python 3.14

New Features and Improvements

  1. avoid parsing sub workflows into memory when parsing the top level workflow that includes them #2148
  2. flux support #2143
  3. Move OSDF transfers to be via condor file transfers in Pegasus Lite instead of relying on pegasus-transfer #2141
  4. pick user provided env script for PegasusLite from the site where the job runs #2136
  5. limit the number of pegasus-monitord launches in pegasus-dagman #2134
  6. Add support for Python 3.14 #2128
  7. Modify Pegasus versioning scheme to use Semantic Versioning Scheme v2 #2126
  8. enable condorio support for bosco/ssh style job submissions for remote HPC clusters #2121
  9. use job classad variables to shorten paths in transfer_input_files key in the job submit directories #2120
  10. Explore condor_dag_checker – something we want to use as part of planning? #2116
  11. update sqlite jar to latest stable 3.49.1.0 or higher #2109
  12. [PM-1833] pmc cpuinfo invalid detection #1946
  13. [PM-1098] encrypt the credentials when transferred with jobs #1212
  14. [PM-1092] Ask Condor team to propagate glite errors in gridmanager #1206

Bugs Fixed

  1. enforce maximum document parsing size for yaml docs to 2047 MB #2152
  2. expand wf_submit_dir when generating AWS Batch job descriptions #2151
  3. cpu atttributes are not included in the job composite event #2150
  4. ensure user JAVA_HEAPMAX and JAVA_HEAPMIN values are propagated to the planner invocations for sub workflows #2147
  5. bypass in condorio mode gets incorrectly triggered if a directory path specified for a local file #2142
  6. cleanup jobs running remotely in nonsharedfs get associated with a container #2137
  7. CLI tools pollute PYTHONPATH #2135
  8. condor quoting is not triggered for arguments for glite and ssh style jobs #2132
  9. monitord fails to parse job.out file if location record is malformed #2131
  10. pegasus-init remote cluster option creates incorrect paths for local site #2125
  11. worker package staging broken in sharedfs for create dir job (if set to run remotely) #2124
  12. monitord overwrites transfer_attempts records found in the job.out file #2123
  13. pegasus-analyzer –debug-job option broken #2122
  14. pegasus-mpi-cluster: cpuinfo validation fails on hybrid CPU architectures #2119

 Merged Pull Requests

  1. Bugfix: removes incorrect use of the ‘local’ keyword in flux_local_submit_attributes.sh #2145(ilumsden)
  2. Adds a “local_submit_attributes.sh” script for the Flux RJMS #2144 (ilumsden)
  3. Update pegasus-halt to fix POSIX compliance issue, bug on OSs where SH-> dash #2139(ahnitz)