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.
0.9.3 - 2019-05-08
- Fixed bug relating to cluster comparison
0.9.2 - 2019-05-08
- When
squeue
returns '(null)' for an account the Slurm adapter will now convert that tonil
0.9.1 - 2019-05-07
- Added logic to
OodCore::Job::ArrayIds
to return an empty array when the array request is invalid
0.9.0 - 2019-05-04
- Job array support for LSF and PBSPro
- Slurm adapter uses
squeue
owner filter (-u
) forinfo_where_owner
- Grid Engine adapter now starts scripts in the current directory like all other adapters
- Fixed issue where Slurm comment field might break job info parsing
- Fixed possible crash when comparing two clusters if the id of one of the clusters is nil
- Fixed bug with the live system test that impacted non-Torque systems
- Fixed bug with Slurm adapter when submit time is not available
0.8.0 - 2019-01-29
- info_all_each and info_where_owner_each super class methods
- job array support for Torque, Slurm, and SGE (currently missing from LSF and PBSPro)
OodCore::Job::Status#precedence
for the ability to get an overall status for a group of jobs
- Fix SGE adapter to specify
-u '*'
when calling qstat to get all jobs
0.7.1 - 2019-01-11
- Fixed crash when libdrmaa is used to query for a job no longer in the queue
0.7.0 - 2018-12-26
- Addition of an optional live system test of a configurable job adapter
- Fix Torque adapter crash by fixing scope resolution on Attrl and Attropl
- Fix SGE adapter crash in
OodCore::Job::Adapters::Sge::Batch#get_info_enqueued_job
when libdrmma is not available (DRMMA constant not defined)
- Always set
SGE_ROOT
env var, for both SGE commands via popen and when using libdrmaa - Use libdrmaa only when libdrmaa is set in the cluster config
0.6.0 - 2018-12-19
- Added ability to override the default password length
- Merge the pbs-ruby gem removing that as a dependency, but adding FFI
- Added support for overriding resource manager client executables using
bin_overrides
in the cluster configs - Add support for the Grid Engine resource manager (tested on GE 6.2u5 and UGE 8.0.1)
- Fixed a bug in password creation where certain locales resulted in invalid passwords #91
0.5.1 - 2018-05-14
0.5.0 - 2018-04-30
- Added missing "Waiting" state to the Torque adapter as
:queued_held
.
- Changed the "Waiting" state in the PBSPro adapter to
:queued_held
.
0.4.0 - 2018-04-20
- Updated Torque adapter to take into account the new
Script#native
format allowing for arrays. #65
0.3.0 - 2018-04-05
- Basic multi-cluster support for LSF by specifying name of cluster for -m argument. #24
- Added
OodCore::Job::Script#shell_path
as an option to all adapters. #82 - Added
header
andfooter
options to a Batch Connect template. #64
- Replaced
Fixnum
code comments withInteger
. #67
0.2.1 - 2018-01-26
- Updated the date in the
LICENSE.txt
file.
- Fixed bug where LSF adapter would sometimes return
nil
when getting job info. #75 - Fixed list of allocated nodes for LSF adapter when single node is expanded for each core. #71
- Clean up children processes in forked Batch Connect main script before cleaning up batch script. #69
- Fix bug when detecting open ports using the bash helpers in the Batch Connect template. #70
0.2.0 - 2017-10-11
- Added Batch Connect helper function to wait for port to be used. #57
- Can include Batch Connect helper functions when writing to files or running remote code. #58
- The Batch Connect helper functions are now available to use in the forked Batch Connect main script. #59
- The
host
andport
environment variables are now available to use in the forked Batch Connect main script. #60
- Fixed a bug with the
nc
command used in the Batch Connect helper functions for CentOS 7. #55 - Fixed not correctly detecting open ports for specific ip address in Batch Connect helper functions. #56
- Fixed a bug when parsing nodes in the Slurm adapter. #54
0.1.1 - 2017-09-08
- fix crash when calling
Adapters::Lsf#info(id:)
with "invalid" id - optimize
Adapters::Lsf#info_where_owner
by usingbjobs -u $USER
when a single user is specified
0.1.0 - 2017-07-17
- Setting the host in a batch_connect batch script can now be directly
manipulated through the
set_host
initialization parameter. #42
0.0.5 - 2017-07-05
- Add wallclock time limit to
OodCore::Job::Info
object. - Add further support for the LSF adapter.
- Add a new Batch Connect template feature that builds batch scripts to launch web servers.
- Add support for the PBS Professional resource manager.
- Add method to filter list of batch jobs for a given owner or owners.
- Torque adapter provides nodes/procs info if available for non-running jobs.
- Slurm adapter provides node info if available for non-running jobs.
- Changed the
CHANGELOG.md
formatting.
- Remove deprecated tests for the Slurm adapter.
- Fix parsing bjobs output for LSF 9.1, which has extra SLOTS column.
0.0.4 - 2017-05-17
- By default all PBS jobs output stdout & stderr to output path unless an error path is specified (mimics behavior of Slurm and LSF)
- Remove
OodCore::Job::Script#min_phys_memory
due to lack of commonality across resource managers. - Remove
OodCore::Job::Script#join_files
due to lack of support in resource managers.
0.0.3 - 2017-04-28
- Provide support for Slurm conf file.
- Correct code documentation for
Script#min_phys_memory
. - Add fix for login feature being allowed on all clusters even if not defined.
0.0.2 - 2017-04-27
- Remove the
OodCore::Job::NodeRequest
object.
- Initial release!