Need 0.5.0 Release

[dongahn]

Keeping abreast of flux-framework/flux-core#1479 (comment)

[dongahn]

I believe #341 should the last merge before tagging v0.5.0.

[grondo]

First cut of NEWS entry gleaned from merged PRs since 0.4.0:
@dongahn, feel free to expand on some the bullet items here if you'd like to give a bigger picture of what changed, this can just be a starting point

flux-sched version 0.4.0 - 2018-05-11

Fixes

• c++ configuration changes to mirror flux-core (Bring in C++ configuration change to mirror flux-core #270)
• track upstream changes in flux-core APIs (track KVS API changes in flux-core #271, avoid deprecated KVS functions, minor simulator cleanup #272, track bulk message handler registration API change #278)
• portability fixes for Ubuntu 16.04 and 17.10 (solve portability issues for Ubuntu 16.04.4 LTS #298, solve portability issues for Ubuntu 17.10  #301)
• fixes for hwloc input xml (Fix for reading hwloc xml files generated from Sierra compute nodes #310)

Features

• add pending job cancellation support (sched: Add pending job cancellation support #291)
• add support for excluding resources for specific nodes (Add node exclusion support #305)
• support for GPU scheduling (GPU scheduling support #313)
• add new, more scalable resource assignment schema in kvs (R_lite) (Add R_lite support #321)
• include host names in R_lite resource assignment data (R_lite: Add the node key to support hostlist #324)
• optimize queue-depth=1 FCFS (sched: Add queue-depth=1 FCFS perf/scalability optimization #294)
• support for enhanced wreck submitted state event (Add support for the new wreck.state.submitted event #295)
• add resource-query utility (resource-query: Integrate resource-query into flux-sched #283)
• initial planner and priority plugin development (planner: initial planner work #274, planner: fix mintime resource tree corruption bug #281, Priority plugin work-in-progress #290)
• support for new completing state for wreck jobs (sched: Add completing state support #341)

Cleanup

• testsuite enhancements (sharness: update flux-sharness.sh script #316, experimental: add valgrind test to flux-sched, fix some leaks #328, Add valgrind wrap specification for dlclose in sched/plugin.c #331)
• refactoring and code cleanup (Refactor the new scheduling infrastructure code + cleanup #323, resource: Refactor policy code + other cleanup #325)
• drop old rank.N method for resource assignment (Drop rank.N support in favor of R_lite #339)

[springme]

Nice job, Flux team! Becky From: Mark Grondona <[email protected]> Reply-To: flux-framework/flux-sched <[email protected]> Date: Friday, May 11, 2018 at 3:21 PM To: flux-framework/flux-sched <[email protected]> Cc: Subscribed <[email protected]> Subject: Re: [flux-framework/flux-sched] Need 0.5.0 Release (#340) First cut of NEWS entry gleaned from merged PRs since 0.4.0: @dongahn<https://github.com/dongahn>, feel free to expand on some the bullet items here if you'd like to give a bigger picture of what changed, this can just be a starting point flux-sched version 0.4.0 - 2018-05-11 Fixes * c++ configuration changes to mirror flux-core (#270<#270>) * track upstream changes in flux-core APIs (#271<#271>, #272<#272>, #278<#278>) * portability fixes for Ubuntu 16.04 and 17.10 (#298<#298>, #301<#301>) * fixes for hwloc input xml (#310<#310>) Features * add pending job cancellation support (#291<#291>) * add support for excluding resources for specific nodes (#305<#305>) * support for GPU scheduling (#313<#313>) * add new, more scalable resource assignment schema in kvs (R_lite) (#321<#321>) * include host names in R_lite resource assignment data (#324<#324>) * optimize queue-depth=1 FCFS (#294<#294>) * support for enhanced wreck submitted state event (#295<#295>) * add resource-query utility (#283<#283>) * initial planner and priority plugin development (#274<#274>, #281<#281>, #290<#290>) * support for new completing state for wreck jobs (#341<#341>) Cleanup * testsuite enhancements (#316<#316>, #328<#328>, #331<#331>) * refactoring and code cleanup (#323<#323>, #325<#325>) * drop old rank.N method for resource assignment (#339<#339>) — You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub<#340 (comment)>, or mute the thread<https://github.com/notifications/unsubscribe-auth/AEdU0GvXTgSUsgM_5oKXQ93dSf9LxL8eks5txg7IgaJpZM4T6Ypp>.

[dongahn]

Looks great, @grondo. I just made minor edits below: 1) decouple priority plugin and planner works as the latter belongs to the new scheduling infrastructure piece; 2) collect all of the new scheduling infrastructure work PRs into planner and resource-query utility support.

Do you want me to post this as a PR to News.md?

flux-sched version 0.5.0 - 2018-05-11

Fixes

• c++ configuration changes to mirror flux-core (Bring in C++ configuration change to mirror flux-core #270)
• track upstream changes in flux-core APIs (track KVS API changes in flux-core #271, avoid deprecated KVS functions, minor simulator cleanup #272, track bulk message handler registration API change #278)
• portability fixes for Ubuntu 16.04 and 17.10 (solve portability issues for Ubuntu 16.04.4 LTS #298, solve portability issues for Ubuntu 17.10  #301)
• fixes for hwloc input xml (Fix for reading hwloc xml files generated from Sierra compute nodes #310)

Features

• add pending job cancellation support (sched: Add pending job cancellation support #291)
• add support for excluding resources for specific nodes (Add node exclusion support #305)
• support for GPU scheduling (GPU scheduling support #313)
• add new, more scalable resource assignment schema in kvs (R_lite) (Add R_lite support #321)
• include host names in R_lite resource assignment data (R_lite: Add the node key to support hostlist #324)
• optimize queue-depth=1 FCFS (sched: Add queue-depth=1 FCFS perf/scalability optimization #294)
• support for enhanced wreck submitted state event (Add support for the new wreck.state.submitted event #295)
• support for new completing state for wreck jobs (sched: Add completing state support #341)
• add planner and resource-query utility support (resource-query: Integrate resource-query into flux-sched #283, Refactor the new scheduling infrastructure code + cleanup #323, resource: Refactor policy code + other cleanup #325, planner: initial planner work #274, planner: fix mintime resource tree corruption bug #281)
• initial priority plugin development (Priority plugin work-in-progress #290)

Cleanup

• testsuite enhancements (sharness: update flux-sharness.sh script #316, experimental: add valgrind test to flux-sched, fix some leaks #328, Add valgrind wrap specification for dlclose in sched/plugin.c #331)
• drop old rank.N method for resource assignment (Drop rank.N support in favor of R_lite #339)

[dongahn]

@grondo: I just took liberty posting a PR (#342) since this is already late on Friday.