Slurm versions 14.11.4 and 15.08.0-pre2 are now available from
http://www.schedmd.com/#repos
Version 14.11.4 contains quite a few bug fixes generated over the past
five weeks. Several of these are related to job arrays, including one
that can cause the slurmctld daemon to abort. Version 15.08.0-pre2
represents the current state of Slurm development for the released
planned in August 2015 and is intended for development and test
purposes only. It includes some development work for burst buffers,
power management, and inter-cluster job dependencies. More details
about the changes are shown below.
* Changes in Slurm 14.11.4
==========================
-- Make sure assoc_mgr locks are initialized correctly.
-- Correct check of enforcement when filling in an association.
-- Make sacctmgr print out classification correctly for clusters.
-- Add array_task_str to the perlapi job info.
-- Fix for slurmctld abort with GRES types configured and no CPU binding.
-- Fix for GRES scheduling where count > 1 per topology type (or GRES types).
-- Make CR_ONE_TASK_PER_CORE work correctly with task/affinity.
-- job_submit/pbs - Fix possible deadlock.
-- job_submit/lua - Add "alloc_node" to job information available.
-- Fix memory leak in mysql accounting when usage rollup happens.
-- If users specify ALL together with other variables using the
--export sbatch/srun command line option, propagate the users'
environ to the execution side.
-- Fix job array scheduling anomaly that can stop scheduling of valid tasks.
-- Fix perl api tests for libslurmdb to work correctly.
-- Remove some misleading logs related to non-consumable GRES.
-- Allow --ignore-pbs to take effect when read as an #SBATCH argument.
-- Fix Slurmdb::clusters_get() in perl api from not returning information.
-- Fix TaskPluginParam=Cpusets from logging error message about not
being able
to remove cpuset dir which was already removed by the release_agent.
-- Fix sorting by time left in squeue.
-- Fix the file name substitution for job stderr when %A, %a %j and %u
are specified.
-- Remove minor warning when compiling slurmstepd.
-- Fix database resources so they can add new clusters to them after
they have
initially been added.
-- Use the slurm_getpwuid_r wrapper of getpwuid_r to handle possible
interrupts.
-- Correct the scontrol man page and command listing which node states can
be set by the command.
-- Stop sacct from printing non-existent stat information for
Front End systems.
-- Correct srun and acct_gather.conf man pages, mention Filesystem instead
of Lustre.
-- When a job using multiple partition starts send to slurmdbd only
the partition in which the job runs.
-- ALPS - Fix depth for MemoryAllocation in BASIL with CLE 5.2.3.
-- Fix assoc_mgr hash to deal with users that don't have a uid yet
when making
reservations.
-- When a job uses multiple partition set the environment variable
SLURM_JOB_PARTITION to be the one in which the job started.
-- Print spurious message about the absence of cgroup.conf at log
level debug2
instead of info.
-- Enable CUDA v7.0+ use with a Slurm configuration of TaskPlugin=task/cgroup
ConstrainDevices=yes (in cgroup.conf). With that configuration
CUDA_VISIBLE_DEVICES will start at 0 rather than the device number.
-- Fix job array logic that can cause slurmctld to abort.
-- Report job "shared" field properly in scontrol, squeue, and sview.
-- If a job is requeued because of RequeueExit or RequeueExitHold sent event
REQUEUED to slurmdbd.
-- Fix build if hwloc is in non-standard location.
-- Fix slurmctld job recovery logic which could cause the last task in a job
array to be lost.
-- Fix slurmctld initialization problem which could cause requeue of the last
task in a job array to fail if executed prior to the slurmctld loading
the maximum size of a job array into a variable in the job_mgr.c module.
-- Fix fatal in controller when deleting a user association of a user which
had been previously removed from the system.
-- MySQL - If a node state and reason are the same on a node state change
don't insert a new row in the event table.
-- Fix issue with "sreport cluster AccountUtilizationByUser" when using
PrivateData=users.
-- Fix perlapi tests for libslurm perl module.
* Changes in Slurm 15.08.0pre2
==============================
-- Add the environment variables SLURM_JOB_ACCOUNT, SLURM_JOB_QOS
and SLURM_JOB_RESERVATION in the batch/srun jobs.
-- Add sview burst buffer display.
-- Properly enforce partition Shared=YES option. Previously oversubscribing
resources required gang scheduling to be configured.
-- Enable per-partition gang scheduling resource resolution (e.g.
the partition
can have SelectTypeParameters=CR_CORE, while the global value is
CR_SOCKET).
-- Make it so a newer version of a slurmstepd can talk to an older srun.
allocation. Nodes could have been added while waiting for an allocation.
-- Expanded --cpu-freq parameters to include min-max:governor specifications.
--cpu-freq now supported on salloc and sbatch.
-- Add support for optimized job allocations with respect to SGI Hypercube
topology.
NOTE: Only supported with select/linear plugin.
NOTE: The program contribs/sgi/netloc_to_topology can be used to build
Slurm's topology.conf file.
-- Remove 64k validation of incoming RPC nodelist size. Validated at 64MB
when unpacking.
-- In slurmstepd() add the user primary group if it is not part of the
groups sent from the client.
-- Added BurstBuffer field to advanced reservations.
-- For advanced reservation, replace flag "License_only" with flag
"Any_Nodes".
It can be used to indicate the an advanced reservation resources (licenses
and/or burst buffers) can be used with any compute nodes.
-- Allow users to specify the srun --resv-ports as 0 in which case no ports
will be reserved. The default behaviour is to allocate one port per task.
-- Interpret a partition configuration of "Nodes=ALL" in slurm.conf as
including all nodes defined in the cluster.
-- Added new configuration parameters PowerParameters and PowerPlugin.
-- Added power management plugin infrastructure.
-- If job already exceeded one of its QOS/Accounting limits do not
return error if user modifies QOS unrelated job settings.
-- Added DebugFlags value of "Power".
-- When caching user ids of AllowGroups use both getgrnam_r() and
getgrent_r()
then remove eventual duplicate entries.
-- Remove rpm dependency between slurm-pam and slurm-devel.
-- Remove support for the XCPU (cluster management) package.
-- Add Slurmdb::jobs_get() interface to perl api.
-- Performance improvement when sending data from srun to stepds when
processing fencing.
-- Add the feature to specify arbitrary field separator when running
sacct -p or sacct -P. The command line option is --separator.
-- Introduce slurm.conf parameter to use Proportional Set Size (PSS) instead
of RSS to determinate the memory footprint of a job.
Add an slurm.conf option not to kill jobs that is over memory limit.
-- Add job submission command options: --sicp (available for inter-cluster
dependencies) and --power (specify power management options) to salloc,
sbatch, and srun commands.
-- Add DebugFlags option of SICP (inter-cluster option logging).
-- In order to support inter-cluster job dependencies, the MaxJobID
configuration parameter default value has been reduced from 4,294,901,760
to 2,147,418,112 and it's maximum value is now 2,147,463,647.
ANY JOBS WITH A JOB ID ABOVE 2,147,463,647 WILL BE PURGED WHEN SLURM IS
UPGRADED FROM AN OLDER VERSION!
-- Add QOS name to the output of a partition in squeue/scontrol/sview/smap.
--
Morris "Moe" Jette
CTO, SchedMD LLC
Commercial Slurm Development and Support
