Merge branch 'release/1.4.0' into production

ULHPC · Mar 30, 2021 · 20cf269 · 20cf269
2 parents 97cbbd6 + 4b7a6c8
commit 20cf269
Show file tree

Hide file tree

Showing 12 changed files with 503 additions and 58 deletions.
diff --git a/README.md b/README.md
@@ -6,7 +6,7 @@
 
 Configure and manage [Slurm](https://slurm.schedmd.com/): A Highly Scalable Resource Manager
 
-      Copyright (c) 2017-2020 UL HPC Team <hpc-sysadmins@uni.lu>
+      Copyright (c) 2017-2021 UL HPC Team <hpc-sysadmins@uni.lu>
       .             see also http://hpc.uni.lu
 
 ## Overview

diff --git a/docs/overview.md b/docs/overview.md
diff --git a/docs/overview.md b/docs/overview.md
diff --git a/manifests/init.pp b/manifests/init.pp
@@ -135,8 +135,6 @@
 # @param enforcepartlimits        [String]      Default: 'ALL'
 # @param epilog                   [String]      Default: ''
 # @param epilogslurmctld          [String]      Default: ''
-# @param fastschedule             [Integer]     Default: 1
-#           Elligible values in [ 0, 1, 2]
 # @param grestypes                [Array]       Default: []
 #           list of generic resources to be managed
 # @param healthcheckinterval      [Integer]     Default: 30
@@ -280,9 +278,9 @@
 # @param switchtype               [String ]     Default: 'none'
 #           Elligible values in ['nrt', 'none']
 # @param taskepilog               [String ]     Default: ''
-# @param taskplugin               [String ]     Default: 'cgroup'
+# @param taskplugin               [String or Array]     Default: ['affinity','cgroup']
 #           Elligible values in ['affinity', 'cgroup','none']
-# @param taskpluginparams         [Array  ]     Default: ['cpusets']
+# @param taskpluginparams         [Array  ]     Default: []
 # @param taskprolog               [String ]     Default: ''
 # @param tmpfs                    [String ]     Default: '/tmp'
 # @param waittime                 [Integer]     Default: 0
@@ -374,7 +372,7 @@
 #            lower bound (in MB) on the memory limits defined by AllowedKmemSpace.
 # @param cgroup_minramspace        [String]     Default: '30M'
 #            lower bound (in MB) on the memory limits defined by AllowedRAMSpace & AllowedSwapSpace.
-# @param cgroup_taskaffinity       [Boolean]    Default: true
+# @param cgroup_taskaffinity       [Boolean]    Default: false
 #            This feature requires the Portable Hardware Locality (hwloc) library
 #
 ############################                      ####################################
@@ -558,7 +556,6 @@
   String  $enforcepartlimits              = $slurm::params::enforcepartlimits,
   String  $epilog                         = $slurm::params::epilog,
   String  $epilogslurmctld                = $slurm::params::epilogslurmctld,
-  Integer $fastschedule                   = $slurm::params::fastschedule,
   Integer $getenvtimeout                  = $slurm::params::getenvtimeout,
   Array   $grestypes                      = $slurm::params::grestypes,
   Integer $healthcheckinterval            = $slurm::params::healthcheckinterval,
@@ -651,7 +648,7 @@
   String  $srunprolog                     = $slurm::params::srunprolog,
   String  $switchtype                     = $slurm::params::switchtype,
   String  $taskepilog                     = $slurm::params::taskepilog,
-  String  $taskplugin                     = $slurm::params::taskplugin,
+  $taskplugin                             = $slurm::params::taskplugin,
   Array   $taskpluginparams               = $slurm::params::taskpluginparams,
   String  $taskprolog                     = $slurm::params::taskprolog,
   String  $tmpfs                          = $slurm::params::tmpfs,

diff --git a/manifests/pam.pp b/manifests/pam.pp
@@ -98,24 +98,36 @@
 
   # PAM limits
   # Ex: Update PAM MEMLOCK limits (required for MPI) + nproc
-  include ::ulimit
+  include ::limits
 
   if $ulimits_source != undef {
-    ulimit::rule { 'slurm':
+    file { "${limits::limits_dir}/50_slurm.conf":
       ensure => $ensure,
+      owner   => 'root',
+      group   => 'root',
+      mode    => '0644',
       source => $ulimits_source,
+
     }
+    # ulimit::rule { 'slurm':
+    #   ensure => $ensure,
+    #   source => $ulimits_source,
+    # }
   }
   else
   {
     $ulimits.each |String $item, $value| {
-      ulimit::rule { "slurm-${item}":
-        ensure        => $ensure,
-        ulimit_domain => '*',
-        ulimit_type   => [ 'soft', 'hard' ],
-        ulimit_item   => $item,
-        ulimit_value  => $value,
+      limits::limits { "*/${item}":
+        ensure => $ensure,
+        both   => $value
       }
+      # ulimit::rule { "slurm-${item}":
+      #   ensure        => $ensure,
+      #   ulimit_domain => '*',
+      #   ulimit_type   => [ 'soft', 'hard' ],
+      #   ulimit_item   => $item,
+      #   ulimit_value  => $value,
+      # }
     }
   }
 }
diff --git a/manifests/params.pp b/manifests/params.pp
@@ -38,6 +38,7 @@
       'libX11-devel',
       'libssh2-devel',
       'libevent-devel',
+      'python3'
     ],
     default => []
   }
@@ -170,7 +171,6 @@
   $enforcepartlimits       = 'ALL'
   $epilog                  = ''
   $epilogslurmctld         = ''
-  $fastschedule            = 1           # in [ 0, 1, 2]
   $grestypes               = []          # list of generic resources to be managed
   $healthcheckinterval     = 30
   $healthchecknodestate    = 'ANY'       # in ['ALLOC', 'ANY','CYCLE','IDLE','MIXED']
@@ -275,8 +275,8 @@
   $srunprolog              = ''
   $switchtype              = 'none' # in ['nrt', 'none']
   $taskepilog              = ''
-  $taskplugin              = 'cgroup' # in ['affinity', 'cgroup','none']
-  $taskpluginparams        = ['cpusets']
+  $taskplugin              = ['affinity','cgroup'] # in ['affinity', 'cgroup','none']
+  $taskpluginparams        = []
   $taskprolog              = ''
   $tmpfs                   = '/tmp'
   $tresbillingweights      = ''
@@ -359,7 +359,7 @@
   $cgroup_maxkmempercent            = 100   # upper bound in percent of total Kmem for a job.
   $cgroup_minkmemspace              = 30    # lower bound (in MB) on the memory limits defined by AllowedKmemSpace.
   $cgroup_minramspace               = 30    # lower bound (in MB) on the memory limits defined by AllowedRAMSpace & AllowedSwapSpace.
-  $cgroup_taskaffinity              = true  # This feature requires the Portable Hardware Locality (hwloc) library
+  $cgroup_taskaffinity              = false # if true, this feature requires the Portable Hardware Locality (hwloc) library
 
   ###
   ### Generic RESource management -- gres.conf
@@ -429,11 +429,11 @@
     default  => undef
   }
   # Which version of Slurm to grab and build
-  $version = '19.05.3-2'
+  $version = '20.11.3'
 
   ### SLURM Sources
   # Checksum for the slurm source archive (empty means no check will be done)
-  $src_checksum      = '6fe2c6196f089f6210d5ba79e99b0656f5a527b4'
+  $src_checksum      = 'dcf328865591b42b6c8f5586b3e396e6eb30dcd7'
   $src_checksum_type = 'sha1'
   # From where the Slurm sources can be downloaded
   $download_baseurl    = 'https://download.schedmd.com/slurm/'
@@ -507,9 +507,9 @@
   ############################
   # See https://pmix.org/code/getting-the-reference-implementation/
   $with_pmix              = false # Whether or not using PMIx
-  $pmix_version           = '3.1.4'
+  $pmix_version           = '3.2.3'
   # Checksum for the pmix source archive (empty means no check will be done)
-  $pmix_src_checksum      = '0f3f575e486d8492441c34276d1d56cbb48b4c37'
+  $pmix_src_checksum      = '97978abcd4da1b2a3d2bf2452247c4d47f8cc6a3'
   $pmix_src_checksum_type = 'sha1'
   # From where the Slurm sources can be downloaded
   $pmix_download_baseurl  = 'https://github.com/openpmix/openpmix/releases/download'
@@ -608,7 +608,7 @@
   ### SLURM DataBase Configuration (slurmdbd.conf)  ###
   #####################################################
   $dbd_configfile      = 'slurmdbd.conf'
-  $dbd_configfile_mode = '0400'
+  $dbd_configfile_mode = '0600'
   $archivedir          = '/tmp'
   $archiveevents       = false # When purging events also archive them?
   $archivejobs         = false # When purging jobs also archive them?

diff --git a/manifests/pmix.pp b/manifests/pmix.pp
@@ -26,21 +26,27 @@
 #          Top directory of the sources builds (i.e. RPMs, debs...)
 #          For instance, built RPMs will be placed under
 #          <builddir>/RPMS/${::architecture}
+# @param do_build                 [Boolean]     Default: true
+#          Do we perform the build of the OpenPMIx packages from sources or not?
+# @param do_package_install       [Boolean]     Default: true
+#          Do we perform the install of the OpenPMIx packages or not?
 #
-# @example install version 3.1.4 of PMIx
+# @example install version 3.2.3 of PMIx
 #
 #     slurm::pmix { '3.1.4':
 #        ensure => 'present',
 #        builddir => "/root/rpmbuild/",
-#
+#     }
 #
 class slurm::pmix(
-  String  $ensure            = $slurm::params::ensure,
-  String  $version           = $slurm::params::pmix_version,
-  String  $srcdir            = $slurm::params::srcdir,
-  String  $src_checksum      = $slurm::params::pmix_src_checksum,
-  String  $src_checksum_type = $slurm::params::pmix_src_checksum_type,
-  String  $builddir          = $slurm::params::builddir,
+  String  $ensure             = $slurm::params::ensure,
+  String  $version            = $slurm::params::pmix_version,
+  String  $srcdir             = $slurm::params::srcdir,
+  String  $src_checksum       = $slurm::params::pmix_src_checksum,
+  String  $src_checksum_type  = $slurm::params::pmix_src_checksum_type,
+  String  $builddir           = $slurm::params::builddir,
+  Boolean $do_build           = $slurm::params::do_build,
+  Boolean $do_package_install = $slurm::params::do_package_install,
 ) {
   include ::slurm::params
 
@@ -52,19 +58,23 @@
     checksum_type => $src_checksum_type,
   }
 
-  # Now build them
-  slurm::pmix::build { $version :
-    ensure  => $ensure,
-    srcdir  => $srcdir,
-    dir     => $builddir,
-    require => Slurm::Pmix::Download[$version],
+  if $do_build {
+    # Now build them
+    slurm::pmix::build { $version :
+      ensure  => $ensure,
+      srcdir  => $srcdir,
+      dir     => $builddir,
+      require => Slurm::Pmix::Download[$version],
+    }
   }
 
-  # And install it
-  slurm::pmix::install { $version :
-    ensure   => $ensure,
-    builddir => $builddir,
-    require  => Slurm::Pmix::Build[$version],
+  if $do_package_install {
+    # And install it
+    slurm::pmix::install { $version :
+      ensure   => $ensure,
+      builddir => $builddir,
+      require  => Slurm::Pmix::Build[$version],
+    }
   }
 
 }
diff --git a/manifests/slurmctld.pp b/manifests/slurmctld.pp
@@ -43,7 +43,7 @@
   Class['slurm::install'] -> Class['slurm::config']
 
   if $slurm::manage_firewall {
-    slurm::firewall { $slurm::slurmctldport:
+    slurm::firewall { "${slurm::slurmctldport}":
       ensure => $slurm::ensure,
     }
   }

diff --git a/manifests/slurmd.pp b/manifests/slurmd.pp
@@ -30,10 +30,10 @@
   Class['slurm::install'] -> Class['slurm::config']
 
   if $slurm::manage_firewall {
-    slurm::firewall { $slurm::slurmdport:
+    slurm::firewall { "${slurm::slurmdport}":
       ensure => $slurm::ensure,
     }
-    slurm::firewall { $slurm::srunportrange:
+    slurm::firewall { "${slurm::srunportrange}":
       ensure => $slurm::ensure,
     }
   }

diff --git a/manifests/slurmdbd.pp b/manifests/slurmdbd.pp
@@ -192,7 +192,7 @@
   Class['slurm::install'] -> Class['slurm::config']
 
   if $slurm::manage_firewall {
-    slurm::firewall { $dbdport:
+    slurm::firewall { "${dbdport}":
       ensure => $slurm::ensure,
     }
   }

diff --git a/metadata.json b/metadata.json
@@ -1,6 +1,6 @@
 {
   "name": "ULHPC-slurm",
-  "version": "1.3.0",
+  "version": "1.4.0",
   "author": "UL HPC Team",
   "summary": "Configure and manage Slurm: A Highly Scalable Resource Manager",
   "license": "Apache-2.0",
@@ -41,8 +41,8 @@
       "version_requirement": ">=1.2.2 <3.0.1"
     },
     {
-      "name": "svarrette-ulimit",
-      "version_requirement": ">=1.0.6 <2.0.0"
+      "name": "saz-limits",
+      "version_requirement": ">=3.0.4 <4.0.0"
     },
     {
       "name": "crayfishx-firewalld",
@@ -65,7 +65,8 @@
       "operatingsystem": "CentOS",
       "operatingsystemrelease": [
         "6",
-        "7"
+        "7",
+        "8"
       ]
     }
   ],

diff --git a/templates/slurm.conf.erb b/templates/slurm.conf.erb
@@ -269,7 +269,6 @@ SuspendTime=<%= scope['slurm::suspendtime'] %>
 AccountingStorageEnforce=<%= scope['slurm::accountingstorageenforce'].join(',') %>
 <% end -%>
 AccountingStorageHost=<%= scope['slurm::accountingstoragehost'] %>
-AccountingStorageLoc=slurm
 #AccountingStoragePass=
 #AccountingStoragePort=
 AccountingStorageType=accounting_storage/slurmdbd
@@ -334,7 +333,6 @@ StateSaveLocation=<%= scope['slurm::statesavelocation'] %>
 # - Backfill
 # - job priority
 ###
-FastSchedule=<%= scope['slurm::fastschedule'] %>
 #SchedulerTimeSlice=30
 SchedulerType=sched/<%= scope['slurm::schedulertype'] %>
 <% if scope['slurm::schedulerparameters'].empty? -%>
@@ -434,7 +432,11 @@ SelectTypeParameters=<%= scope['slurm::selecttype_params'].join(',') %>
 <% else -%>
 TaskEpilog=<%= scope['slurm::taskepilog'] %>
 <% end -%>
+<% if scope['slurm::taskplugin'].is_a?(Array) -%>
+TaskPlugin=task/<%= scope['slurm::taskplugin'].join(',task/') %>
+<% else -%>
 TaskPlugin=task/<%= scope['slurm::taskplugin'] %>
+<% end -%>
 <% if scope['slurm::taskpluginparams'].empty? -%>
 #TaskPluginParam=
 <% else -%>

diff --git a/tests/vagrant/puppet/hieradata/defaults.yaml b/tests/vagrant/puppet/hieradata/defaults.yaml
@@ -9,7 +9,6 @@ slurm::builddir: '/vagrant/tests/vagrant/rpmbuild'
 
 
 # SLURM general settings
-#slurm::version: '19.05.3-2'
 slurm::uid: 900
 slurm::gid: 900
 slurm::service_manage: true
@@ -71,9 +70,9 @@ slurm::prologflags:
   - 'X11'
 slurm::messagetimeout: 30
 slurm::resumetimeout: 600
-slurm::taskplugin: 'cgroup'
+slurm::taskplugin: ['affinity', 'cgroup']
 slurm::taskpluginparams :
-  - 'cpusets'
+  - 'none'
 
 ###############################
 ### cgroup.conf settings     ###