Change default value to TreeWidth

As a result of fanout investigation we decided to set the default TreeWidth value to 30 since it is the best balance between performances and robustness
aws · Apr 17, 2023 · 709565b · 709565b
1 parent 01d7d03
commit 709565b
Show file tree

Hide file tree

Showing 2 changed files with 2 additions and 0 deletions.
diff --git a/CHANGELOG.md b/CHANGELOG.md
@@ -29,6 +29,7 @@ This file is used to list changes made in each version of the AWS ParallelCluste
 - Set Slurm BatchStartTimeout to 3 minutes so to allow max 3 minutes Prolog execution during compute node registration.
 - Upgrade Slurm to version 23.02.1.
 - Upgrade munge to version 0.5.15.
+- Set Slurm default `TreeWidth` to 30. 
 
 **BUG FIXES**
 - Fix IP association on instances with multiple network cards.

diff --git a/cookbooks/aws-parallelcluster-slurm/templates/default/slurm/slurm.conf.erb b/cookbooks/aws-parallelcluster-slurm/templates/default/slurm/slurm.conf.erb
@@ -30,6 +30,7 @@ SlurmctldParameters=idle_on_node_suspend,power_save_min_interval=30,node_reg_mem
 TreeWidth=65533
 <% else -%>
 SlurmctldParameters=idle_on_node_suspend,power_save_min_interval=30,cloud_dns,node_reg_mem_percent=<%= node['cluster']['slurm_node_reg_mem_percent'] %>
+TreeWidth=30
 <% end -%>
 CommunicationParameters=NoAddrCache
 SuspendProgram=<%= node['cluster']['scripts_dir'] %>/slurm/slurm_suspend