From 544a89bbce051108ee83d158df7f9ba4842548df Mon Sep 17 00:00:00 2001 From: Matt Crees Date: Thu, 14 Aug 2025 14:33:00 +0000 Subject: [PATCH 1/3] Set 1500 MTU for slurm-staging-control network --- etc/openstack-config/openstack-config.yml | 1 + 1 file changed, 1 insertion(+) diff --git a/etc/openstack-config/openstack-config.yml b/etc/openstack-config/openstack-config.yml index 59fc042..90e62d3 100644 --- a/etc/openstack-config/openstack-config.yml +++ b/etc/openstack-config/openstack-config.yml @@ -337,6 +337,7 @@ openstack_network_slurm_staging_control: project: "{{ openstack_project_slurm_staging.name }}" shared: false external: false + mtu: 1500 port_security_enabled: false provider_network_type: "vlan" provider_physical_network: "physnet3" From d67e45d0801a39773d48c5554f28918886a9cfec Mon Sep 17 00:00:00 2001 From: Matt Crees Date: Thu, 14 Aug 2025 15:53:47 +0000 Subject: [PATCH 2/3] Add physnet3 network for slurm-production - Rename to indicate network dedicated to RDMA - RDMA network has port security disabled - Control network on physnet3, providing default gateway --- etc/openstack-config/openstack-config.yml | 54 ++++++++++++++++------- 1 file changed, 37 insertions(+), 17 deletions(-) diff --git a/etc/openstack-config/openstack-config.yml b/etc/openstack-config/openstack-config.yml index fbcf3ed..a99b6cd 100644 --- a/etc/openstack-config/openstack-config.yml +++ b/etc/openstack-config/openstack-config.yml @@ -160,9 +160,10 @@ openstack_networks: #- "{{ openstack_network_external_ceph }}" #- "{{ openstack_network_stackhpc }}" #- "{{ openstack_network_stackhpc_vlan }}" - - "{{ openstack_network_slurm_staging_comm }}" + - "{{ openstack_network_slurm_staging_rdma }}" - "{{ openstack_network_slurm_staging_control }}" - - "{{ openstack_network_slurm_production }}" + - "{{ openstack_network_slurm_production_rdma }}" + - "{{ openstack_network_slurm_production_control }}" openstack_networks_rbac: - "{{ openstack_rbac_external_ceph }}" @@ -313,8 +314,8 @@ secgroup_nvidia_dls: ################################################################################ # Networks for Slurm staging # -openstack_network_slurm_staging_comm: - name: "{{ openstack_project_slurm_staging.name }}-comm-net" +openstack_network_slurm_staging_rdma: + name: "{{ openstack_project_slurm_staging.name }}-rdma-net" project: "{{ openstack_project_slurm_staging.name }}" shared: false external: false @@ -322,10 +323,10 @@ openstack_network_slurm_staging_comm: provider_network_type: "vlan" provider_physical_network: "physnet2" subnets: - - "{{ openstack_subnet_slurm_staging_comm }}" + - "{{ openstack_subnet_slurm_staging_rdma }}" -openstack_subnet_slurm_staging_comm: - name: "{{ openstack_project_slurm_staging.name }}-comm-subnet" +openstack_subnet_slurm_staging_rdma: + name: "{{ openstack_project_slurm_staging.name }}-rdma-subnet" project: "{{ openstack_project_slurm_staging.name }}" cidr: "192.168.1.0/24" no_gateway_ip: true @@ -338,7 +339,6 @@ openstack_network_slurm_staging_control: shared: false external: false mtu: 1500 - port_security_enabled: false provider_network_type: "vlan" provider_physical_network: "physnet3" subnets: @@ -378,29 +378,49 @@ openstack_secgroup_slurm_staging_https: ################################################################################ # Networks for Slurm production # -openstack_network_slurm_production: - name: "{{ openstack_project_slurm_production.name }}" +openstack_network_slurm_production_rdma: + name: "{{ openstack_project_slurm_production.name }}-rdma-net" project: "{{ openstack_project_slurm_production.name }}" - shared: true + shared: false external: false + port_security_enabled: false provider_network_type: "vlan" provider_physical_network: "physnet2" subnets: - - "{{ openstack_subnet_slurm_production }}" + - "{{ openstack_subnet_slurm_production_rdma }}" -openstack_subnet_slurm_production: - name: "{{ openstack_project_slurm_production.name }}" +openstack_subnet_slurm_production_rdma: + name: "{{ openstack_project_slurm_production.name }}-rdma-subnet" project: "{{ openstack_project_slurm_production.name }}" cidr: "192.168.8.0/21" - gateway_ip: "192.168.8.1" + no_gateway_ip: true allocation_pool_start: "192.168.8.10" allocation_pool_end: "192.168.15.250" +# +openstack_network_slurm_production_control: + name: "{{ openstack_project_slurm_production.name }}-control-net" + project: "{{ openstack_project_slurm_production.name }}" + shared: true + external: false + mtu: 1500 + provider_network_type: "vlan" + provider_physical_network: "physnet3" + subnets: + - "{{ openstack_subnet_slurm_production_control }}" + +openstack_subnet_slurm_production_control: + name: "{{ openstack_project_slurm_production.name }}-control-subnet" + project: "{{ openstack_project_slurm_production.name }}" + cidr: "192.168.0.0/21" + gateway_ip: "192.168.0.1" + allocation_pool_start: "192.168.0.10" + allocation_pool_end: "192.168.7.250" openstack_router_slurm_production: - name: "{{ openstack_project_slurm_production.name }}" + name: "{{ openstack_project_slurm_production.name }}-router" project: "{{ openstack_project_slurm_production.name }}" interfaces: - - "{{ openstack_network_slurm_production.name }}" + - "{{ openstack_subnet_slurm_production_control.name }}" network: "{{ openstack_network_external_internet.name }}" openstack_secgroup_slurm_production_default: From b71d1d82392492cf05bdd556f923926db6456ee0 Mon Sep 17 00:00:00 2001 From: Stig Telfer Date: Fri, 12 Sep 2025 18:09:47 +0000 Subject: [PATCH 3/3] Feedback from Matt --- etc/openstack-config/openstack-config.yml | 8 ++++---- 1 file changed, 4 insertions(+), 4 deletions(-) diff --git a/etc/openstack-config/openstack-config.yml b/etc/openstack-config/openstack-config.yml index a99b6cd..8ffa568 100644 --- a/etc/openstack-config/openstack-config.yml +++ b/etc/openstack-config/openstack-config.yml @@ -347,10 +347,10 @@ openstack_network_slurm_staging_control: openstack_subnet_slurm_staging_control: name: "{{ openstack_project_slurm_staging.name }}-control-subnet" project: "{{ openstack_project_slurm_staging.name }}" - cidr: "192.168.0.0/24" - gateway_ip: "192.168.0.1" - allocation_pool_start: "192.168.0.10" - allocation_pool_end: "192.168.0.250" + cidr: "192.168.16.0/24" + gateway_ip: "192.168.16.1" + allocation_pool_start: "192.168.16.10" + allocation_pool_end: "192.168.16.250" openstack_router_slurm_staging: