Apstra ConnectorOps RNIC कॉन्फ़िगरेटर
मार्गदर्शक

Apstra ConnectorOps RNIC कॉन्फ़िगरेटर

प्रकाशित
2025-12-09

जुनिपर नेटवर्क्स, इंक.
1133 इनोवेशन वे
सनीवेल, कैलिफोर्निया 94089
यूएसए
408-745-2000
www.juniper.net

जुनिपर नेटवर्क, जुनिपर नेटवर्क लोगो, जुनिपर और जूनो संयुक्त राज्य अमेरिका और अन्य देशों में जुनिपर नेटवर्क, इंक. के पंजीकृत ट्रेडमार्क हैं। अन्य सभी ट्रेडमार्क, सेवा चिह्न, पंजीकृत चिह्न, या पंजीकृत सेवा चिह्न उनके संबंधित स्वामियों की संपत्ति हैं।
जुनिपर नेटवर्क इस दस्तावेज़ में किसी भी अशुद्धि के लिए कोई जिम्मेदारी नहीं लेता है। जुनिपर नेटवर्क बिना किसी सूचना के इस प्रकाशन को बदलने, संशोधित करने, स्थानांतरित करने या अन्यथा संशोधित करने का अधिकार सुरक्षित रखता है।
Apstra ConnectorOps RNIC Configurator Guide
कॉपीराइट © 2025 जुनिपर नेटवर्क, इंक। सभी अधिकार सुरक्षित।
इस दस्तावेज़ की जानकारी शीर्षक पृष्ठ पर दिनांक के अनुसार वर्तमान है।

वर्ष 2000 नोटिस
जुनिपर नेटवर्क हार्डवेयर और सॉफ्टवेयर उत्पाद वर्ष 2000 के अनुरूप हैं। जूनोस ओएस की वर्ष 2038 तक कोई ज्ञात समय-संबंधी सीमाएं नहीं हैं। हालांकि, एनटीपी एप्लिकेशन को वर्ष 2036 में कुछ कठिनाई के लिए जाना जाता है।

अंत उपयोगकर्ता लाइसेंस समझौता
जुनिपर नेटवर्क उत्पाद जो इस तकनीकी दस्तावेज का विषय है, जुनिपर नेटवर्क सॉफ्टवेयर से बना है (या उपयोग के लिए अभिप्रेत है)। इस तरह के सॉफ़्टवेयर का उपयोग अंतिम उपयोगकर्ता लाइसेंस अनुबंध ("ईयूएलए") के नियमों और शर्तों के अधीन है, जो यहां पोस्ट किया गया है https://support.juniper.net/support/eula/ऐसे सॉफ़्टवेयर को डाउनलोड, इंस्टॉल या उपयोग करके, आप उस EULA के नियमों और शर्तों से सहमत होते हैं।

इस गाइड के बारे में

This guide walks you through the setup and deployment of Apstra ConnectorOps RNIC Configurator.
You define QoS and DCQCN settings in Apstra, and ConnectorOps RNIC reads those settings to configure your GPU server RNICs accordingly. The service then generates and pushes vendor-specific configuration scripts to your GPU server RNICs over SSH. ConnectorOps RNIC runs as a service on a VM or host with access to both Apstra and the GPU servers. It continuously monitors selected blueprints for configuration changes and ensures that RDMA traffic is always managed according to the latest Apstra intent.

1 CHAPTER: Introduction

परिचय

सारांश
ConnectorOps RDMA Network Interface Card (RNIC) Configurator automates host-side RNIC configuration for GPU servers in your fabric.

ConnectorOps RDMA Network Interface Card (RNIC) Configurator is a service designed to automate host-side RDMA NIC configuration for GPU servers in your fabric. The service continuously monitors Apstra blueprints for changes in topology and QoS configuration, such as updates to the Data Center Quantized Congestion Notification (DCQCN) Configlets. If an update event occurs, the service automatically generates scripts that are delivered to GPU servers based on the vendor. This keeps the network interface card Differentiated Services Code Point (DSCP) and Priority-based Flow Control (PFC) settings aligned with the intended fabric policy.

संबंधित दस्तावेज़ीकरण
Apstra ConnectorOps Setup Guide

2 CHAPTER: Prerequisites

आवश्यक शर्तें
Blueprint Prerequisites

• GPU servers connected as “generic systems” in the blueprint
• All interface cabling is mapped in the blueprint
• Configlets assigned to all leaf switches with connected GPU hosts

Apstra ConnectorOps RNIC Configurator Prerequisites

• DCQCN Configlets created in Apstra
• Property Set imported with GPU server SSH credentials
• ConnectorOps RNIC host can access Apstra and GPU servers
• SSH is enabled on all GPU servers
• configure_nic flag set to “enable” for RNIC configuration automation

3 CHAPTER: Deploy ConnectorOps RNIC Configurator and Verify Connectivity

Deploy ConnectorOps RNIC Configurator and Verify कनेक्टिविटी

सारांश
Use the following steps to deploy ConnectorOps RNIC Configurator and Verify Connectivity

1. Create a directory for the ConnectorOps image.
   mkdir ~/connectorops-rnic
2. Download the latest Apstra ConnectorOps RNIC image onto the VM/host/server.
   Use the following download link under Application Tools: https://support.juniper.net/support/downloads/?p=apstra.
3. Install Docker and Docker Compose.
4. Load the Docker image.
   root@test-vm:~/connectorops-rnic$ docker load -i <connectorops-rnic-image>
5. Create the docker-compose.yml file.
   निम्नलिखित एक पूर्व हैample Docker Compose. yml file.
   सेवाएं:
   connector-ops-rnic:
   image: <image-name>:<tag>
   container_name: connector-ops-rnic
   पर्यावरण:
   – AOS_URL=<aos-url>
   – AOS_USER=<aos-username>
   – AOS_PASSWORD=<aos-password>
   – CONFIGURE_NIC=enable
   बंदरगाह:
   – “<host-port>:8080”
   मात्राएँ:
   – <path-to-host-directory>:/app/data
   restart: always
6. Start ConnectorOps RNIC Configurator.
   root@test-vm:~/connectorops-rnic$ docker-compose up -d
   The service begins caching blueprint data and monitoring for configuration changes.
7. Access the debugging dashboard to verify the service is active.
   एक खोलो web browser and navigate to http://<host-ip>:<8080> (the same port in your docker-compose.yml) to view the dashboard tool.
   तुम्हें देखना चाहिए:
   • Active blueprints being monitored
   • Cached config-context data
   • Last poll timestamps
   • Any detected configuration changes

4 CHAPTER: Create and Import a Global Property Set

To Create and Import a Global Property Set

सारांश
You must define a Property Set with input information for ConnectorOps to automate configuration generation to the RNICs. ConnectorOps uses this Property Set as the data source for configuration rendering and to establish connectivity between devices.
You must configure a Property Set with the “connectorops_config” prefix. The property set must include the following:

• gpu_host credentials: username/password to access the GPU host or hosts.
• gpu_details: username/password to access the GPUs.

निम्नलिखित एक पूर्व हैample Property Set. Note that the jump_host details are only required if your topology contains a jump host.

Name connectorops_config
मान
{
“jump_host”: “<jump_host_ip>”,
“jump_host_user”: “<username>”,
“jump_host_password”: “<password>”,
“gpu_hosts_user”: “<username>”,
“gpu_hosts_password”: “<password>”,
“gpu_details”: {
“H100-01”: {
“user”: “<username>”,
“password”: “<password>”
},
“H100-0[2-4]”: {
“user”: “<username>”,
“password”: “<password>”
}
}
}

1. Navigate to Design > Property Sets > Create Property Set.
   The Create Property Set window displays.
2. Enter a Name and define a Property Set configuration in Values.
   Make sure to follow the proper naming convention.
3. बनाएँ पर क्लिक करें.
4. From within your Blueprint, navigate to Staged > Catalog > Property Sets > Import Property Set.
   The Import Property Set from Global Catalog window displays.
5. Select the Property Set you created from the dropdown, and click Import Property Set.
   This triggers an event, and ConnectorOps uses the Property Set as input to generate the configuration.

5 CHAPTER: Create and Import a DCQCN Configlet

Create and Import a DCQCN Configlet
For instructions about how to create and import a DCQCN Configlet into your Apstra blueprint, see Enable DCQCN in your AI Fabric.

For more information about DCQCN, see Data Center Quantized Congestion Notification (DCQCN).
निम्नलिखित एक पूर्व हैample DCQCN Configlet:

class-of-service {
classifiers {
dscp mydscp {
forwarding-class CNP {
loss-priority low code-points 110000;
}
forwarding-class NO-LOSS {
loss-priority low code-points 011010;
}
}
}
drop-profileएस {
dp1 {
interpolate {
fill-level [ 55 90 ];
drop-probability [ 0 100 ];
}
}
}
shared-buffer {
ingress {
buffer-partition lossless {
percent 66;
dynamic-threshold 10;
}
buffer-partition lossless-headroom {
percent 24;
}
buffer-partition lossy {
percent 10;
}
}
egress {
buffer-partition lossless {
percent 66;
}
buffer-partition lossy {
percent 10;
}
}
}
dedicated-buffer {
ingress {
percent 15;
}
egress {
percent 30;
}
}
forwarding-classes {
class CNP queue-num 3;
class NO-LOSS queue-num 4 no-loss pfc-priority 3;
}
congestion-notification-profile {
cnp {
pfc-watchdog;
input {
dscp {
code-point 011010 {
pfc;
}
}
}
output {
ieee-802.1 {
code-point 011 {
flow-control-queue 4;
}
}
}
}
}
इंटरफ़ेस {
et-* {
congestion-notification-profile cnp;
scheduler-map sm1;
unit * {
classifiers {
dscp mydscp;
}
}
}
}
scheduler-maps {
sm1 {
forwarding-class CNP scheduler s2-cnp;
forwarding-class NO-LOSS scheduler s1;
}
}
schedulers {
s1 {
drop-profile-map loss-priority any protocol any drop-profile dp1;
explicit-congestion-notification;
}
s2-cnp {
transmit-rate percent 5;
priority strict-high;
}
}
}

After you import and commit, Apstra includes the DCQCN policies in the config-context for the relevant switches in your fabric. ConnectorOps RNIC Configurator automatically reads these policies and enforces them on all GPU server RNICs.

6 CHAPTER: Troubleshooting

समस्या निवारण
Check the following to help with troubleshooting:

• Ensure that the GPU servers are connected as “generic systems” in the blueprint.
• Verify that all user-inputted Property Sets follow the correct naming conventions and are properly imported into the relevant blueprints.
• Confirm that both Apstra and the GPU servers are reachable from the VM/host/server where ConnectorOps is installed.
• Check that the ConnectorOps RNIC Configurator service is up and running. Review logs for any errors using:

docker logs <container_name_or_id>

To access the debugging tool dashboard, visit http://<host-ip>:<8080>.
You can use the debugging dashboard tool to troubleshoot by checking which blueprints are being polled, when data was last updated, and whether any errors occurred while processing Apstra configuration changes.

दस्तावेज़ / संसाधन

Juniper NETWORKS Apstra ConnectorOps RNIC Configurator [पीडीएफ] उपयोगकर्ता गाइड Apstra ConnectorOps RNIC Configurator, ConnectorOps RNIC Configurator, RNIC Configurator

संदर्भ

• उपयोगकर्ता पुस्तिका