NVIDIA NEO REST API Guide v2.7.20
v2.7.20

Performance

Description

Runs a performance test between all/random hosts in the fabric

Request URL

  • For the Random Candidate option:

    POST /neo/app/performance

  • For the All to All option:

    POST /neo/app/performance?all=true

test_type options

  • RoCE latency

  • RoCE bandwidth

  • RoCE message rate (telemetry only)

  • TCP/UDP latency

  • TCP/UDP bandwidth single-stream

  • TCP bandwidth multi-stream

  • Multicast bandwidth by message size

  • Multicast latency by message size

  • Multicast latency under load (telemetry only)

  • Ping (network health only)

Modes

  • Telemetry mode: 1_to_1

  • Network health mode: health

Request Data

See code directly below

Status Codes

Note:

Status code 409 CONFLICT means that a Job is already running and the system is without capability.

Status code 500 INTERNAL SERVER ERROR means that there is a bug in the system.

Status code 503 SERVICE UNAVAILABLE means that performance provider is down.

Copy
Copied!
            

{ "role":{ "servers":["10.209.36.108"], "clients":["10.209.36.113"] }, "type":[" RoCE latency"] , "mode": " 1_to_1 }

Warning

You must have PerfTest capability enabled on the systems

Request URL

POST /neo/app/performance/job_id/save

Request Data

{
"title" : "Performance Test"
"description" : ""
}

Request URL

GET /neo/app/perfomance/job_id

Response:

Copy
Copied!
            

{ “status” : “Done” “tests”:[ { “name”:” Ethernet TCP_RR”, “direction”:” unidir”, “message_size”:2, “stream”:1, “host_id”:” 10.209.36.108”, “result”: “1.43 [us]” , “expected”: “(35,40)” } ], "params": { "role":{ "servers":[ "10.209.36.108": “ports”:[“eth0”] ], "clients":[ "10.209.36.113": “ports”:[“eth1”] ] } “type”:” health” “mode”: 1_to_1 } }

© Copyright 2023, NVIDIA. Last updated on Nov 16, 2023.