HTTP Node --> RunPod API

Hey guys,

I designed a workflow that use HTTP nodes to create on runpod (via API) pods and NFS store. I created the necessary API commands via Create a new Pod - Runpod Documentation (also tested there and working) and use the CURL import to import the API comannds. I run then in different problems:

If i run the node the API responding:

  1. All parameters with a value NULL was not working ( i fixed that with removal of the parameters, and retested is against the API and is working)

  2. If i run the node in the workflow with the exact same command:

    500 - "{\“error\”:\“create pod: There are no instances currently available\”,\“status\”:500}\n

It looks like that the curl import changing something or the values datatypes are not correct but i have no idea. Any run into same problems?

Json BODY

{“allowedCudaVersions”: [“12.8”],“cloudType”: “SECURE”,“computeType”: “GPU”,“containerDiskInGb”: 50,“containerRegistryAuthId”: “”,“countryCodes”: [“”],“cpuFlavorIds”: [“cpu3c”],“cpuFlavorPriority”: “availability”,“dataCenterIds”: [“EU-SE-1”],“dataCenterPriority”: “availability”,“dockerEntrypoint”: ,“dockerStartCmd”: ,“env”: {“ENV_VAR”: “value”},“globalNetworking”: false,“gpuCount”: 1,“gpuTypeIds”: [“NVIDIA A40”],“gpuTypePriority”: “availability”,“imageName”: “runpod/pytorch:1.0.2-cu1281-torch280-ubuntu2404”,“interruptible”: false,“locked”: false,“minDiskBandwidthMBps”: 123,“minDownloadMbps”: 123,“minRAMPerGPU”: 8,“minUploadMbps”: 123,“minVCPUPerGPU”: 2,“name”: “vibevoice”,“networkVolumeId”: “{{ $json.id }}”,“ports”: [“8888/http,22/tcp”,“22/tcp”],“supportPublicIp”: true,“vcpuCount”: 2,“volumeInGb”: 20,“volumeMountPath”: “/workspace”}

Hello, first check quotes and, change it to " (just copy form runpod’s response). You may check the syntax at https://jsonformatter.org

Then you’ll find a syntax problem here:

Parse error on line 1:
..."dockerEntrypoint": ,"dockerStartCmd": ,
-----------------------^
Expecting 'STRING', 'NUMBER', 'NULL', 'TRUE', 'FALSE', '{', '[', got ','

This happens because dockerEntrypoint and dockerStartCmd need a list as an argument:

"dockerEntrypoint": [] ,"dockerStartCmd": []

You may find a similar request here, runpod-gitlab-ci/.gitlab-ci.yml at 5243ec4b2743320a4fb30787dcdfe6e23345a10d · rapekas/runpod-gitlab-ci · GitHub - probably your next steps are running vLLM and a model inside it. That code may help to achieve it. I am the author.