Requesting a pod for cpu via python sdk seems not possible

[Lhemamou]

Describe the bug
When I want to create a only cpu pod, it's needed to choose a gpu_type_id but when i test all the gpu_type_id with gpu_count=0 , I have not access to any server.

To Reproduce
`list_gpus=[x['id'] for x in runpod.get_gpus()]

for gpu in list_gpus:
try:
runpod.create_pod(name="test_pod",image_name="docker/image",
gpu_count=0,gpu_type_id=gpu,
container_disk_in_gb=20,
min_vcpu_count=2,min_memory_in_gb=8,ports="8080/http;22/tcp")
break
except runpod.error.QueryError as e:
print(e)`

-> print statement :
There are no longer any instances available with the requested specifications. Please refresh and try again.

Expected behavior
A clear and concise description of what you expected to happen.
Expect to obtain a cpu only server

Screenshots
If applicable, add screenshots to help explain your problem.

Desktop (please complete the following information):

• OS: [e.g. iOS]
• Browser [e.g. chrome, safari]
• Version [e.g. 22]
  Ubuntu

Smartphone (please complete the following information):

• Device: [e.g. iPhone6]
• OS: [e.g. iOS8.1]
• Browser [e.g. stock browser, safari]
• Version [e.g. 22]

Additional context
Add any other context about the problem here.
python==3.10.12
runpod==1.6.2

[frankvp11]

Are there any updates on this issue and when it might be fixed? I too am running into a similar issue but then with the gpus itself (not trying to have only cpus)

[zaddy6]

I have the same issue and it pretty annoying

[matcha-mochi-rm]

I wrote to RunPod support about this, and received the following guidance:

• CPU pods are not supported yet via the Python SDK, but they're working on it.
• There is an undocumented method of achieving this via the graphQL API (which we can call e.g. using python's requests library). Here is what it looks like as a curl command:

curl --request POST \
  --header 'content-type: application/json' \
  --url 'https://api.runpod.io/graphql?api_key=${API_KEY}' \
  --data '{"query": "mutation { deployCpuPod(input: { instanceId: \"cpu3c-2-4\", cloudType: SECURE, containerDiskInGb: 5, deployCost: 0.06, dataCenterId: null, networkVolumeId: null, startJupyter: true, startSsh: true, templateId: \"runpod-ubuntu\", volumeKey: null, ports: \"22/tcp\" }) { id imageName env machineId machine { podHostId } } }"}'

I verified that this works, and it also works with custom template ID, and you can also pass a name for your pod as part of the input.

[AdonaiVera]

🚀 I’ve submitted a PR to add basic pod creation support in the Python SDK! It addresses this issue by implementing a minimal pod creation function, along with a unit test and README update.

PR: #403

Would love any feedback! Thanks! 🙌

[user41pp]

The undocumented deployCpuPod mutation works. But it has one issue: it does not support custom docker images (e.g. using imageName arg fails). Furthermore, creating CPU only templates seems to be impossible / unsupported; I could not do it (neither via web UI nor grqphQL).

However, dockerArgs work. We can use it to execute custom scripts on startup. Just be sure to add /start.sh at the end; otherwise, you won't be able to connect (if you use the default template). Here is an example query which works well for me:

    deployCpuPod(
      input: {
        instanceId: "cpu3c-2-4"
        cloudType: SECURE
        containerDiskInGb: 5
        deployCost: 0.06
        dataCenterId: null
        networkVolumeId: null
        startJupyter: true
        startSsh: true
        templateId: "runpod-ubuntu"
        dockerArgs: "bash -c 'mkdir -p /workspace && git clone https://${GITHUB_TOKEN}@github.com/${USER_NAME}/${PROJECT_NAME}.git /workspace/${PROJECT_NAME} && /workspace/${PROJECT_NAME}/entrypoint.sh && exec /start.sh'"
        volumeKey: null
        ports: "22/tcp"
        env: [{key: "GITHUB_TOKEN", value: "${GITHUB_TOKEN}"}]
      }
    ) {
      id
      imageName
      env
      machineId
      machine {
        podHostId
      }
    }
  }