1308979 - Specific worker for DeepSpeech tensorflow builds

Status: RESOLVED FIXED

(Taskcluster :: Operations and Service Requests, task)

Description

[:gerard-majax]

DeepSpeech has tensorflow builds running on tc with the github integration. However, sharing the workers when triggering cpu and gpu builds for tensorflow results in OOM. This should be avoided :)

Comment 1

[Dustin J. Mitchell [:dustin] (he/him)]

I think we'll use `deepspeech-worker` as the workerType name.  Here's the beginning of the workerType definition:

{
  "workerType": "deepspeech-worker",
  "minCapacity": 1,
  "maxCapacity": 20,
  "scalingRatio": 0,
  "minPrice": 0,
  "maxPrice": 0.2,
  "canUseOndemand": false,
  "canUseSpot": true,
  "instanceTypes": [
    {
      "instanceType": "r3.xlarge",
      "capacity": 1,
      "utility": 1,
      "secrets": {},
      "scopes": [],
      "userData": {
        "isolatedContainers": true
      },
      "launchSpec": {}
    },
    {
      "instanceType": "m3.xlarge",
      "capacity": 1,
      "utility": 1,
      "secrets": {},
      "scopes": [],
      "userData": {
        "isolatedContainers": true
      },
      "launchSpec": {}
    },
    {
      "instanceType": "c3.xlarge",
      "capacity": 1,
      "utility": 1,
      "secrets": {},
      "scopes": [],
      "userData": {
        "isolatedContainers": true
      },
      "launchSpec": {}
    }
  ],