docs/developers/provider/compute-endpoints.md

8.1 KiB

Compute Endpoints

All compute endpoints respond with an Array of status objects, each object describing a compute job info.

Each status object will contain:

    owner:The owner of this compute job
    documentId: String object containing document id (e.g. a DID)
    jobId: String object containing workflowId
    dateCreated: Unix timestamp of job creation
    dateFinished: Unix timestamp when job finished (null if job not finished)
    status:  Int, see below for list
    statusText: String, see below
    algorithmLogUrl: URL to get the algo log (for user)
    resultsUrls: Array of URLs for algo outputs
    resultsDid: If published, the DID

Status description (statusText): (see Operator-Service for full status list)

status Description
1 Warming up
10 Job started
20 Configuring volumes
30 Provisioning success
31 Data provisioning failed
32 Algorithm provisioning failed
40 Running algorith
50 Filtering results
60 Publishing results
70 Job completed

Create or restart compute job

Endpoint: POST /api/services/compute

Start a new job

Parameters

{% code overflow="wrap" %}

    signature: String object containg user signature (signed message) (required)
    consumerAddress: String object containing consumer's ethereum address (required)
    nonce: Integer, Nonce (required)
    environment: String representing a compute environment offered by the provider
    dataset: Json object containing dataset information
        dataset.documentId: String, object containing document id (e.g. a DID) (required)
        dataset.serviceId: String, ID of the service the datatoken is attached to (required)
        dataset.transferTxId: Hex string, the id of on-chain transaction for approval of datatokens transfer
            given to the provider's account (required)
        dataset.userdata: Json, user-defined parameters passed to the dataset service (optional)
    algorithm: Json object, containing algorithm information
        algorithm.documentId: Hex string, the did of the algorithm to be executed (optional)
        algorithm.meta: Json object, defines the algorithm attributes and url or raw code (optional)
        algorithm.serviceId: String, ID of the service to use to process the algorithm (optional)
        algorithm.transferTxId: Hex string, the id of on-chain transaction of the order to use the algorithm (optional)
        algorithm.userdata: Json, user-defined parameters passed to the algorithm running service (optional)
        algorithm.algocustomdata: Json object, algorithm custom parameters (optional)
    additionalDatasets: Json object containing a list of dataset objects (optional)

    One of `algorithm.documentId` or `algorithm.meta` is required, `algorithm.meta` takes precedence

{% endcode %}

Returns: Array of status objects as described above, in this case the array will have only one object

Example:

POST /api/compute
payload:
{
    "signature": "0x00110011",
    "consumerAddress": "0x123abc",
    "nonce": 1,
    "environment": "env",
    "dataset": {
        "documentId": "did:op:2222...",
        "serviceId": "compute",
        "transferTxId": "0x0232123..."
    }
}

Response:

[
    {
      "jobId": "0x1111:001",
      "status": 1,
      "statusText": "Job started",
      ...
    }
]

Status and Result

GET /api/services/compute

Get all jobs and corresponding stats

Parameters

{% code overflow="wrap" %}

    signature: String object containg user signature (signed message)
    documentId: String object containing document did  (optional)
    jobId: String object containing workflowID (optional)
    consumerAddress: String object containing consumer's address (optional)

    At least one parameter from documentId, jobId and owner is required (can be any of them)

{% endcode %}

Returns

Array of status objects as described above

Example:

GET /api/services/compute?signature=0x00110011&documentId=did:op:1111&jobId=012023

Response:

[
  {
    "owner": "0x1111",
    "documentId": "did:op:2222",
    "jobId": "3333",
    "dateCreated": "2020-10-01T01:00:00Z",
    "dateFinished": "2020-10-01T01:00:00Z",
    "status": 5,
    "statusText": "Job finished",
    "algorithmLogUrl": "http://example.net/logs/algo.log",
    "resultsUrls": [
      "http://example.net/logs/output/0",
      "http://example.net/logs/output/1"
    ],
    "resultsDid": "did:op:87bdaabb33354d2eb014af5091c604fb4b0f67dc6cca4d18a96547bffdc27bcf"
  },
  {
    "owner": "0x1111",
    "documentId": "did:op:2222",
    "jobId": "3334",
    "dateCreated": "2020-10-01T01:00:00Z",
    "dateFinished": "2020-10-01T01:00:00Z",
    "status": 5,
    "statusText": "Job finished",
    "algorithmLogUrl": "http://example.net/logs2/algo.log",
    "resultsUrls": [
      "http://example.net/logs2/output/0",
      "http://example.net/logs2/output/1"
    ],
    "resultsDid": ""
  }
]

GET /api/services/computeResult

Allows download of asset data file.

Parameters

    jobId: String object containing workflowId (optional)
    index: Integer, index of the result to download (optional)
    consumerAddress: String object containing consumer's address (optional)
    nonce: Integer, Nonce (required)
    signature: String object containg user signature (signed message)

Returns: Bytes string containing the compute result.

Example:

{% code overflow="wrap" %}

GET /api/services/computeResult?index=0&consumerAddress=0xA78deb2Fa79463945C247991075E2a0e98Ba7A09&jobId=4d32947065bb46c8b87c1f7adfb7ed8b&nonce=1644317370

{% endcode %}

Response:

b'{"result": "0x0000000000000000000000000000000000000000000000000000000000000001"}'

Stop

PUT /api/services/compute

Stop a running compute job.

Parameters

{% code overflow="wrap" %}

    signature: String object containg user signature (signed message)
    documentId: String object containing document did (optional)
    jobId: String object containing workflowID (optional)
    consumerAddress: String object containing consumer's address (optional)

    At least one parameter from documentId,jobId and owner is required (can be any of them)

{% endcode %}

Returns

Array of status objects as described above

Example:

PUT /api/services/compute?signature=0x00110011&documentId=did:op:1111&jobId=012023

Response:

[
    {
      ...,
      "status": 7,
      "statusText": "Job stopped",
      ...
    }
]

Delete

DELETE /api/services/compute

Delete a compute job and all resources associated with the job. If job is running it will be stopped first.

Parameters

    signature: String object containg user signature (signed message)
    documentId: String object containing document did (optional)
    jobId: String object containing workflowId (optional)
    consumerAddress: String object containing consumer's address (optional)

    At least one parameter from documentId, jobId is required (can be any of them)
    in addition to consumerAddress and signature

Returns

Array of status objects as described above

Example:

DELETE /api/services/compute?signature=0x00110011&documentId=did:op:1111&jobId=012023

Response:

[
    {
      ...,
      "status": 8,
      "statusText": "Job deleted successfully",
      ...
    }
]

GET /api/services/computeEnvironments

Allows download of asset data file.

Parameters

{% code overflow="wrap" %}

chainID: Int object representing the chain ID that the Provider is connected to (mandatory)

{% endcode %}

Returns: List of compute environments.

Example:

GET /api/services/computeEnvironments?chainId=8996

Response:

[
    {
        "cpuType":"AMD Ryzen 7 5800X 8-Core Processor",
        "currentJobs":0,
        "desc":"This is a mocked environment",
        "diskGB":2,
        "gpuType":"AMD RX570",
        "id":"ocean-compute",
        "maxJobs":10,
        "nCPU":2,
        "nGPU":0,
        "priceMin":2.3,
        "ramGB":1
    },
    ...
]