Error:Provisioning succeeded without instancePoolId or numOfWorker RRS feed

  • Question

  • I’ve repeatedly been getting this error:

    { "errorCode": "130", "message": "Internal Server Error:Provisioning succeeded without instancePoolId or numOfWorker for {0} ''", "failureType": "SystemError", "target": "XYZ", "details": [] }

    { "errorCode": "130", "message": "Internal Server Error:Provisioning succeeded without instancePoolId or numOfWorker for {0} ''", "failureType": "SystemError", "target": "PQR", "details": [] }

    XYZ and PQR are dummy names in place of the names of my pipeline. See the picture below. I don’t know if it’s some kind of issue with me having set a too low TTL on my runtime and it has timed out? But I would imagine that all that would mean, is that I had to incur AcquringCompute time again? At least, I can’t really glean anything from the error message, except that it provisioned something without getting a worker (at least, that’s what I understand from it).

    Please note that this is a triggered run, so I’m not debugging.

    If I click “Rerun from failed activity” it runs fine without the above error.

    Friday, February 14, 2020 11:44 AM

All replies

  • Hello tbjensen and thank you for bringing this issue to our attention.  Could you provide a few details for me?

    • Does this happen every trigger run, or only some times?
    • Could you share the TTL / integration runtime settings?
    • What region is the integration runtime using?
    • Is the issue still occurring?
    Friday, February 14, 2020 9:34 PM
  • Hi,

    1) It only happens sometimes, but running a complex pipeline of say 20 sub pipelines and 30 data flows, will result in having to rerun from failed activity around 5 to 6 times

    2+3) For running my entire pipeline I am using two different run times to try to cost optimize. One I use for sequential activities and one I use for parallel activities (such as flows being called from a pipeline in a foreach loop. The reason for this differentiation is that if I spawn say 15 data flows in parallel a cluster will be allocated for each of these, which will mean that I have to pay TTL for each of them as they finalize, whereas with the sequential path in the flow, I can use a TTL to make sure the cluster is ready for reuse for the next flow to minimize compute acquisition time.

    Parellel IR:

    Region: Auto Resolve

    Compute type: General purpose

    Core count: 8 (+ 8 Driver cores)

    Time to live: 0 minutes

    Sequential IR:

    Region: Auto Resolve

    Compute type: General purpose

    Core count: 8 (+ 8 Driver cores)

    Time to live: 15 minutes

    4) I haven't tested it today, but will later. Last week I was consistently getting the problem every day.

    Follow up: I still get the erros today (17th of February 2020), and the errors occur regardless me choosing the runtime, I've defined my self or the auto resolve runtime.

    • Edited by tbjensen Monday, February 17, 2020 12:46 PM
    Monday, February 17, 2020 9:50 AM
  • Thank you for the detailed response tbjensen.  I have reached out internally for more information.  I expect the experts will want to take a more detailed look from their end.  To facilitate this, could you please sent me an email at AzCommunity@Microsoft.com with the following information:

    Thread URL: (this page url)
    Subscription ID: (the subscription to which the factory belongs)
    Pipeline run IDs: (the IDs of several of the failed pipeline runs where this issue occurred. label whether this is the 'master' pipeline, or the 'subpipeline'.)

    Please let me know once you have sent the email.  Thank you for your patience.

    Tuesday, February 18, 2020 8:47 PM
  • Done :)
    Wednesday, February 19, 2020 9:01 AM