Why are so many of my nodes unusable? RRS feed

  • Question

  • Hello,

    I do not have any special configurations of my VMs, no changes to firewall, ports or anything like that. I don't even have any application package. However right now I have put up 100 VMs, and out of them 70 are set as unusable... This is extremely frustrating, and there is not even very much useful information as to way they are not usable.

    What can I do to make them usable?

    Tuesday, November 5, 2019 12:22 PM


All replies

  • Can you provide some more details about your setup? Were you following any documentation? What is the end goal? 

    You can also review the docs we have on troubleshooting nodes in Batch

    Tuesday, November 5, 2019 6:33 PM
  • I would recommend filing a ticket for us through the Azure portal - ensure that you include your account name and the pool name which is experiencing issues and we can investigate.
    Tuesday, November 5, 2019 10:14 PM
  • I use setup details:

    VM Size - standard_a1

    sku - 2016-datacenter-smalldisk

    offer - WindowsServer

    version - latest

    Resize timeout: 45 min

    And that is it, those are all the options. However a lot of the nodes 70% go into an unusable state.

    The end goal is to run a couple of C# tasks on the nodes, but I don't see how that can be relevant, as the nodes never even get to the state where they can run the tasks.

    • Edited by cenh Thursday, November 14, 2019 9:45 AM
    Thursday, November 14, 2019 9:45 AM
  • Well one thing that could be the issue is the size of the VM you are choosing. That A size has 1CPU and 2GB of ram. Given that is the minimum requirement needed, you might be running into resource constraint issues. 

    Would you be able to try the same processes using a bit larger size VM to see if the issue repros? 

    Else, have you tried simply removing the Batch Account and deploying it again? Just to make sure its not a random error that occurred during the initial setup? 

    As Matt said you can always open a ticket to have it looked into further as well. That way the engineer can see what status the nodes are in and any backend errors. If you cannot open a ticket I can enable your subscription for one if you email me at and provide me with your Azure SubscriptionID and link to this issue. 

    Thursday, November 14, 2019 6:02 PM