scale set azure agents pipelines are too slow in azure devops

muhammed aneesh 0 Reputation points
2025-04-14T16:35:53.7066667+00:00

I am trying to configure scale set agent in azure devops so I configured scale set with a predefined image which have all the software required installed.
Image: image1
size:d4s_v3

configured it as an agent and when the pipeline executed it works well but took around 1hr, which is not usual

so I configured a normal VM with same image and same size. and connected it as azure agent then the same pipeline exacted and it took only 12 min which is usual. any reason why scale set agents taking too long

Azure DevOps
0 comments No comments
{count} votes

1 answer

Sort by: Most helpful
  1. Suresh Chikkam 1,100 Reputation points Microsoft External Staff
    2025-04-15T13:22:53.5966667+00:00

    Hey muhammed aneesh,

    The delay you're seeing with the scale set agent like a classic cold start issue with scale set agents. Even though the image and size are the same, scale sets spin up new VMs on demand, and that whole process booting the VM, initializing the agent, and any background setup can add a lot of time before the actual pipeline even starts.

    Try this few things:

    Keep a couple of VMs always running in the scale set to avoid cold starts.

    • Check that the DevOps agent and all dependencies are fully baked into the image to avoid extra setup time.

    Also worth checking if there are delays in networking or agent registration right after VM spin-up.

    If the standalone VM runs the job in 12 minutes, you should be able to get similar performance from scale sets with the right setup.

    Hope it helps!


    Please do not forget to click "Accept the answer” and Yes wherever the information provided helps you, this can be beneficial to other community members.

    User's image

    If you have any other questions or still running into more issues, let me know in the "comments" and I would be happy to help you.


Your answer

Answers can be marked as Accepted Answers by the question author, which helps users to know the answer solved the author's problem.