Stuck with Lambda concurrency errors after deploying from our Windows build pipeline

Jenifer Teffani 20 Reputation points
2026-10-08T16:32:17.3466667+00:00

I was attempting to set up provisioned concurrency for our backend serverless services running alongside our Windows for Business workloads, configured the pre-warming settings, then got into a “ProvisionedConcurrencyConfigNotFoundException” loop during our automated deployments.

I deleted the concurrency configuration in an attempt to “start over”, and now I’m in an endless loop where the functions need pre-warmed instances to avoid cold-start timeouts on our business apps, but the deployment fails because the provisioned concurrency config cannot be found on the target. Every time the pipeline triggers from our build agent, it restarts the loop.

I now see the pipeline attempt to allocate the warm capacity, but then it fails immediately because it's trying to attach the settings directly to an unqualified function or an unpinned revision rather than a published version alias, which I can't get past because the new code isn't aliased yet.

How do we properly automate version alias targeting in SAM or CDK pipelines so provisioned concurrency applies cleanly without throwing this exception?help please

Windows for business | Windows 365 Business
0 comments No comments

1 answer

Sort by: Most helpful
  1. Tracy Le 13,530 Reputation points Independent Advisor
    2026-10-08T17:40:52.6033333+00:00

    Hello Jenifer Teffani,

    The error is consistent with provisioned concurrency being applied to an unqualified function reference rather than a published version or alias. Provisioned concurrency can only be associated with a version or alias, so if the deployment pipeline is targeting $LATEST, the deployment will fail.

    Since the problem started after the configuration was manually removed, I would also verify whether the CloudFormation stack and the actual Lambda resources are still in sync. When a provisioned concurrency configuration is deleted outside the deployment process, subsequent deployments can fail while trying to update or remove a resource that no longer exists.

    For SAM or CDK deployments, the recommended approach is to publish a version automatically, create or update an alias, and then attach provisioned concurrency to that alias. If you can share whether the pipeline uses SAM, CDK, or raw CloudFormation, it will be easier to identify the exact deployment step causing the exception. If this helps, please accept the answer.

    Tracy Le

    Was this answer helpful?

    0 comments No comments

Your answer

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