Share via

Will API specification update or policy updates to APIM cause transient error

Guru Pasupathy 60 Reputation points
2026-04-22T09:04:18.1666667+00:00

In a Premium (classic) tier APIM instance with availability zones enabled, does updating an API specification or API/operation policy cause a 5xx or 4xx response for requests that are already in flight on a gateway unit at the moment the update is applied to that unit?

The docs tell me what happens when a zone fails, but nothing about what happens inside a unit when it's applying a config change.

My specific question is — when a unit is applying an API spec or policy update, what happens to a request that has already been accepted by that unit? Does it complete against the old config? Does it get dropped? Or does the platform swap the config in a way that the in-flight request never even sees?

I'm not asking about zone failures or propagation between zones. I'm asking about the millisecond window on a single unit where the old config ends and the new one begins. Is that window even observable from a request's perspective?

Retry handles it — but I'd like to understand what I'm actually retrying against.

Azure API Management
Azure API Management

An Azure service that provides a hybrid, multi-cloud management platform for APIs.

0 comments

Answer accepted by question author

  1. Pravallika KV 14,235 Reputation points Microsoft External Staff Moderator
    2026-04-22T09:51:49.1366667+00:00

    Hi @Guru Pasupathy ,

    Thanks for reaching out to Microsoft Q&A.

    You won't see dropped or error-ing calls when you patch an API spec or policy on a single gateway unit.

    Here’s what happens under the hood in Premium-classic:

    1. Configuration is loaded in the background onto each unit
    2. Once the new config is fully in memory, it’s atomically swapped in
    3. Any requests that were already being processed continue to run against the old config and complete normally
    4. Only new requests see the updated API spec or policy

    In other words, there isn’t a moment where a unit is half-configured and returns 4xx/5xx for in-flight calls. The switch from old to new is instantaneous from the request’s point of view, so you won’t observe a transient break. (Zone-failure scenarios are different, they terminate in-flight calls and force retries but routine config updates on a healthy unit don’t.)

    Hope this helps!

    If the resolution was helpful, kindly take a moment to click on User's imageand click on Yes for was this answer helpful. And, if you have any further query do let us know.

    1 person found this answer helpful.
    0 comments

1 additional answer

Sort by: Most helpful
Most helpful Newest Oldest
  1. Guru Pasupathy 60 Reputation points
    2026-04-22T11:10:36.3333333+00:00

    Hello @Pravallika KV
    Thank you, that's really helpful! Appreciate the quick response you provided.

    Is there a public doc that talks about this atomic swap behaviour? Would love to reference it.

    Guru Pasupathy

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.