Power Platform Community Forum Thread Details

The problem

After our production environment was rebuilt via an environment copy (copied into a new Sandbox, then converted to Production), every single cloud flow in the environment now fails with:

UnresolvableHostName
Http request failed with status code 'HostNotFound' and status message: 'No such host is known.'

It fails on the first Dataverse action in every flow. Different flows, different tables, different actions, different connections — same error every time. Nothing has succeeded since the copy.

The part that makes no sense to me

Dataverse itself is completely healthy.

The environment URL loads the maker portal normally

Environment State = Ready

Not in Administration mode

I can browse tables, run apps, and query data in the UI without any issue

The Dataverse connection shows Connected with a green check

So the org is up and reachable — but the Power Automate runtime apparently cannot resolve it.

Key clue: the failure takes 0 seconds

This is what I keep coming back to. From the retry detail pane:

Retry       : 1 of 12
Duration    : 0 seconds
Start time  : 1:36:52 PM
End time    : 1:36:52 PM   <-- identical
Status      : UnresolvableHostName

Start and end times are identical on all 12 retries.

A network problem has to time out — it takes time to fail. This fails instantly, which says the hostname is being rejected at name resolution, before a single packet leaves. The ~3.5 hour total run duration is purely retry-policy backoff between 12 instant failures.

It is also not an auth failure — a bad identity gives 401/403, not "no such host is known."

No org URL in the action inputs

There is nothing hardcoded for me to fix. Raw inputs:

host:
  connectionReferenceName : shared_commondataserviceforapps
  operationId             : ListRecords
parameters:
  entityName              : xxx_campaignteams
  accept                  : application/json;odata.metadata=minimal

The Environment parameter is on (Current), so it isn't trying to reach another environment. The host is resolved at runtime from the connection reference.

Everything I have already tried (none of it worked)

Confirmed environment not in Administration mode, State = Ready

Checked Service Health — no active incident

Verified Dataverse connection status = Connected

Verified connection reference is bound, green check

Recreated the Dataverse connection from scratch

Repointed the connection reference to a different known-good connection

Verified the owning service account is Active with correct security roles

Turned flows Off, waited, turned back On

Resubmitted failed runs many times

Confirmed the flow definition was regenerated (Modified timestamp updated) — still fails identically

Per-flow fixes have zero effect, which is what makes me think this is environment-level rather than anything to do with the flows themselves.

Also broken: dataflows

Dataflows in the same environment fail to refresh with Credentials are required to connect to the CommonDataService source. The data source path in the error is correct — it matches the live environment URL. So the copy seems to have left credential and binding state broken across multiple component types, not just cloud flows.

Why I can't use the documented workaround

The known fix floating around for UnresolvableHostName is to open the flow and re-save it, which regenerates the binding.

I can't. The flows are in a managed solution in production and I don't have edit rights — re-saving would create an unmanaged layer over a managed solution in prod, which our release governance prohibits. And with every flow in the environment affected, a per-flow workaround wouldn't scale anyway.

My questions

Has anyone seen UnresolvableHostName affect an entire environment after a copy — as opposed to a single flow? Everything I can find online describes it hitting one flow at a time.

Does the Sandbox → Production conversion re-register the environment with the Power Automate runtime? My working theory is that the runtime registration still points at the source org of the copy, while Dataverse itself is correctly on the new org. Can anyone confirm or kill this theory?

Is there any way to see what hostname a Dataverse connection actually resolves to at runtime? Nothing in the maker portal or admin center seems to expose it, and without it I'm guessing.

Is there a supported way to force a re-registration or re-bind at the environment level — PowerShell, pac CLI, admin API — rather than flow by flow?

If you've hit this: what actually fixed it? Did it need Microsoft to intervene server-side, or was there something customer-side that worked?

Context

This has been an active production incident for two weeks. The environment copy was itself performed on Microsoft's advice to remediate an earlier problem, and it has changed the failure mode rather than fixed it. I have open support tickets, but I'm posting here in case someone has hit this exact combination and knows the shortcut.