Investigative information
- Timestamp: First observed
2026-08-12T22:08:37.966Z
- Function App version:
azurefunctions: 4.1053.200.26403
- Hosting plan: Y1 Dynamic / classic Consumption
- Invocation/operation ID: Empty
- Region: North Europe
- Runtime: .NET isolated
Repro steps
We do not have a reliable application-level reproduction. The exceptions appear during host/worker lifecycle activity shortly before otherwise successful scheduled invocations.
Expected behavior
Normal worker shutdown or specialization should terminate the gRPC stream gracefully, or the expected disconnect should not be emitted as Error/Critical exception telemetry.
Actual behavior
After the North Europe rollout of host build 4.1053.200.26403, we observed a substantial increase in this exception fingerprint:
System.Net.Sockets.SocketException at
Grpc.AspNetCore.Server.Internal.PipeExtensions+<ReadStreamMessageAsync>d__15`1.MoveNext
Nested chain:
IOException: The request stream was aborted
→ HTTP/2 connection faulted
→ ConnectionResetException
→ SocketException: remote host forcibly closed the connection
→ FunctionRpcService.EventStream
There are no application-owned frames or operation IDs.
During 2026-08-12T20:58Z–2026-08-17T06:00Z:
- Host 4.1053: 93 events across 39 Function App roles.
- Host 4.1052: one event across one role.
- For one equivalent integration population:
- 4.1053: 79 events across 26 roles.
- 4.1052: zero events.
All 35 apps observed on 4.1053 were in North Europe, so host version and region are confounded. This demonstrates a strong rollout association, not confirmed release causation.
For 78 of 79 integration events, the next same-role request occurred 62–86 seconds later. All 78 requests succeeded, and no failed requests occurred within ±120 seconds.
Known workarounds
No runtime workaround is known. We suppress only this exact exception fingerprint in alerting when the operation ID is empty.
Related information
This appears equivalent to:
The increase is strongly associated with the rollout of hosted build 4.1053.200.26403. That build has no corresponding public tag, so we could not determine the exact source delta.
PR #11853 modifies worker-channel shutdown and disposal and may be relevant, but we cannot establish that it is included in the hosted build.
The expected outcome is that normal worker shutdown does not emit Error/Critical exception telemetry.
Investigative information
2026-08-12T22:08:37.966Zazurefunctions: 4.1053.200.26403Repro steps
We do not have a reliable application-level reproduction. The exceptions appear during host/worker lifecycle activity shortly before otherwise successful scheduled invocations.
Expected behavior
Normal worker shutdown or specialization should terminate the gRPC stream gracefully, or the expected disconnect should not be emitted as Error/Critical exception telemetry.
Actual behavior
After the North Europe rollout of host build
4.1053.200.26403, we observed a substantial increase in this exception fingerprint:Nested chain:
There are no application-owned frames or operation IDs.
During
2026-08-12T20:58Z–2026-08-17T06:00Z:All 35 apps observed on 4.1053 were in North Europe, so host version and region are confounded. This demonstrates a strong rollout association, not confirmed release causation.
For 78 of 79 integration events, the next same-role request occurred 62–86 seconds later. All 78 requests succeeded, and no failed requests occurred within ±120 seconds.
Known workarounds
No runtime workaround is known. We suppress only this exact exception fingerprint in alerting when the operation ID is empty.
Related information
This appears equivalent to:
The increase is strongly associated with the rollout of hosted build
4.1053.200.26403. That build has no corresponding public tag, so we could not determine the exact source delta.PR #11853 modifies worker-channel shutdown and disposal and may be relevant, but we cannot establish that it is included in the hosted build.
The expected outcome is that normal worker shutdown does not emit Error/Critical exception telemetry.