Container Apps: Simultaneous Container Apps preemptions and platform component restarts in UK South on 14 October 2026

Question

Container Apps: Simultaneous Container Apps preemptions and platform component restarts in UK South on 14 October 2026

D C 20

On 14 October 2026 at 15:06:25–15:06:28 UTC, eleven apps in one Azure Container Apps environment in UK South were rescheduled. Seven logged Preempted by pod … across two worker nodes. Dapr scheduler and placement connections also closed during the same interval.

We checked the following:

No customer-initiated updates, restarts, scaling operations, or environment writes appeared in the subscription Activity Log for 15:00–15:10 UTC.
All affected revisions existed before the incident.
Environment diagnostics reported eight platform component restart events, including an ingress restart.
Azure Support reported overlapping managed-environment deployment or reconciliation activity but could not identify the underlying operation.

This was a single observed incident. We preserved the system logs, preemption messages, and revision timestamps.

The AI support agent referred me here after several exchanges without establishing the cause.

How can we determine whether Azure maintenance, environment reconciliation, or an unplanned node event caused the rescheduling? Can the platform restart and deployment signals be traced to the initiating operation?

Sridevi Machavarapu 33,820 Reputation points Microsoft External Staff Moderator

2026-10-15T16:47:48.32+00:00

Hello D C,

Thank you for sharing the detailed findings and logs.

Based on the information provided, the app rescheduling and platform component restarts occurred within the same time window as managed-environment deployment or reconciliation activity. However, the initiating operation and root cause have not yet been confirmed.

We will investigate further with the Azure Container Apps product group to correlate the platform telemetry and determine whether the events were related to planned maintenance, environment reconciliation, or an unexpected worker-node event. We will share an update once further findings are available.

1 answer

Your answer

Sridevi Machavarapu 33,820 Reputation points Microsoft External Staff Moderator

2026-10-15T16:47:48.32+00:00

Hello D C,

Thank you for sharing the detailed findings and logs.

Based on the information provided, the app rescheduling and platform component restarts occurred within the same time window as managed-environment deployment or reconciliation activity. However, the initiating operation and root cause have not yet been confirmed.

We will investigate further with the Azure Container Apps product group to correlate the platform telemetry and determine whether the events were related to planned maintenance, environment reconciliation, or an unexpected worker-node event. We will share an update once further findings are available.

Answer 1

Hello , Welcome to MS Q&A

I would suggest below actions to take further -

Check Azure Service Health history. In the Azure portal, open Service Health → Health history and filter for UK South, Azure Container Apps and 14 October. Also check Planned maintenance and Health advisories. This can show subscription-targeted events that never appeared on the public status page.
Ask Azure Support to correlate the platform events. The eight component restarts, including ingress, and the overlapping deployment or reconciliation activity are the strongest leads. Ask Microsoft to check the internal control-plane and infrastructure logs for the exact three-second window. These records may not be exposed in the customer Activity Log.
Establish what happened to the worker nodes. Request confirmation of whether the two nodes experienced maintenance, replacement, resource pressure, health degradation or an unexpected host event. The Preempted by pod messages alone do not prove that a node failed.
Correlate the Dapr and ingress events. Ask whether the scheduler and placement connection closures occurred as a consequence of the same platform operation or were independent events.

Pls check and let us know if any ques

Thanks

Deepanshu

D C 20 Reputation points

2026-10-15T14:32:57.78+00:00

Hi Deepanshu!

Thank you for this!

I checked out the Service Health History - sadly nothing Uk South related to Container Apps - same goes for Planned maintenance. Thank you for pointing it out to me!

Regarding points 2, 3 and 4 I did actually attempt that with support on Azure Portal - but got no where, and when I asked for escalation the agent which I believe is a fully AI agent just referred me to here. Do you know if I can flag this for a real agent?

Thanks!