Unable to Access PostgreSQL Flexible Servers – Request for Restoration to Supported SKU/Region

Rumsan Associates 20 Reputation points
2026-05-19T09:30:06.9333333+00:00 
Hi Azure Support Team,

We are currently experiencing a critical issue with our Azure Database for PostgreSQL Flexible Servers. Both of our servers have become unresponsive, and we are unable to establish any connections.

We also received a notification regarding planned maintenance and deprecation of the current SKU, which we suspect may be related to the issue.

Details:

Service: Azure Database for PostgreSQL Flexible Server
Region: West Europe
Servers affected:
giga-production-server
giga-staging

Issue:

Servers are not responding
Unable to connect via any client or service
Impacting production and staging environments

We understand that certain SKUs (e.g., Intel V5 confidential compute) have been deprecated. We would like your assistance with:

Restoring both servers to a supported configuration (e.g., AMD V5 or equivalent)
Migrating them to a stable and available region if required
Ensuring data integrity during the restoration process
Providing guidance on preventing similar issues in the future

Reference link:
https://app.azure.com/h/GGQY-G1G/f6b5f5

This is impacting our live systems, so we would appreciate urgent support on this matter.

Please let us know if any additional information is required from our side.
Azure Database for PostgreSQL
Answer accepted by question author
Manoj Kumar Boyini 19,590 Reputation points Microsoft External Staff Moderator
2026-06-04T04:04:29.2133333+00:00

Hi @Rumsan Associates

We have received confirmation from the PostgreSQL engineering team that the affected server has now been successfully removed from the backend.

Root cause: The server entered an inconsistent state due to an internal sync issue between the Azure Resource Manager (ARM) layer and the backend service. As a result:

  • The resource was not visible/manageable from your side
  • Deletion operations failed
  • However, the resource remained active in the backend, which is why billing continued

The PG team performed a manual cleanup and forced deletion from the backend, ensuring the resource and all associated components are now removed.

For the billing concern, we recommend raising a billing support request for review and adjustment.

Was this answer helpful?

1 person found this answer helpful.

1 additional answer

Sort by: Most helpful
Most helpful Newest Oldest

  1. AI answer

    2026-05-19T09:30:39.7333333+00:00

    For unresponsive Azure Database for PostgreSQL flexible servers after planned maintenance and SKU deprecation, use the following actions and checks:

    1. Check for Azure-side outage or maintenance
      • In the Azure portal, open Service Health and check Service issues and Planned maintenance for the West Europe region to see if there is an active outage or ongoing maintenance impacting Azure Database for PostgreSQL flexible server.
      • Also check the Resource health blade for each affected server to confirm whether Azure reports them as unavailable or in an error state.
    2. Validate whether this is a planned maintenance impact
      • Planned maintenance (including minor version upgrades and new software deployments) can cause a few minutes of downtime while the service restarts the server with updated software.
      • If high availability (HA) is enabled, maintenance is applied to the standby first and then failover is performed to minimize downtime; if HA is not enabled, the primary server is restarted and can be briefly unavailable.
      • If the servers remain unavailable significantly beyond the expected few minutes, treat this as a service issue and proceed with support escalation.
    3. Confirm server state and restart if needed
      • In the Azure portal, open each flexible server (giga-production-server and giga-staging) and confirm the server state.
      • If a server is in a failed or stopped state, perform a manual restart. Be aware that if maintenance could not be applied while the server was stopped, it will be applied on restart, which can extend restart time by several minutes.
    4. If servers remain unavailable, open an urgent support request
      • Because this is production-impacting and related to platform maintenance/SKU deprecation, open an Azure support request under the Azure Database for PostgreSQL flexible server category.
      • Select a high severity level and clearly state:
        • Both server names and region (West Europe).
        • That they became unresponsive following planned maintenance and SKU deprecation.
        • That they host production and staging workloads.
      • For capacity or provisioning-related constraints (for example, if the deprecated SKU cannot be re-provisioned in the current region), request:
        • Restoration or migration to a supported SKU (for example, AMD-based v5 or other supported SKU) in the same region if possible.
        • Region access or migration guidance if West Europe capacity or SKU availability is constrained.
    5. Handle capacity and region constraints
      • If the current SKU or availability zone is no longer available or has capacity constraints, Azure guidance is to:
        • Choose a different availability zone in the same region when the error indicates zone capacity issues.
        • Choose a different SKU when the error indicates the specified SKU is not supported or temporarily unavailable in that region.
      • If provisioning or restoration continues to fail due to capacity or program-region limits, open a Region access request (under the Azure PostgreSQL flexible server support topic) and specify the vCores required so that Azure can review and potentially grant access to additional regions.
    6. Data integrity and business continuity
      • Azure Database for PostgreSQL flexible server is designed to perform automatic recovery after unplanned downtime, including provisioning a new server instance if the original host fails, without requiring manual intervention.
      • For planned maintenance (including compute/storage scaling or software updates), Azure drains connections, cancels uncommitted transactions, and reattaches storage to a new instance, performing recovery before accepting new connections.
      • If high availability is configured, failover to the standby is used to minimize downtime during maintenance and unplanned failures.
      • If the outage threatens SLAs and geo-replication is configured, consider promoting a read replica in another region to primary or to an independent server, then update application connection strings as needed. This requires user action and may involve some data loss equal to the replication lag at the time of promotion.
    7. Preventing similar issues in the future
      • Monitor and act on planned maintenance notifications:
        • Use Azure Service Health → Planned maintenance to track upcoming maintenance and see all impacted servers in a region.
        • Export the impacted resources list to CSV to plan changes and maintenance windows.
      • Avoid stopping servers during scheduled maintenance windows; if a server is stopped, maintenance will be applied on the next restart, which can extend restart time.
      • For critical workloads:
        • Enable high availability to reduce downtime during maintenance and unplanned failures.
        • Consider geo-replication and a tested promotion/runbook for regional failures.
      • For capacity and SKU lifecycle changes:
        • Proactively plan migration from deprecated SKUs to supported SKUs and, if necessary, to regions with sufficient capacity, coordinating with Azure support when capacity or program-region limits apply.

    Because the current issue involves unresponsive production servers and a deprecated SKU, direct engagement with Azure Support is required to:

    • Restore the servers on a supported SKU in West Europe if capacity allows, or
    • Assist with migration to another region/SKU while preserving data integrity.

    References:

    AI-generated content may be incorrect. Read our transparency notes for more information.

    Was this answer helpful?

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.