Share via

How to resume Azure database stuck in Pausing Status

Gustavo Davila 25 Reputation points
2025-10-06T14:14:46.3+00:00

I have one SQL Database on the serverless tier which is stuck in the 'pausing' state and therefore allows no connections at all. The database paused correctly yesterday evening, but this morning will not resume. The change analysis in the Azure Portal shows status going from Paused to Pausing approximately 90 minutes ago, whereas it would normally show it as going from Paused to Online.

The database is serverless, general purpose, standard series, hosted in East US 2 . I see no Azure status alerts for this service in the region. Any ideas on how to resolve?

Azure SQL Database

Answer recommended by moderator
  1. Victor Olazabal 20 Reputation points
    2025-10-06T15:29:29.2366667+00:00

    I am also experiencing the same issue, getting the following message from Resource Health

    At Monday, October 6, 2025 at 11:26:14 AM EDT, the Azure monitoring system received the following information regarding your SQL database:

    We're sorry your SQL database is experiencing transient login failures. Currently, Azure shows the impacted time period for your SQL database resource at a two-minute granularity. The actual impact is likely less than a minute – average is 2s. We're working to determine the source of the problem. If we are able to find the reason for unavailability, it will appear here within an hour after the downtime event. Please check back later to see if more information is provided.

    2 people found this answer helpful.
    0 comments No comments

1 additional answer

Sort by: Most helpful
  1. Sina Salam 28,281 Reputation points Volunteer Moderator
    2025-12-27T14:23:00.7566667+00:00

    Hello Gustavo Davila,

    Welcome to the Microsoft Q&A and thank you for posting your questions here.

    I understand that you need how you can resume Azure database stuck in Pausing Status.

    In most documented cases, the recommended action is to allow time for Azure to self-heal. Customers are advised to wait anywhere from 2 to 6 hours, while monitoring the database state in the Azure portal and reviewing Resource Health for transient backend errors such as login failures.

    During this waiting period, it is also best practice to verify whether the issue correlates with a regional service degradation. Checking Azure Service Health in the portal can help confirm whether SQL Database services are experiencing incidents that may delay resume operations. This step prevents unnecessary configuration changes when the root cause is external and already being addressed by Microsoft.

    If the database does not recover after a reasonable waiting window, a manual resume trigger may be attempted. This can be done either through the Azure portal by navigating to Compute + Storage and selecting Resume, or by nudging the compute allocation using Azure CLI. The following command can sometimes force the platform to re-evaluate the database state:

    az sql db update \
      --resource-group <ResourceGroup> \
      --server <ServerName> \
      --name <DatabaseName> \
      --compute-model Serverless \
      --capacity <existing_or_new_capacity>
    

    This approach does not change the database configuration but may prompt the backend to reassign compute resources.

    If the database remains stuck despite manual attempts, the only reliable resolution is to engage Microsoft Support. Support engineers have access to backend controls that are not exposed to customers and can directly reset the database lifecycle state. This escalation is strongly recommended when business-critical workloads are impacted, as no customer-side operation can fully remediate a platform-level pause failure.

    I hope this is helpful! Do not hesitate to let me know if you have any other questions or clarifications.


    Please don't forget to close up the thread here by upvoting and accept it as an answer if it is helpful.

    0 comments No comments

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.