SQL Warehouse stuck on "Cluster Start-up Delayed"

[Nid-cbs]

Hi everyone,

I'm running into an issue with my Starter Warehouse on Databricks and would appreciate any help or pointers.

Problem: My SQL Warehouse has been stuck in a Starting state with the following warning:

Cluster Start-up Delayed. Please wait while we continue to try and start the cluster. No action is required from you.
cluster-id: 0603-142345-jqh97kz4

It hasn't progressed past this state despite waiting.

[Ashwin_DSA]

Hi @Nid-cbs,

Sorry you're running into this.

A few minutes of startup time can occur while the warehouse is provisioning, but if it has been stuck in Starting for quite a while, it could also be related to cloud quota or backend capacity, so it would help to narrow down the environment a bit.

Could you share which cloud this is on (AWS, Azure, or GCP), whether this is a paid workspace, a trial workspace, or Databricks Free Edition, roughly how long it has been stuck in Starting, and whether retrying the warehouse start shows the same behaviour each time?

For reference, these public docs may help: SQL warehouses overview, SQL warehouse sizing, scaling, and queuing behavior, Databricks Free Edition FAQ, and Databricks support.

If this is a paid workspace and the issue is recurring, I’d recommend opening a support ticket so the team can investigate the workspace and backend logs. If this is Databricks Free Edition, please mention that as well, since the support path is different.

If this answer resolves your question, could you mark it as “Accept as Solution”? That helps other users quickly find the correct fix.

[Nid-cbs]

Hi Ashwin, thanks for the detailed response!

To answer your questions:

- Cloud: Azure
- Workspace type: Trial workspace (14-day trial, created via the Azure Portal around 2 weeks ago)
- How long stuck: More than 10 minutes each time
- Retry behaviour: Yes, I tried stopping and restarting the warehouse multiple times, and also attempted resizing it, but the same Starting delay occurs every time

One thing I was unsure about , since this is a trial workspace nearing its end, could the pricing tier be a factor? I assumed it would be handled automatically, but wondering if there is something I need to change manually.

Thanks again for your help!

[Ashwin_DSA]

Hi @Nid-cbs,

Thanks for the extra detail. Since this is on Azure and you’re seeing the same behaviour after multiple retries and even after resizing, it sounds less like a one-off startup delay and more like an environmental issue.

I can’t say for certain that the trial status is the cause, but it’s reasonable to check. On Azure, there may be required setup steps for serverless SQL warehouses in supported regions, and for classic or pro SQL warehouses, there must be sufficient Azure vCPU quota available. 

Since this is a 14-day Azure trial workspace and the issue is happening consistently, I’d suggest opening a support ticket through your Azure / Databricks support path if you have that available, because they’ll be able to confirm whether this is related to workspace state, serverless eligibility, quota, or another backend provisioning issue. From what you described, it doesn’t sound like you need to make any obvious manual pricing-tier change to get a Starter Warehouse past Starting.

If this answer resolves your question, could you mark it as “Accept as Solution”? That helps other users quickly find the correct fix.

[Nid-cbs]

Hi Ashwin,

Thank you.

Just to update you on my end : my 14-day trial was expiring so I upgraded to a Pay-As-You-Go Premium subscription. However, after upgrading I tried starting the SQL Warehouse again and got the same startup error.

Following your advice on vCPU quota, I attempted to increase it from 10 to 16 cores in Azure but the request failed. I've now raised a support ticket with Microsoft to get the quota increased.

My plan is to wait for the quota increase to go through and then retry the warehouse start. If it still doesn't work after that, I'll raise a Databricks support ticket as well.

Will keep this thread updated. Thanks again for your guidance!