B2C PreTokenIssuance API Connector is timing out after 5 seconds instead of 20 seconds

Hello @Adam Gatt,

I see that you've already raised a support ticket with our team — 2504070030009160. Please correct me if this is not the ticket you submitted.

I’ve reviewed the case, and I can confirm that our support engineer has already notified the engineering team to make the necessary changes to your tenant. I’ll continue to monitor the progress and will notify you once your tenant has been added to the exemption list.

Thanks, @Raja Pothuraju

Yes, that is my support case. This hasn't been communicated to me as yet by the support team.

Can you please explain why a change to the tenant is necessary? Why did Microsoft change the behaviour, and where was it communicated / linked documentation not updated?

We will need this rectified on all of our B2C tenants as we have multiple for different environments.

@Raja Pothuraju , I will need my tenants exempted as well. Should I open a ticket and reference this thread or is there a preferred way of submitting this request?

Thank you.

Hello @Adam Gatt,

The exemption has been applied to all five of your tenants. I’m sharing the tenant details via private message, as they contain personally identifiable information (PII).

Please note that this is a temporary exemption, valid only until December 31st, 2025. After that date, the changes will be re-applied.

There has been a change in the way API connectors operate—the default timeout has been reduced to 10 seconds, and this behavior is now by design:

Public Doc: API connectors overview

Our team analyzed the response times of APIs linked to this service. Out of 506 records, 223 requests completed in under 10 seconds, while 283 requests hit the 10-second timeout. This means approximately 56% of the requests are failing due to the timeout limit.

As demonstrated, the majority of API responses in your environment take longer than 10 seconds, which is the root cause of the timeouts. This is why our engineering team has temporarily exempted your tenants from the timeout changes—to give you time to assess and optimize your API response times.

For your production tenant ehmpb2c, only 1.36% of requests took more than 10 seconds. However, we identified a clear pattern: the longer response times are tied to processing errors, where the API returns invalid or unprocessable responses (e.g., 409 or 401 errors). These errors not only cause failures but also increase processing time, often leading to timeouts.

Because of the reduced timeout threshold, you may now see timeout errors instead of specific processing errors—since the 10 seconds elapse before the original error is surfaced. It's important to investigate and fix the scenarios where your APIs return such invalid content to prevent future disruptions.

Hello @Mike Palgon,

Yes, but still, it will be a temporary solution. As I mentioned above, there has been a change in the way API connectors operate—the default timeout has been reduced to 10 seconds, and this behavior is now by design:

Public Doc: API connectors overview

Please review your API failures and try to prevent them.

Thank you Raja Pothuraju. If this is just temporary, then I'll work on resolving it in the API connector. The API endpoint is an Azure Function that connects to Azure SQL and Graph API. I believe the latency is coming from Azure SQL (it's thinly provisioned to save $$ in a dev environment).

Hi @Raja Pothuraju - I've responded to the private message. Best to take this conversation off a public forum. Thanks for your help.

Hi @Adam Gatt,

Thank you for taking this up over private message.