ICP Connection Data API: Many failures with code:500

  • This discussion is locked, please start a new discussion.
  • 247 Views
  • Last post 05 February 2021
  • Discussion is solved
Clive Gifford posted this 01 February 2021

Starting over the recent weekend, the ICP Connection Data API has been behaving very erratically.

In particular, the "get by ID list" call appears to be failing with code 500 ({"Code":500,"Message":"API is currently unavailable to process this request."}) a high proportion of the time, even though it also does sometimes succeed with exactly the same parameters.

Is there some kind of issue with the registry database that is causing this?

When is it likely to be fixed?

Matthew Keir posted this 05 February 2021

Hi Clive,

The aggregate volumes of calls from our subscribers to the ICP connection data API have been causing issues for the underlying registry infrastructure managed by our service provider. To address this issue the registry manager has significantly reduced the number of concurrent calls they will manage before an error is generated and returned. You'll receive the error "API is currently unavailable to process this request" when this threshold binds. This message may appear intermittent, as it depends on how many calls our other users are generating at the same time as you.

To preserve the integrity of our API, especially for our subscribers using the get by ID calls to engage consumers and offer real-time advice, we are making changes to our API management policy. We will increase throttling on the 'get by ID list' call by reducing the set size to 50 ICPs and also reduce the MB threshold across all calls each user can make to 5 MB per 15 minutes. The change will help spread the load our users generate in aggregate that is adversely impacting the underlying registry infrastructure. This change will impact all our users and they may need to modify scripts accordingly.

Cheers,

Matthew

This discussion is locked, please start a new discussion.