We are using CosmosDB with SQL API and Serverless.
We are getting strange BadRequest CosmosException's without a reason text and a substatus 0 text.
We enabled AllowBulkExecution to load some data quickly running 64 tasks in parallel.
When ever this happens I am able to find the item and run it alone and I am able to call UpsertItemAsync on it. but whenever I try and do it in parallel, it fails almost immediately.
Nothing have changed in the data. Its the same /id value and partition key.
I've excluded everything from the index include path, and using /id as the partition key.
The diagnostic data do not show any obvious difference between a successful call and a failed one.
Have anyone experienced anything like this before and know what the root cause is?