Describe the bug
During network io exception we do the region failover to other available region.
All the subsequent requests goes to other region.
However current request keep retrying over the same region for 120 times before floating error to user if not succeeded.
To Reproduce
Have two read regions. Induce the IO exception in one region, and you will see that flight request keep retrying in same region
Expected behavior
In flight request should move to other region once it mark the current region unavailable temporally
Actual behavior
In flight request keep retrying in same region after marking it unavailble
Describe the bug During network io exception we do the region failover to other available region. All the subsequent requests goes to other region. However current request keep retrying over the same region for 120 times before floating error to user if not succeeded.
To Reproduce Have two read regions. Induce the IO exception in one region, and you will see that flight request keep retrying in same region
Expected behavior In flight request should move to other region once it mark the current region unavailable temporally
Actual behavior In flight request keep retrying in same region after marking it unavailble
Environment summary SDK Version: V2/V3/V4
Additional context This problem probably exist in .NET cosmosdb sdk too. Below is the relevant code where it keep retrying in same region. Setting retryReguestOnPreferredRegion false Which is getting use here