Throughput greatly impacted between Core version 4.0.31.0 and 4.0.32.0 #4053
Description
Describe the bug
We can see this issue in our service doing a lot of batches of events to SQS, and performing a lot of actions concurrently. Upgrading from a Core version 4.0.31.0 to 4.0.32.0 completely kills off the throughput and grind our service to a halt.
We run the service within EKS and have credentials renewing. Therefore I suspect the commit c43d041 being the root cause for this. Reusing the same AWSCredentials for the single pod instance we are running will essentially only allow us to fetch signing credentials for one message at a time.
Regression Issue
- Select this option if this issue appears to be a regression.
Expected Behavior
We would expect the same throughput for our service doing that minor upgrade.
Current Behavior
No exceptions, no errors in logs and limited visibility for the issue. But we can see that our requests throughput raise alarms on our side.
Reproduction Steps
.
Possible Solution
If possible, return from GetCredentials/GetCredentialsAsync if we have valid credentials within the ttl before entering the semaphore locks.
Additional Information/Context
No response
AWS .NET SDK and/or Package version used
AWSSDK.Core 4.0.32.0
Targeted .NET Platform
.NET 9
Operating System and version
Debian / Docker