We are facing the above mentioned error in our 30 minute test with 4080 Vusers from 16 Azure RM Load Servers (XL) from US East location. (Each server generating 4080/16 = 255 Vusers). The clip just contains a single POST API that we need to scale upto 10K/second. (We have successfully scaled to that level with CT using some related APIs of the same application, so this should not be an issue). Ramp up time was 5 minutes.
The error is spread across the test duration more or less evenly and in total ~2 % of total messages in the test face this error.
On running the exact same scenario with the same 16 Azure boxes, with 1/3rd of the load mentioned (4080/3 = 1360 Users), not a single error is observed.
Nothing seems to be wrong with load/result servers, looking at monitoring stats.
(1) Can you please share more information on what this error means from soasta's end?
(2) I don't see any errors on New Relic. Why?