error: 429 too many requests
endpoint: /api/v2/communication/interactions
region: ap-southeast-1
hey guys, im running into a wall with the platform api rate limits again. i know i mentioned this before with the analytics endpoint but this time its the core interactions endpoint. im trying to simulate a spike in inbound traffic for a stress test using jmeter.
my setup is pretty standard. i have a thread group with 100 active threads, each making a GET request to fetch interaction details. the ramp-up is set to 10 seconds so it should be a gradual increase. however, as soon as the request rate hits around 60 rps, the api starts throwing 429 errors consistently.
i checked the x-rate-limit headers in the response and it says:
x-rate-limit: 100
x-rate-limit-reset: 1678886400
x-rate-limit-remaining: 0
this is confusing because the documentation states that the global rate limit for most endpoints is higher, or at least that burst capacity should allow for short spikes. also, i am using a service account with admin privileges so i dont think user-level throttling is the issue here.
i tried adding a delay between requests in jmeter (constant timer set to 100ms) which brings the rps down to about 10, and the errors stop. but that doesnt help me test the actual load capacity of our integration. we need to know how the system behaves under high concurrency, not just steady low traffic.
is there a specific header i need to include to increase the limit? or maybe a different endpoint i should be hitting for bulk interaction data? i feel like im missing something obvious about how the rate limiter calculates the window. also, does the region matter? i read somewhere that ap-southeast-1 might have different quotas than us-east-1.
any help would be appreciated. im stuck on this load test phase and cant move forward with the performance report. thanks in advance.