0

I have a Jelastic environment to build a scalable backend api. It used to run on a single vps but when the API has lots of requests the vps can't handle it. So My environment A HAProxy loadbalancer. Default scale 2 cloudless and can scale to 40 cloudlets Behind that 2 nodes of 2 cloudless. They can vertical scale to 40 cloudlets And then a galera maria db cluster on 3 nodes with each 6 cloudless scalable to 40 each.

Still. When I run a test with loaderio I can only manage to get 120 calls a second. When I test to handle for example 500 calls a second I assumed that jelastic would scale vertical up to those 40 cloudless. But the nodes, even the haproxy has not much of increase in cpu and ram while running the test. In fact. It is almost nothing. But at certain point the calls fail with a handshake error on SSL. We use lets encrypt.

All hints are welcome to find why it can't handle this because this is the reason I'm trying to switch to jelastic. To make sure it can handle 10000 calls a second if needed.

Kind regads

Roel Van der Planken
  • 153
  • 1
  • 6
  • OK. I disabled the ratelimiter in HAProxy. But I thought this would only check on same IP. Probably loader.io runs on the same IP address. Or is the haproxy ratelimiter global? #rate limiting backend per_ip_rates stick-table type ip size 1m expire 10s store http_req_rate(10s) – Roel Van der Planken Jul 09 '20 at 12:04
  • As for the rate limiting, it is the custom setting. There are no rate limits in our default configuration of the HAproxy. If the analysis is needed - please provide us with the environment name and the Hosting provider to check your haproxy config. – Virtuozzo Jul 14 '20 at 15:43

0 Answers0