0

I'm in a company where we have an elastic beanstalk configuration and it works fine with our CICD. The only issue is that earlier today i ran a stress test (basically just disabled AWS shield and went hammer on the DDOS). Regardless of what we did we couldn't get our medium sized server to crash (this was with a i9 cpu - so just a gaming pc) where we ran a multithreaded python script just sending GET requests. We then downgraded to t2.small because regardless of what we put the servers through the target response time went to almost 8 seconds before the CPU utilization got over 50% every time. The autoscaling works as it's supposed to but even after the upgrade (2 servers per trigger) the response time was still around 4-6 seconds. The trigger we're using right now is that it should create 2 new instances if the response time exeeds 1.5 seconds and then cool down for 360 seconds if it gets below 1 second.

The system is running PHP with apache2.4. There hasn't really been made the biggest configs other than virtualhost configs.

It can't be the db since its read and write both are below 0.1 seconds. I can't seem to figure out how to get the response time down?

Edit for further clarification of the setup: We have an entry at route 53 that points to our EBS where we have an nginx proxy to handle the requests to the servers. The servers that are running are t2.small. each server has a docker container setup and inside that we have an apache with php. We haven't setup any of the autoscaling on our own but used elastic beanstalks default options for the group.

LAMG
  • 101
  • 4
  • I'm confused about your situation, your problem, and what you're asking. Your title mentions EBS target response time - do you mean ALB? Suggest you rewrite your question for clarity. – Tim Mar 26 '21 at 07:01
  • to be honest i'm not completely sure what to edit since i'm no expert in aws compared to some of you guys. i've added a edit for clarification of the architecture. thanks – LAMG Mar 26 '21 at 14:35
  • 1
    Try breaking up your question into sections. Clearly describe the problem you're facing. Have another section with any logs or diagnostics. Then have a separate section for each thing you've tried to resolve the problem. – Tim Mar 26 '21 at 18:47
  • 1
    By EBS, do you mean EBS volume, or are you saying Elastic Beanstalk (EB)? For your actual issue: 1) Make sure you're not running out of CPU credits. 2) Check if you're hitting the max concurrent request limit in your webserver config 3) check if there's disk or memory issues (these often happen before CPU with web servers) – Shahad Apr 06 '21 at 01:22
  • It's EB not EBS. Has been edited in the title. – LAMG Apr 11 '21 at 22:13
  • Could very well be request limit. haven't even looked at the nginx (load balancer) or php.ini config. the apache config should not be the issue – LAMG Apr 11 '21 at 22:15

0 Answers0