I'm using JMeter CLI to load tests our micro services (Nginx+PHP-FPM+Laravel).
Micro services are all under an AWS Application Load Balancer which performs a health-check.
For that micro service, load test runs with an average of 50req/s. After few minutes, health check is failing with
50#50: *835 epoll_wait() reported that client prematurely closed connection, so upstream connection is closed too while sending request to upstream, client: **.**.**.**, server: _, request: "GET /health HTTP/1.1", upstream: "fastcgi://unix:/var/run/php-fpm.sock:", host: "**.**.**.**:32807"
While the requests from JMeter are still ok.
It's the only micro service that gives the error. They all have the same nginx config though.
I'm really clueless.