기억 파편

• 인스턴스 여러개 띄운거 -> multi-process
• 한 인스턴스 안에서 동시 요청 -> multi-thread
  • controller는 스프링에 의해 자동 multi thread(여러 요청이 동시에 들어올 경우)
  • 서비스 로직을 multi-thread로 구현한다는 것은 보통 비동기로 구현하는 것을 의미

multi-thread파악

11번 우르르쾅쾅 요청하기

curl -v http://localhost:5000/api/setting/enums & curl -v http://localhost:5000/api/setting/enums & curl -v http://localhost:5000/api/setting/enums & curl -v http://localhost:5000/api/setting/enums & curl -v http://localhost:5000/api/setting/enums & curl -v http://localhost:5000/api/setting/enums & curl -v http://localhost:5000/api/setting/enums & curl -v http://localhost:5000/api/setting/enums & curl -v http://localhost:5000/api/setting/enums & curl -v http://localhost:5000/api/setting/enums & curl -v http://localhost:5000/api/setting/enums

서버 설정

스래드 맥스를 5로 잡았다. 

springboot 2.7기준

server.tomcat.threads.max=5
server.tomcat.threads.min-spare=1

10초 딜레이를 넣은 api를 요청했다.

결과

5-5-1 순으로 쌓였다가 순차 실행한다.

2023-04-25 07:25:52 INFO  [c.n.d.common.filter.RequestLogFilter    .doFilter            :  36][http-nio-5000-exec-4] [REQUEST] [GET] /api/setting/enums
Headers : {host=localhost:5000, user-agent=curl/7.64.1, accept=*/*}
Request body :  - 

2023-04-25 07:25:52 INFO  [c.n.d.common.filter.RequestLogFilter    .doFilter            :  36][http-nio-5000-exec-2] [REQUEST] [GET] /api/setting/enums
Headers : {host=localhost:5000, user-agent=curl/7.64.1, accept=*/*}
Request body :  - 

2023-04-25 07:25:52 INFO  [c.n.d.common.filter.RequestLogFilter    .doFilter            :  36][http-nio-5000-exec-5] [REQUEST] [GET] /api/setting/enums
Headers : {host=localhost:5000, user-agent=curl/7.64.1, accept=*/*}
Request body :  - 

2023-04-25 07:25:52 INFO  [c.n.d.common.filter.RequestLogFilter    .doFilter            :  36][http-nio-5000-exec-1] [REQUEST] [GET] /api/setting/enums
Headers : {host=localhost:5000, user-agent=curl/7.64.1, accept=*/*}
Request body :  - 

2023-04-25 07:25:52 INFO  [c.n.d.common.filter.RequestLogFilter    .doFilter            :  36][http-nio-5000-exec-3] [REQUEST] [GET] /api/setting/enums
Headers : {host=localhost:5000, user-agent=curl/7.64.1, accept=*/*}
Request body :  - 

---> 응답 다 받고

2023-04-25 07:26:02 INFO  [c.n.d.common.filter.RequestLogFilter    .doFilter            :  36][http-nio-5000-exec-2] [REQUEST] [GET] /api/setting/enums
Headers : {host=localhost:5000, user-agent=curl/7.64.1, accept=*/*}
Request body :  - 

2023-04-25 07:26:02 INFO  [c.n.d.common.filter.RequestLogFilter    .doFilter            :  36][http-nio-5000-exec-4] [REQUEST] [GET] /api/setting/enums
Headers : {host=localhost:5000, user-agent=curl/7.64.1, accept=*/*}
Request body :  - 

2023-04-25 07:26:02 INFO  [c.n.d.common.filter.RequestLogFilter    .doFilter            :  36][http-nio-5000-exec-3] [REQUEST] [GET] /api/setting/enums
Headers : {host=localhost:5000, user-agent=curl/7.64.1, accept=*/*}
Request body :  - 

2023-04-25 07:26:02 INFO  [c.n.d.common.filter.RequestLogFilter    .doFilter            :  36][http-nio-5000-exec-5] [REQUEST] [GET] /api/setting/enums
Headers : {host=localhost:5000, user-agent=curl/7.64.1, accept=*/*}
Request body :  - 

2023-04-25 07:26:02 INFO  [c.n.d.common.filter.RequestLogFilter    .doFilter            :  36][http-nio-5000-exec-1] [REQUEST] [GET] /api/setting/enums
Headers : {host=localhost:5000, user-agent=curl/7.64.1, accept=*/*}
Request body :  - 

---> 응답 다 받고

2023-04-25 07:26:12 INFO  [c.n.d.common.filter.RequestLogFilter    .doFilter            :  36][http-nio-5000-exec-1] [REQUEST] [GET] /api/setting/enums
Headers : {host=localhost:5000, user-agent=curl/7.64.1, accept=*/*}
Request body :  -

다른 방법:

postman > runner > iteration 을 이용한 run도 응답을 다 받을 때 까지 기다림

https://stackoverflow.com/questions/36157105/postman-how-to-make-multiple-requests-at-the-same-time

참고

브라우저로는 테스트가 힘들다..

탭으로 여러개 띄우면 안되고, 브라우저 자체를 여러개 해야하는데 동시 요청이 사실상 힘듦..

정 탭으로 해야하면 url 뒤에 fragment를 다르게 요청하면 된다고 한다.

• In Chrome, the calls get queued when you call the same resource from two tabs, but executed in parallel when you make the calls from different windows.
• In IE11, they always get executed in parallel
• In Firefox, it doesn't matter whether the calls are from different tabs or windows, they always get queued.

All of them execute them in parallel when the URLs are slightly different, by adding a different fragment or parameter.

서버 설정이 1이면 당연히 순차겠지..

https://stackoverflow.com/questions/19774193/subsequent-rest-call-is-blocked-until-previous-one-is-finished