Cost a lot cpu after 24h

[Ylarod]

Hello, I use docker to host opengrok, the docker-compose file is like below.

version: "3"

services:
  aosp_openresty:
    container_name: aosp_openresty
    image: openresty/openresty:alpine
    profiles:
       - service
    ports:
      - 8080:80
    volumes:
      - ./conf.d:/etc/nginx/conf.d
      - ./html:/usr/local/openresty/nginx/html
    restart: unless-stopped
    networks: 
      vpn:
        ipv4_address: 172.168.22.99

  aosp_tomcat_sdk21:
    container_name: aosp_tomcat_sdk21
    image: tomcat:10.1-jdk11
    profiles:
       - service
    volumes:
      - '/data/aospxref/webapps/android-5.0.0_r1:/usr/local/tomcat/webapps'
      - '/data/aospxref/src/android-5.0.0_r1:/opengrok/src/'
      - '/data/aospxref/etc/android-5.0.0_r1:/opengrok/etc/'
      - '/data/aospxref/data/android-5.0.0_r1:/opengrok/data/'
    restart: unless-stopped
    networks: 
      vpn:
        ipv4_address: 172.168.22.121

  aosp_opengrok_sdk21:
    container_name: aosp_opengrok_sdk21
    image: opengrok/docker:latest
    environment:
      SYNC_PERIOD_MINUTES: '0'
      URL_ROOT: '/android-5.0.0_r1'
    profiles:
       - init
    volumes:
       - '/data/aospxref/webapps/android-5.0.0_r1:/usr/local/tomcat/webapps'
       - '/data/aospxref/src/android-5.0.0_r1:/opengrok/src/'
       - '/data/aospxref/etc/android-5.0.0_r1:/opengrok/etc/'
       - '/data/aospxref/data/android-5.0.0_r1:/opengrok/data/'
    restart: unless-stopped
    networks: 
      vpn:
        ipv4_address: 172.168.22.221

I use opengrok to generate index and use a standalone tomcat image to provide service cause it's more stable.
But every day, I need to restart the container because it costs a lot of cpu and memory(before I limited it to 2GB)

I don't know why this occured, any help?

Thanks in advance.

After restart the container and wait for 1 minute (compile):

top

[vladak]

I am pretty sure this is the Suggester trying to reinitialize the suggester data. First thing I'd do is to look in the webapp logs for any clues. If this is indeed the suggester, then you have 2 options: either tune the webapp appropriately or turn the suggester off. If there is not enough heap in the webapp (which the suggester needs some good amount based on the input data, definitely for stuff like AOSP), the JVM might spend inordinate amounts of time in the Garbage Collector and there will be very little progress.