為什么 docker-entrypoint-initdb.d 中的腳本在數據庫開始偵聽導致 pg_restore 失敗的連接之前運行？

Question

根據https://hub.docker.com/_/postgres的“初始化腳本”部分，我在 docker-entrypoint-initdb.d 中有兩個文件：

1. 初始化-db.sh
2. 備份轉儲

init-db.sh 包含：

#!/bin/sh
pg_restore -h localhost -p 5432 -U postgres -d postgres -v "/docker-entrypoint-initdb.d/backup.dump"
exit

它似乎正在執行，但出現服務器連接錯誤，因為服務器似乎尚未開始偵聽端口 5432，如下面的日志文件似乎表明的那樣。

我究竟做錯了什么？ 一旦 postgres docker 容器已初始化並開始運行，我基本上想恢復數據庫。

一旦容器運行，如果我在終端 session 中手動運行 init-db.sh 腳本，它將完全按照我的意願創建數據庫......

2019-10-23T17:09:43.032659100Z /usr/local/bin/docker-entrypoint.sh: ignoring /docker-entrypoint-initdb.d/backup.dump
2019-10-23T17:09:43.032699100Z
2019-10-23T17:09:43.032704200Z /usr/local/bin/docker-entrypoint.sh: running /docker-entrypoint-initdb.d/init-db.sh
2019-10-23T17:09:43.069092100Z pg_restore: connecting to database for restore
2019-10-23T17:09:43.069518400Z pg_restore: error: connection to database "postgres" failed: could not connect to server: Connection refused
2019-10-23T17:09:43.069534000Z  Is the server running on host "localhost" (127.0.0.1) and accepting
2019-10-23T17:09:43.069538000Z  TCP/IP connections on port 5432?
2019-10-23T17:09:43.069540800Z could not connect to server: Cannot assign requested address
2019-10-23T17:09:43.069543700Z  Is the server running on host "localhost" (::1) and accepting
2019-10-23T17:09:43.069546600Z  TCP/IP connections on port 5432?
2019-10-23T17:09:45.514371200Z 2019-10-23 17:09:45.514 UTC [1] LOG:  starting PostgreSQL 12.0 (Debian 12.0-2.pgdg100+1) on x86_64-pc-linux-gnu, compiled by gcc (Debian 8.3.0-6) 8.3.0, 64-bit
2019-10-23T17:09:45.514563000Z 2019-10-23 17:09:45.514 UTC [1] LOG:  listening on IPv4 address "0.0.0.0", port 5432
2019-10-23T17:09:45.514629800Z 2019-10-23 17:09:45.514 UTC [1] LOG:  listening on IPv6 address "::", port 5432
2019-10-23T17:09:45.522068300Z 2019-10-23 17:09:45.521 UTC [1] LOG:  listening on Unix socket "/var/run/postgresql/.s.PGSQL.5432"
2019-10-23T17:09:45.547131800Z 2019-10-23 17:09:45.547 UTC [23] LOG:  database system was interrupted; last known up at 2019-10-23 17:09:42 UTC
2019-10-23T17:09:45.977904500Z 2019-10-23 17:09:45.977 UTC [23] LOG:  database system was not properly shut down; automatic recovery in progress
2019-10-23T17:09:45.980523000Z 2019-10-23 17:09:45.980 UTC [23] LOG:  invalid record length at 0/16453B0: wanted 24, got 0
2019-10-23T17:09:45.980590000Z 2019-10-23 17:09:45.980 UTC [23] LOG:  redo is not required
2019-10-23T17:09:45.994889300Z 2019-10-23 17:09:45.994 UTC [1] LOG:  database system is ready to accept connections

docker-compose.yml：

version: '3.4'

networks:
  dev01:
    driver: bridge

services:
  webapplication2:
    image: ${DOCKER_REGISTRY-}webapplication2
    depends_on:
      - "pgdev01"
    build:
      context: .
      dockerfile: WebApplication2/Dockerfile
    networks:
     - dev01
  pgdev01:
    image: postgres 
    restart: always 
    volumes:
      - db_volume:/var/lib/postgresql/data
      - ./dbscripts/:/docker-entrypoint-initdb.d/
    networks:
      - dev01
volumes:
  db_volume:

docker-compose.override.yml：

version: '3.4'

services:
  webapplication2:
    environment:
      - ASPNETCORE_ENVIRONMENT=Development
      - ASPNETCORE_URLS=https://+:443;http://+:80
      - ASPNETCORE_HTTPS_PORT=44361
    ports:
      - "61225:80"
      - "44361:443"
    volumes:
      - ${APPDATA}/Microsoft/UserSecrets:/root/.microsoft/usersecrets:ro
      - ${APPDATA}/ASP.NET/Https:/root/.aspnet/https:ro
  pgdev01:
    ports:
      - "5434:5432" 
    environment:
      POSTGRES_PASSWORD: "redacted"

Answer 1

您遇到此問題是因為您在init-db.sh文件中包含了-h localhost 。 如果你刪除它，一切都會奏效。

發生這種情況是因為在postgres映像提供的默認docker-entrypoint.sh文件中，將在初始化階段清空listen_addresses 。 因此，它不會在localhost上監聽。

披露：我為Enterprisedb (EDB)工作

Answer 2

改變

pg_restore -h localhost -p 5432 -U postgres -d postgres -v "/docker-entrypoint-initdb.d/backup.dump"

至

pg_restore -U postgres -d postgres -v "/docker-entrypoint-initdb.d/backup.dump"

解決了這個問題。 我不知道為什么......這是一個瘋狂的猜測，基於顯示無法使用主機和端口連接到 postgres 的日志......