ThingsBoard——Docker重启失败，报错Connection to localhost:5432 refused的解决方法

Posted 2022-03-07 a10615

一、问题

最近在写自定义规则节点，因为还没编译好thingsboard源代码，用的是docker搭建起来的环境。写好的规则节点后，要打包好扔到docker里，再重启docker。发现经常重启失败，报错的日志也都是这样：

2022-03-05 08:53:23,164 [main] ERROR com.zaxxer.hikari.pool.HikariPool - HikariPool-1 - Exception during pool initialization.
org.postgresql.util.PSQLException: Connection to localhost:5432 refused. Check that the hostname and port are correct and that the postmaster is accepting TCP/IP connections.
	at org.postgresql.core.v3.ConnectionFactoryImpl.openConnectionImpl(ConnectionFactoryImpl.java:303)
...

2022-03-05 08:53:25,187 [main] ERROR o.s.boot.SpringApplication - Application run failed
org.springframework.context.ApplicationContextException: Unable to start web server; nested exception is org.springframework.boot.web.server.WebServerException: Unable to start embedded Tomcat
	at org.springframework.boot.web.servlet.context.ServletWebServerApplicationContext.onRefresh(ServletWebServerApplicationContext.java:161)
...

pg_ctl: could not open PID file "/data/db/postmaster.pid": Operation not permitted

二、原因

试过删除后重新run，也试过重新赋权限（好像第一次能成功），试过用docker-compose，结果都是一样。最后还是在issues里找到了解决方案，感谢alyf80提供的思路：

I think this is due to postgres not being properly shut down when the container is stopped; this leaves a stale postmaster.pid file in the data/db directory, which on the next run causes pg_ctl to try talking to a non-existing daemon and ultimately failing to start the db server. Deleting that file prior to starting the container reliably fixes the problem in my case.

I have no idea why this is happening. It looks to me that the TB process is ignoring (or simply not getting) the SIGTERM that Docker sends to initiate a graceful shutdown of the container, so after the default 10s shutdown timeout expires everything gets killed without stop-db.sh having had a chance to run.

大致意思：PostgreSQL在docker关闭时，可能没有正常接收到关闭指令，所以数据库进程一直在使用中，没有被释放，导致docker启动后无法连接上数据库

补充：应该是大多数时机都没关闭，偶尔才正常关闭，所以重试几次后，会有成功的

三、解决

既然是进程未释放原因，那就好办了，就在重启前，先把db给结束掉。这里提供多种方法（扩展思路，试过都OK）

# 【荐】法1：安全关闭容器里的数据库，也是alyf80所提到的
$ docker exec -it mytb stop-db.sh

# 法2：强制关闭
$ docker exec -it mytb killall postgres 

# 法3：强制关闭，进入容器内操作
$ docker exec -it mytb bash
# 或先用ps aux找到postgres的PID，然后kill -9 [PID]
$ killall postgres

根据这个原因分析，应该可以推断出修改权限后，为什么第一次成功了。修改权限后，db停止了，所以重启后也就正常了