Docker安装PostgreSQL重启后/run/postgresql/*丢失问题
创建Docker并安装PostgreSQL
#创建docker
docker run --privileged -dit --name test --net staticnet --ip 192.168.2.6 centos-ssh /usr/sbin/init
#安装postgressql11.2
yum install -y https://download.postgresql.org/pub/repos/yum/11/redhat/rhel-7-x86_64/pgdg-centos11-11-2.noarch.rpm
yum install -y postgresql11 postgresql11-server
#初始化data
/usr/pgsql-11/bin/postgresql-11-setup initdb
#设置开机自启并启动
systemctl enable postgresql-11
systemctl start postgresql-11
查看/run/postgresql
[root@06b0e1e15b1b run]# pwd
/run
[root@06b0e1e15b1b run]# ll
total 16
-rw-r--r-- 1 root root 3 Oct 17 17:06 atd.pid
drwxr-xr-x 2 root root 40 Oct 17 17:11 console
-rw-r--r-- 1 root root 3 Oct 17 17:06 crond.pid
---------- 1 root root 0 Oct 17 17:06 cron.reboot
drwx------ 2 root root 40 Oct 17 17:11 cryptsetup
drwxr-xr-x 2 root root 60 Oct 17 17:06 dbus
drwxr-xr-x 2 root root 40 Oct 17 17:11 faillock
drwxr-xr-x 4 root root 80 Oct 17 17:11 lock
drwxr-xr-x 2 root root 40 Oct 17 17:11 log
drwxr-xr-x 2 root root 40 Oct 17 17:06 mount
drwxrwxr-x 2 root root 40 Oct 17 17:11 netreport
drwxr-xr-x 2 postgres postgres 40 Oct 17 17:17 postgresql
drwxr-xr-x 2 root root 6 Oct 17 17:06 secrets
drwxr-xr-x 2 root root 40 Oct 17 17:11 sepermit
drwxr-xr-x 2 root root 40 Oct 17 17:11 setrans
-rw-r--r-- 1 root root 3 Oct 17 17:06 sshd.pid
drwx--x--x 3 root root 60 Oct 17 17:11 sudo
drwx------ 2 root root 40 Oct 17 17:11 svnserve
-rw------- 1 root root 2 Oct 17 17:06 syslogd.pid
drwxr-xr-x 15 root root 400 Oct 17 17:15 systemd
drwxr-xr-x 2 root root 40 Oct 17 17:11 user
[root@06b0e1e15b1b run]# ll postgresql/ -a
total 4
drwxr-xr-x 2 postgres postgres 80 Oct 17 17:31 .
drwxr-xr-x 18 root root 460 Oct 17 17:17 ..
srwxrwxrwx 1 postgres postgres 0 Oct 17 17:31 .s.PGSQL.5432
-rw------- 1 postgres postgres 63 Oct 17 17:31 .s.PGSQL.5432.lock
可以看到/run/postgresql中包含.s.PGSQL.5432和.s.PGSQL.5432.lock
重启docker容器
docker restart test
进入容器查看
[root@Server]# docker exec -ti test /bin/bash
[root@06b0e1e15b1b /]# cd /run/
[root@06b0e1e15b1b run]# ll
total 16
-rw-r--r-- 1 root root 3 Oct 17 17:34 atd.pid
-rw-r--r-- 1 root root 3 Oct 17 17:34 crond.pid
---------- 1 root root 0 Oct 17 17:34 cron.reboot
drwxr-xr-x 2 root root 60 Oct 17 17:34 dbus
drwxr-xr-x 2 root root 40 Oct 17 17:34 lock
drwxr-xr-x 2 root root 40 Oct 17 17:34 mount
drwxr-xr-x 2 root root 6 Oct 17 17:34 secrets
-rw-r--r-- 1 root root 3 Oct 17 17:34 sshd.pid
-rw------- 1 root root 2 Oct 17 17:34 syslogd.pid
drwxr-xr-x 8 root root 260 Oct 17 17:34 systemd
重启后会发现/run/postgresql丢失了
问题分析：
执行df
[root@06b0e1e15b1b run]# df
Filesystem 1K-blocks Used Available Use% Mounted on
overlay 41931776 12979836 28951940 31% /
tmpfs 4005220 0 4005220 0% /dev
/dev/vda1 41931776 12979836 28951940 31% /etc/hosts
shm 65536 0 65536 0% /dev/shm
tmpfs 65536 24 65512 1% /run
tmpfs 65536 0 65536 0% /run/lock
tmpfs 65536 0 65536 0% /var/log/journal
tmpfs 4005220 0 4005220 0% /tmp
发现第7行，/run使用的是tmpfs文件系统，简单来说就是一个基于内存的文件系统，断电重启后系统中的文件不会保存。详见：tmpfs
解决方法：
1.可以手动创建该目录并修改权限
mkdir /run/postgresql
chown -R postgres:postgres postgresql/
2.可以通过systemd-tmpfiles来创建
systemd-tmpfiles --create /usr/lib/tmpfiles.d/*
因为/run目录的这个特殊性，CentOS7特意在/usr/lib/tmpfiles.d/下提供了可以自定义.conf的方式来解决这个问题。详见：tmpfiles.d
[root@06b0e1e15b1b run]# cat /usr/lib/tmpfiles.d/postgresql-11.conf
d /run/postgresql 0755 postgres postgres -
PG为我们提供了这个配置文件，这行的意思就是有/run/postgresql就直接设置权限，没有就创建后再设置权限，这个指令在安装PG的时候会执行，因为安装时一般都在root权限下，而执行systemctl start postgresql-11时实际是用/usr/pgsql-11/bin/postmaster -D ${PGDATA}来启动，且使用postgres用户组和用户，导致无法创建该文件夹。而且执行systemctl start postgresql-11看不到任何报错，PG的log中也没有错误日志。只有再切换到postgres用户下执行pg_ctl start才会看到错误，报了Permission denied。
bash-4.2$ /usr/pgsql-11/bin/pg_ctl start -D /var/lib/pgsql/11/data
waiting for server to start....2019-10-16 21:21:33.801 CST [277] LOG: number of prepared transactions has not been configured, overriding
2019-10-16 21:21:33.801 CST [277] DETAIL: max_prepared_transactions is now set to 200
2019-10-16 21:21:33.802 CST [277] LOG: listening on IPv4 address "0.0.0.0", port 5432
2019-10-16 21:21:33.802 CST [277] LOG: listening on IPv6 address "::", port 5432
2019-10-16 21:21:33.805 CST [277] FATAL: could not create lock file "/var/run/postgresql/.s.PGSQL.5432.lock": Permission denied
2019-10-16 21:21:33.805 CST [277] LOG: database system is shut down
stopped waiting
pg_ctl: could not start server
Examine the log output.
3.修复systemd-tmpfiles服务
systemd-tmpfiles服务是可以开机自启的，为什么在docker中就是有问题的，不能执行呢？
[root@06b0e1e15b1b run]# cat /usr/lib/systemd/system/systemd-tmpfiles-setup.service
# This file is part of systemd.
#
# systemd is free software; you can redistribute it and/or modify it
# under the terms of the GNU Lesser General Public License as published by
# the Free Software Foundation; either version 2.1 of the License, or
# (at your option) any later version.
[Unit]
Description=Create Volatile Files and Directories
Documentation=man:tmpfiles.d(5) man:systemd-tmpfiles(8)
DefaultDependencies=no
Conflicts=shutdown.target
After=systemd-readahead-collect.service systemd-readahead-replay.service local-fs.target systemd-sysusers.service
Before=sysinit.target shutdown.target
RefuseManualStop=yes
[Service]
Type=oneshot
RemainAfterExit=yes
ExecStart=/usr/bin/systemd-tmpfiles --create --remove --boot --exclude-prefix=/dev
可以看到systemd-tmpfiles-setup.service中的ExecStart正是我们要执行的命令，通过执行systemctl start systemd-tmpfiles-setup.service看下结果
[root@06b0e1e15b1b systemd]# systemctl start systemd-tmpfiles-setup.service
Failed to start systemd-tmpfiles-setup.service: Unit is masked.
[root@06b0e1e15b1b systemd]# systemctl unmask systemd-tmpfiles-setup.service
Removed symlink /etc/systemd/system/systemd-tmpfiles-setup.service.
[root@06b0e1e15b1b systemd]# systemctl start systemd-tmpfiles-setup.service
会发现报Unit is masked，执行ummask后发现服务可以正常start，再次重起发现问题解决。