Harbor 开机自启遇到的问题
Harbor 是一个企业级的 Docker 私有仓库项目，使用 docker-compose 启动时，它本身由多个 Docker Containers 组成，通过 docker-compose 管理 containers 之间的依赖关系。
安装
安装参见官方文档：Harbor - Installation and Configuration Guide，
修改 harbor.yml 时，需要注意的有：
- 使用 https 时，要自己生成证书，然后在 harbor.yml 里配置好证书的路径。
- 客户端也需要安装你生成的这个证书，否则会报错。（所以证书要做好备份）
- 详见 Docker - Use self-signed certificates
- data_volume 可以指定为
/data/harbor，然后要定期备份这个文件夹。（或者直接在新机器上装个 harbor，用 harbor 自带的仓库复制功能做定期备份。） - 企业的话，可能还需要配置 ldap 集成验证。
然后 sudo ./install.sh 安装 harbor（貌似必须用 sudo，因为生成出来的配置文件的 owner 都是 root，而且权限设得很严格。）
安装完成后会自动启动 harbor.
自动启动的问题及解决方法
查看 harbor 目录下的 docker-compose.yml 会发现，所有的 containers 都配置了 restart: always：
version: '2.3'
services:
log:
image: goharbor/harbor-log:v1.8.1
container_name: harbor-log
restart: always
dns_search: .
# ......
registry:
image: goharbor/registry-photon:v2.7.1-patch-2819-v1.8.1
container_name: registry
restart: always
# ......
这表示所有的容器在意外关闭后都会自动重启，比如 docker 重启或服务器重启。（手动 stop 不会自动重启）
但是我在手动运行 docker-compose up -d，然后重启服务器后，发现有几个 container 并没有自动重启：
[ryan@ryan-pc ~]$ docker ps -a
CONTAINER ID IMAGE COMMAND CREATED STATUS PORTS NAMES
f30d802002a4 goharbor/nginx-photon:v1.8.1 "nginx -g 'daemon of…" 13 hours ago Exited (128) 27 minutes ago 0.0.0.0:80->80/tcp nginx
21472ce8a993 goharbor/harbor-portal:v1.8.1 "nginx -g 'daemon of…" 13 hours ago Exited (128) 27 minutes ago 80/tcp harbor-portal
5d866bb17c58 goharbor/harbor-jobservice:v1.8.1 "/harbor/start.sh" 13 hours ago Exited (137) 26 minutes ago harbor-jobservice
0cf0f93b5a87 goharbor/harbor-core:v1.8.1 "/harbor/start.sh" 13 hours ago Up 11 seconds (health: starting) harbor-core
cba280d9b945 goharbor/redis-photon:v1.8.1 "docker-entrypoint.s…" 13 hours ago Exited (137) 26 minutes ago 6379/tcp redis
473e46d1f746 goharbor/harbor-registryctl:v1.8.1 "/harbor/start.sh" 13 hours ago Up 11 seconds (health: starting) registryctl
51f105f1691d goharbor/registry-photon:v2.7.1-patch-2819-v1.8.1 "/entrypoint.sh /etc…" 13 hours ago Exited (137) 26 minutes ago 5000/tcp registry
c41594ec7779 goharbor/harbor-db:v1.8.1 "/entrypoint.sh post…" 13 hours ago Up 11 seconds (health: starting) 5432/tcp harbor-db
713bd4961772 goharbor/harbor-log:v1.8.1 "/bin/sh -c /usr/loc…" 13 hours ago Up 11 seconds (health: starting) 127.0.0.1:1514->10514/tcp harbor-log
可以看到下列五个容器都处于 Exited 状态：
goharbor/nginx-photon:v1.8.1goharbor/harbor-portal:v1.8.1goharbor/harbor-jobservice:v1.8.1goharbor/redis-photon:v1.8.1goharbor/registry-photon:v2.7.1-patch-2819-v1.8.1
搜索发现有人提过这个 issue: https://github.com/goharbor/harbor/issues/7008
这里有人提到，问题在于容器只有在 docker-compose up 时，才会按照 depends_on 定义的顺序启动。
docker 本身并不记录容器之间的依赖顺序，容器们的重启是相互独立的，各自独立的重启导致服务器重启后，harbor 无法正常启动。
issue 中也有大佬给出了基于 systemd 的解决方案。
我于是尝试将 harbor 配成 systemd 的 service，添加配置文件 /usr/lib/systemd/system/harbor.service，内容如下：
[Unit]
Description=Harbor
After=docker.service systemd-networkd.service systemd-resolved.service
Requires=docker.service
Documentation=http://github.com/vmware/harbor
[Service]
Type=simple
Restart=on-failure
RestartSec=5
ExecStart=/usr/local/bin/docker-compose -f {{ harbor_install_path }}/harbor/docker-compose.yml up
ExecStop=/usr/local/bin/docker-compose -f {{ harbor_install_path }}/harbor/docker-compose.yml down
[Install]
WantedBy=multi-user.target
其中 {{ harbor_install_path }} 换成自己的 harbor 安装路径。
还有 docker-compose 的绝对路径，请通过 which docker-compose 查看。
然后启动该项服务：
sudo systemctl enable harbor
sudo systemctl start harbor
现在查看下 harbor.service 的情况：
● harbor.service - Harbor
Loaded: loaded (/usr/lib/systemd/system/harbor.service; enabled; vendor preset: disabled)
Active: active (running) since Sun 2019-07-07 12:54:27 CST; 2min 4s ago
Docs: http://github.com/vmware/harbor
Main PID: 9734 (docker-compose)
Tasks: 11 (limit: 4915)
Memory: 35.7M
CGroup: /system.slice/harbor.service
└─9734 /usr/bin/python /usr/bin/docker-compose -f /home/ryan/harbor/docker-compose.yml up
7月 07 12:54:30 ryan-pc docker-compose[9734]: [128B blob data]
7月 07 12:54:31 ryan-pc docker-compose[9734]: [167B blob data]
7月 07 12:54:31 ryan-pc docker-compose[9734]: registryctl | WARNING: no logs are available with the 'syslog' log driver
7月 07 12:54:31 ryan-pc docker-compose[9734]: registry | WARNING: no logs are available with the 'syslog' log driver
7月 07 12:54:31 ryan-pc docker-compose[9734]: harbor-db | WARNING: no logs are available with the 'syslog' log driver
7月 07 12:54:31 ryan-pc docker-compose[9734]: redis | WARNING: no logs are available with the 'syslog' log driver
7月 07 12:54:31 ryan-pc docker-compose[9734]: harbor-core | WARNING: no logs are available with the 'syslog' log driver
7月 07 12:54:31 ryan-pc docker-compose[9734]: harbor-jobservice | WARNING: no logs are available with the 'syslog' log driver
7月 07 12:54:31 ryan-pc docker-compose[9734]: harbor-portal | WARNING: no logs are available with the 'syslog' log driver
7月 07 12:54:31 ryan-pc docker-compose[9734]: nginx | WARNING: no logs are available with the 'syslog' log driver
重启服务器再确认，发现 harbor 容器组终于全部 up 了：
[ryan@ryan-pc ~]$ docker ps -a
CONTAINER ID IMAGE COMMAND CREATED STATUS PORTS NAMES
90943210e354 goharbor/nginx-photon:v1.8.1 "nginx -g 'daemon of…" 37 seconds ago Up 35 seconds (healthy) 0.0.0.0:80->80/tcp nginx
bc70706d7b5f goharbor/harbor-portal:v1.8.1 "nginx -g 'daemon of…" 38 seconds ago Up 36 seconds (healthy) 80/tcp harbor-portal
54e132bd1a10 goharbor/harbor-jobservice:v1.8.1 "/harbor/start.sh" 38 seconds ago Up 36 seconds harbor-jobservice
28ada4a941ee goharbor/harbor-core:v1.8.1 "/harbor/start.sh" 39 seconds ago Up 37 seconds (healthy) harbor-core
de44686f7e72 goharbor/registry-photon:v2.7.1-patch-2819-v1.8.1 "/entrypoint.sh /etc…" 39 seconds ago Up 38 seconds (healthy) 5000/tcp registry
e5efdae34c13 goharbor/harbor-db:v1.8.1 "/entrypoint.sh post…" 39 seconds ago Up 38 seconds (healthy) 5432/tcp harbor-db
a5fe52e1f184 goharbor/harbor-registryctl:v1.8.1 "/harbor/start.sh" 39 seconds ago Up 38 seconds (healthy) registryctl
4a74b5f3499a goharbor/redis-photon:v1.8.1 "docker-entrypoint.s…" 39 seconds ago Up 38 seconds 6379/tcp redis
677fd84d0a70 goharbor/harbor-log:v1.8.1 "/bin/sh -c /usr/loc…" 40 seconds ago Up 39 seconds (healthy) 127.0.0.1:1514->10514/tcp harbor-log
