Deployment Recipes - fieldenms/tg GitHub Wiki

Note that much of this is now irrelevant since deployment under Docker has become more widespread.

1. Abstract
2. Linux server configuration - firewall
3. Linux server configuration - systemd
4. Deployment to Linux
5. References
6. TODO

---

This document provides some brief notes that may be useful for the deployment of TG-based applications, and were initially created whilst deploying the TridentFleet pilot.

---

This is suitable for CentOS 7.1.

This only needs to be done once per server - provided the changes are made permanent they will take effect each time the server is rebooted.

1. Login as, or become, root:

   $ su -
   

2. Identify the active firewall zones:

   # firewall-cmd --get-active-zones
   

   For example:

   # firewall-cmd --get-active-zones
   public
     interfaces: eth0
   

3. Determine if the port is already open:

   # firewall-cmd --zone=public --query-port=<port>/<protocol>
   

   Or if it is a known service:

   # firewall-cmd --zone=public --query-service=<service>
   

   For example:

   # firewall-cmd --zone=public --query-port=80/tcp
   no
   # firewall-cmd --zone=public --query-service=http
   no
   

4. Add the specific port:

   # firewall-cmd --permanent --zone=public --add-port=<port>/<protocol>
   

   Or if it is a known service:

   # firewall-cmd --permanent --zone=public --add-service=<service>
   

5. Reload the firewall to take effect:

   # firewall-cmd --reload
   

---

This is suitable for CentOS 7.1, but should also work for any other system using systemd.

This only needs to be done once per server - provided the service is appropriately configured it will start automatically each time the server is rebooted.

1. Login as, or become, root:

   $ su -
   

2. Create a user <username>, which will be used to run the service:

   # useradd -m <username>
   

   Note that the -m option will create the user's login directory under /home. To create a user with a specific login directory:

   # useradd -d /path/to/dir <username>
   

3. Create the file /etc/systemd/system/<name>.service with content like:

   [Unit]
   Description=<name> Service
   After=network.target
   
   [Service]
   Type=simple
   User=<username>
   ExecStart=/path/to/executable --options
   Restart=on-abort
   
   [Install]
   WantedBy=multi-user.target
   

   After=network.target - specifies that the service should start after the network is available.

   Type=simple - other options are forking, oneshot, dbus, notify or idle.

   ExecStart=... - specifies the executable file or script that should be used to start the service. This does not set the working directory, so it is suggested to use a shell script that initially sets the working directory, then starts the application as required.

   Restart=on-abort - specifies that the service should be automatically restarted should it terminate uncleanly. Other options are no, on-success, on-failure, on-abnormal, on-watchdog, or always. no is probably the only other suggested option, which does not restart the service should it terminate for any reason.

   WantedBy=multi-user.target - starts the service when the system boots into multi-user mode (i.e. boots normally).

4. Configure the service to start at boot:

   # systemctl enable <name>
   

5. Start the service immediately:

   # systemctl start <name>
   

6. Allow the service sufficient time to start, and check its status:

   # systemctl status <name>
   

---

This is suitable for CentOS 7.1.

1. Receive a zip file for deployment.

2. Transfer the zip file to the server.

3. As root, stop the service:

   # systemctl stop <name>
   

4. Rename or archive the old version.

5. Extract the new version into the same directory or, if extracted into a different directory, ensure that the startup script is altered accordingly.

   This document does not cover changing the systemd service configuration to cater for placement of the startup script in a different location.

6. As root, start the service:

   # systemctl start <name>
   

7. Allow the service sufficient time to start, and check its status:

   # systemctl status <name>
   

---

• systemd

---

• Nothing.