RopeWiki server

Introduction

This repository contains the information necessary to deploy the RopeWiki technical infrastructure (though it does not contain the database content nor the images folder content of the real site).

Site architecture

• The MySQL database stores all the non-file site data
• The webserver runs MediaWiki with extensions, and also stores all the file-based site data/uploads
  • Exposes port 8080 for direct access to the webserver that bypasses the reverse proxy
• The reverse proxy manages TLS termination and also redirects according to target site
  • Exposes ports 80 & 443 as the external-facing webserver
• The backup manager exposes site data for backup
  • Exposes port 22001 for SSH access to backupreader users

Site deployment

Execute the steps below to produce a server running RopeWiki. The instructions assume an Ubuntu machine. On Windows, the easiest option is probably to install VirtualBox and host a virtual Ubuntu system. Alternately, all of the steps after the firewall (which can be skipped) should be possible directly in a Windows command prompt as long as Python 3 is installed and added to the PATH (python3 --version to verify). Ignore all apt commands and instead perform the Windows alternative.

Protect the system

1. Setup firewall from scratch
   1. Block everything incoming (sudo ufw default deny incoming)
   2. Allow everything outgoing (sudo ufw default allow outgoing)
   3. Allow SSH to host machine (sudo ufw allow OpenSSH)
   4. Allow web server requests (sudo ufw allow 80/tcp && ufw allow 443/tcp)
   5. Allow SSH directly to containers (sudo ufw allow 22001/tcp && ufw allow 22002/tcp)
   6. Optionally, allow debugging: (sudo ufw allow 8080/tcp)
2. Alternately, setup firewall by importing rules
   1. Copy /etc/ufw/user.rules
   2. Copy /etc/ufw/user6.rules
3. Enable firewall (sudo ufw enable)

Install necessary tools

1. Update packages (sudo apt-get update)
2. Install docker
   1. Fix this issue with sudo apt update && apt install rng-tools (note: this may no longer be necessary with docker compose v2)
   2. Verify installation with docker compose version; result should be 2.x.x
3. Install git (sudo apt-get install git)
4. Clone this repository into an appropriate folder (perhaps /rw)

Define site to be deployed

1. Ensure that there is a .json file in site_configs corresponding to the site to be deployed
2. Create a new .json file, modeled after dev.json, if necessary
   1. Create a folder that will hold persistent mount data (perhaps /rw/mount) and define folders relative to that folder
3. SITE_NAME is the name of the .json file without extension (e.g., example.json implies a SITE_NAME of example)

Transfer site data

1. Get latest SQL backup
   1. If transferring from an old server, run python3 deploy_tool.py <SITE_NAME> get_sql_backup_legacy
2. Get images folder
   1. If transferring from an old server, run python3 deploy_tool.py <SITE_NAME> get_images_legacy
3. Create a shortcut to declare passwords: create ~/rw_passwords.sh (or another location) with content like:

   #!/bin/bash
   # These are the manditory environment variables needed to start a copy of the site
   export WG_DB_PASSWORD=<The password for the `ropewiki` DB user>
   export RW_ROOT_DB_PASSWORD=<The password for the `root` DB user>
   export RW_SMTP_USERNAME=<The username for logging into the smtp relay>
   export RW_SMTP_PASSWORD=<The password for logging into the smtp relay>

Deploy site

1. Build the necessary images: python3 deploy_tool.py <SITE_NAME> dc build
2. Make password variables accessible in the terminal: source ~/rw_passwords.sh
3. Create an empty database using python3 deploy_tool.py <SITE_NAME> create_db
4. Restore content into database using python3 deploy_tool.py <SITE_NAME> restore_db
   1. Or, for a development deployment, create a minimal database using python3 deploy_tool.py <SITE_NAME> restore_empty_db
5. Bring site up with python3 deploy_tool.py <SITE_NAME> start_site
6. (Optional) Confirm that the webserver container is working, apart from the reverse proxy, by visiting http://<hostname>:8080
7. Confirm that the site is working via HTTP by visiting http://<hostname>
8. Enable TLS with python3 deploy_tool.py <SITE_NAME> enable_tls
   1. Note that the certs should be persisted in ${proxy_config_folder}/letsencrypt; select option 1 to reinstall the existing cert if prompted
   2. Enable redirection (option 2) when prompted
   3. Verify success by visiting https://<hostname>
   4. Create cronjob to automatically update certificates
      1. From this working directory, run python3 deploy_tool.py <SITE_NAME> add_cert_cronjob
      2. To edit or delete crontabs, crontab -e

Backups

Direct SSH access is provided to the backup manager at port 22001 for the backupreader user for clients who possess the private key to any of the public keys listed in authorized_keys.

Database

In the backup manager, the backupreader's home directory has a backups folder where complete backups of the database will be created daily and named all-backup-YYYY-MM-DD-HHMMSS.tar.gz. An off-site backup client should connect to this container and copy the latest all-backup file to back up the database.

Images

In the backup manager, the backupreader's home directory has a symlink to the images folder which contains most of the file-based data uploaded to the site. An off-site backup client should connect to this container and synchronize the full content of the images folder to back them up.

Site maintenance

Refreshing TLS certificates manually

This should be performed by a cron job, but in the event of needing to do it manually, run python3 deploy_tool.py <SITE_NAME> renew_certs

Arbitrary docker compose commands

The docker-compose.yaml configuration requires a number of environment variables to be set before it can be used. To avoid the need to set these variables yourself (apart from WG_DB_PASSWORD and RW_ROOT_DB_PASSWORD), use python3 deploy_tool.py <SITE_NAME> dc "<YOUR COMMAND>". For instance, python3 deploy_tool.py dev dc "up -d".

Updating webserver

To deploy changes to the webserver Dockerfile: python3 deploy_tool.py redeploy webserver

OR, manually:

1. Build the Dockerfile (docker image build -t ropewiki/webserver . from the root of this repo)
2. Kill and remove the webserver (python3 deploy_tool.py <SITE_NAME> dc "rm -f -s -v ropewiki_webserver")
3. Restore the full deployment (python3 deploy_tool.py <SITE_NAME> start_site)

Troubleshooting

The site can't be reached after updating the webserver

Reset the containers and redeploy:

1. Tear the site down (python3 deploy_tool.py <SITE_NAME> dc "down -v")
2. Bring the site back up (python3 deploy_tool.py <SITE_NAME> start_site)
3. Re-enable TLS (python3 deploy_tool.py <SITE_NAME> enable_tls then run the specified script, choosing to reinstall the certificate)

Sorry! This site is experiencing technical difficulties.

If this is accompanied by "(Cannot contact the database server)", it means the MediaWiki app (the ropewiki_webserver container) is not configured properly to contact the database (the ropewiki_db container). The most likely problem is that you have not specified the WG_DB_PASSWORD environment variable to match the one in the database backup you restored. WG_DB_PASSWORD should be specified to match the password used in the database you restored; see instructions above.

To verify whether a connection with a particular username and password can be established, open a terminal in the database container: docker container exec -it dev_ropewiki_db_1 /bin/bash (but with an appropriate container; python3 deploy_tool.py <SITE_NAME> dc ps to list containers). Then, attempt to connect to the database with mysql -u <USERNAME> -p<PASSWORD. If successful, check users with select host, user, password from mysql.user;.

If the above is successful, verify that the connection can be made from the webserver container by opening a terminal in the webserver container via a similar process as above. Add the hostname to the mysql command like: mysql -h ropewiki_db -u <USERNAME> -p<PASSWORD>.

Running a local development instance

These instructions can be used to run a local development instance with just a few adjustments. Simply make sure there is a site_config appropriate to your local machine, and otherwise follow the instructions above normally. The site_config "local" is excluded from git tracking, so it is an ideal place to define a system configuration that other people are unlikely to use. However, if your local development instance setup is likely to be reusable by others, feel free to add it to site_configs; a local_windows.json site_config would probably be helpful to others, for instance.

Note that, in all cases, the following resources (not included in this repo) are necessary to bring up a functional site:

• A backup of the site database in .sql format (1.2+ GB)
• The images folder of the site (18.3+ GB)

TLS is not necessary, and sometimes not feasible, on a local development instance. In that case, simply don't enable TLS and instead access the site with http. If the site_config hostname is localhost, then port 80 will be used and the site should be accessible at http://localhost. To use a different port, specify, e.g., a hostname of localhost:8081 making the site available at http://localhost:8081. Do not use port 8080 because it is already used to provide debug access directly to the webserver without going through the reverse_proxy.

Exploring the system

_The commands below assume the use of the local SITE_NAME/siteconfig; change the commands to reflect your site configuration name if necessary.

• Print webserver stdout + stderr (there should not be much here)
  • python3 deploy_tool.py local dc logs ropewiki_webserver
• Run an interactive shell inside the webserver
  • Determine the name of the webserver container with python3 deploy_tool.py local dc ps
  • docker container exec -it local_ropewiki_webserver_1 /bin/bash (but substitute your webserver container name)
  • Print normal access log
    • cat /var/log/nginx/access.log
  • Print error log
    • cat /var/log/nginx/error.log
• Delete the system
  • python3 deploy_tool.py local dc down
    • Note that this will disconnect the docker volume automatically created for the mysql container from any container (leaving it dangling). If you have restored the database, this will leave a very large volume dangling
    • See your dangling volumes and their space
      • docker system df -v
    • Delete your dangling volumes to free disk space
      • docker volume rm `docker volume ls -q -f dangling=true`