Backups

From Meta
This page contains changes which are not marked for translation. Please make a request at Meta:Administrators' noticeboard or contact any translation administrator or administrator via their user talk page to mark the updated page for translation.

Miraheze has an internal and external backup procedure. Following the schedule below, certain critical parts of our infrastructure are automatically backed up to an external server (external meaning under our control but provided by a different server host and in a different country than all our current servers). These 'internal' backups include full database dumps, which include user account information and CheckUser information. These backups are only accessible by our Site Reliability Engineering team, and can be used in the event of a catastrophic site failure to quickly bring the entire site up to a recent state.

Backups of a wiki

In addition to these private backups, any wiki administrator can create an XML or image backup of their wiki by going to Special:DataDump on their wiki and selecting XML or image type. These backups can then be stored securely wherever you like. XML backups do not include user account information or CheckUser information but contain wiki page text and logs that you can import to any MediaWiki site. Image dumps contain all file types uploaded to the wiki, but without descriptions or licensing type.

General backup Schedules

We are working on introducing a new permanent backup solution in the future. In the meantime, XML backups of public wikis are regularly uploaded to archive.org and private wiki XML dumps will be kept on servers. (last updated 29 November 2022)

This follow is OUDATED and no longer currently applies Miraheze runs two backup schedules in production:

  • The following are backed up in their entirety every Sunday:
    • Our private Git repository (stored on puppet1)
      • This includes configured passwords, private keys, and certificates for our domains
      • This includes the original source of private keys and certificates, and includes our account information for Let's Encrypt (the CA we use for free certificates)
  • The following are backed up completely on the first Sunday of every month. Changed files and new files are backed up on the third Sunday of the month:
    • Databases, including user information, for all wikis
    • Our static content (wiki images, user XML dumps)
    • Phabricator static (content used by our tracking software)


Local backups

A fast and simple method to backup Miraheze wikis is Elsie Hupp's Python 3 fork of the WikiTeam dumpgenerator.py Python 2 script.

Note: This doesn't yet work for every Miraheze wiki, some debugging is still needed.

Example usage, this will produce an XML dump, a folder of images (and other files) and a JSON file:
dumpgenerator.py --xmlrevisions --xml --images --api=https://yourwiki.miraheze.org/w/api.php
or on Windows:
dumpgenerator --xmlrevisions --xml --images --api=https://yourwiki.miraheze.org/w/api.php

The presence of a siteinfo.json file probably indicates a succesful XML dump.

Full instructions are at elsiehupp/wikiteam3.

See also

For more technical details on our automatic backup server, see Tech:Bacula.