Actor migration
Historically, MediaWiki has stored references to users (such as the author of a revision or an image) as an [<id>, <name>]
pair, with [0, <IP address>]
for anonymous edits.
This was a waste of storage space, caused performance issues on wikis with a huge number of revisions, and made it very hard to rename users or introduce new types of authors such as T133452.
MediaWiki 1.31 introduces the concept of actors (someone who makes an action, such as an edit or a log event; currently either a registered user or an anonymous one), and a migration process towards a new database schema, with fields like revision.rev_user
and revision.rev_user_text
replaced by fields like revision.rev_actor
.
Due to the huge size of the English Wikipedia revision
table, which had be fully migrated before the changes can be cleaned up, the migration was done gradually, over several MediaWiki releases.
It finished in MediaWiki 1.39.
The migration created a actor table record for every distinct *_user
/ *_user_text
pair, and replaced those pairs with a reference to the actor row ID.
The following columns were migrated:
Table | Old columns | New column |
---|---|---|
revision | rev_user, rev_user_text | rev_actor |
archive | ar_user, ar_user_text | ar_actor |
ipblocks | ipb_by, ipb_by_text | ipb_by_actor |
image | img_user, img_user_text | img_actor |
oldimage | oi_user, oi_user_text | oi_actor |
filearchive | fa_user, fa_user_text | fa_actor |
recentchanges | rc_user, rc_user_text | rc_actor |
logging | log_user, log_user_text | log_actor |
To avoid a costly ALTER TABLE on the extremely huge revision
tables of large Wikipedia wikis, rev_user
and rev_user_text
were initially copied to the revision_actor_temp table and from there back to revision
.
The revision_actor_temp
table was dropped afterwards.
If you are a wiki owner, you probably don't need to do anything; the migration will happen automatically when you run update.php
.
If you want to control when the migration happens, see $wgActorTableSchemaMigrationStage and the migrateActors.php maintenance script.
If you are an extension developer, and your extension does direct DB access to some of the affected fields, you'll need to update your queries (and ideally use the getQueryInfo()
methods, such as Revision::getQueryInfo()
, instead of manually constructing your query).
During the migration period, you can also use the ActorMigration
class to get schema/field information that respects any $wgActorTableSchemaMigrationStage
settings.
If you are a tool developer using Toolforge , there might be performance implications, even if you don't use any of the affected fields directly. See wikitech:News/Actor storage changes on the Wiki Replicas for additional information.