Out-of-sync data in the Wikidata database
Open, Needs TriagePublic
Actions

Assigned To

None

Authored By

	Mohammed_Sadat_WMDE
	Mar 7 2023, 10:40 AM

Project Tags

Referenced Files

None

Subscribers

• Aklapper

Bdijkstra

ItamarWMDE

• Lucas_Werkmeister_WMDE

Lydia_Pintscher

Mohammed_Sadat_WMDE

Description

User story:
Using Quarry on the Wikidata database, I notice the database is not always accurate.

Problem:
After saving an edit, the alias still had the old value according to the database. There are potentially many redirected Items which, according to the database, still have labels, descriptions and/or aliases.

Examples:

In this edit the page was successfully saved, but according to the database, the alias still had the old value.
a quarry for Items that are redirects but do have terms: https://quarry.wmcloud.org/query/71784

Previous discussion:
See discussion on Wikidata:Report a technical problem#Database out of sync

Related Objects

Mentioned Here: T309445: [Investigation] Terms storage missing most item labels after merge, resulting in item IDs being shown

Event Timeline

Mohammed_Sadat_WMDE created this task.Mar 7 2023, 10:40 AM

Restricted Application added a subscriber: • Aklapper. · View Herald TranscriptMar 7 2023, 10:40 AM

Bugreporter added a project: Quarry.Mar 7 2023, 11:47 AM

taavi renamed this task from Quarry showing out-of-sync data in the Wikidata database to Out-of-sync data in the Wikidata database.Mar 7 2023, 11:51 AM

taavi removed a project: Quarry.

Bdijkstra subscribed.Mar 7 2023, 12:51 PM

Bugreporter added a project: Data-Services.Mar 9 2023, 4:49 AM

JJMC89 moved this task from Backlog to Wiki replicas on the Data-Services board.Mar 29 2023, 4:53 PM

Lydia_Pintscher updated the task description. (Show Details)Apr 6 2023, 10:17 AM

taavi removed a project: Data-Services.Apr 6 2023, 10:18 AM

Devs: Do we know where in the chain the problem is? Is it that our primary term store still has the terms when a redirect is created? Or is it from there to the replica? Or somewhere else?

@Lydia_Pintscher, we will take a look at this asap to determine the source.

a quarry for Items that are redirects but do have terms: https://quarry.wmcloud.org/query/71784

At least the first item mentioned there also has terms on the production database:

lucaswerkmeister-wmde@mwdebug2002:~$ mwscript sql wikidatawiki
> SELECT COUNT(*) FROM wbt_item_terms WHERE wbit_item_id = 10000336;
stdClass Object
(
    [COUNT(*)] => 129
)

> SELECT page_is_redirect FROM page WHERE page_namespace = 0 AND page_title = 'Q10000336';
stdClass Object
(
    [page_is_redirect] => 1
)

So this is probably a general Wikibase issue and nothing specific to the cloud replicas. Possibly the same as T309445: [Investigation] Terms storage missing most item labels after merge, resulting in item IDs being shown.

Bdijkstra added a comment.Apr 11 2023, 4:04 PM

This comment was removed by Bdijkstra.

ItamarWMDE moved this task from Incoming to Radar on the wmde-wikidata-tech board.May 9 2023, 12:52 PM

Out-of-sync data in the Wikidata databaseOpen, Needs TriagePublicActions

Description

Related Objects

Event Timeline

Out-of-sync data in the Wikidata database
Open, Needs TriagePublic
Actions