SPARQL shows redirect with original data months after merge
Closed, ResolvedPublic3 Estimated Story PointsBUG REPORT
Actions

Assigned To

Authored By

	Nux
	May 7 2025, 10:46 PM

Description

Steps to replicate the issue (include links if applicable):

Merge item on September 2024 (8 months ago).
- History of merged item: https://www.wikidata.org/w/index.php?title=Q121884648&action=history (redirect)
- History of destination item: https://www.wikidata.org/w/index.php?title=Q4801376&action=history
Query for items by coords: https://w.wiki/E322

What happens?:

The query shows both items (higher Q is a redirect).

What should have happened instead?:

Redirects should be purged from the graph and not be part of any SPARQL... or at least they should have some status.

Software version (on Special:Version page; skip for WMF-hosted wikis like Wikipedia):
WD

Other information (browser name/version, screenshots, etc.):

Event Timeline

Nux created this task.May 7 2025, 10:46 PM

Restricted Application added a subscriber: Aklapper. · View Herald TranscriptMay 7 2025, 10:46 PM

Pppery added a project: Wikidata-Query-Service.May 7 2025, 11:40 PM

Maintenance_bot added a project: Wikidata.May 8 2025, 12:29 AM

Aklapper renamed this task from SPARQL shows redirct with original data months after merge to SPARQL shows redirect with original data months after merge.May 8 2025, 5:28 PM

The data / logs are probably too old so a full investigation on the causes of the issue is unlikely to be successful. We should still reload that item to fix the data drift.

Gehel moved this task from Incoming to Ready for Dev on the Discovery-Search (2025.05.02 - 2025.05.23) board.May 12 2025, 3:42 PM

Gehel set the point value for this task to 3.

There is also a 2nd one in Poland if that helps:
https://www.wikidata.org/w/index.php?title=Q122261336&action=history

Though that's around same time, so I guess it won't help that much.

Gehel edited projects, added Discovery-Search (2025.05.24 - 2025.06.13); removed Discovery-Search (2025.05.02 - 2025.05.23).May 23 2025, 1:01 PM

Gehel moved this task from Incoming to Ready for Dev on the Discovery-Search (2025.05.24 - 2025.06.13) board.

dcausse claimed this task.May 28 2025, 8:16 AM

dcausse moved this task from Ready for Dev to In Progress on the Discovery-Search (2025.05.24 - 2025.06.13) board.

Thanks for the report, I see two stale revisions for these items indeed:

select * {
  VALUES ?stale_revision {1962416023 1969436471}
  ?item schema:version ?stale_revision
}

returning:

item            stale_revision
wd:Q121884648	1962416023
wd:Q122261336	1969436471

From the revision-create stream I can't find any events related to

Q122261336 on 2024-08-27
Q121884648 on 2024-09-11

Both

select * from mediawiki_revision_create where datacenter in ('eqiad', 'codfw') and year = 2024 and month=8 and day=27 and page_title='Q122261336'

and

select * from mediawiki_revision_create where datacenter in ('eqiad', 'codfw') and year = 2024 and month=9 and day=11 and page_title='Q121884648'

return no result in the event_sanitized database.

I suspect that the event was missed...
Missing events is sadly a known issue for which there's no consensus yet on how to best mitigate those.

I reconciled these two items to fix the state.

Gehel moved this task from Done to Reported on the Discovery-Search (2025.05.24 - 2025.06.13) board.May 30 2025, 11:59 AM

SPARQL shows redirect with original data months after mergeClosed, ResolvedPublic3 Estimated Story PointsBUG REPORTActions

Description

Event Timeline

SPARQL shows redirect with original data months after merge
Closed, ResolvedPublic3 Estimated Story PointsBUG REPORT
Actions