Page MenuHomePhabricator

Deletion of Lexemes appears to leak triples related to its forms and senses
Closed, ResolvedPublic3 Estimated Story Points

Description

In T302189 it was reported:

This report of grammatical features is wrong because it includes deleted data. Like with the previous queries I mentioned, I'm unable to fix it because that takes it from running in under a second to timing out.

This query returns a form which was deleted 11 months ago.

(Here's 100 forms which need cleaning up)

It suggests that forms of deleted lexemes are leaked in the triple store.

Triples whose subjects have a form that is attached to a deleted Lexeme using ontolex:lexicalForm should be deleted as well.
It might be possible that senses are leaked too (attached with ontolex:sense).

AC:

  • when a Lexeme is deleted all its forms and senses should be removed from WDQS.

Event Timeline

Gehel triaged this task as High priority.Jan 9 2023, 4:23 PM
Gehel moved this task from Incoming to Tech Debt on the Wikidata-Query-Service board.
dr0ptp4kt lowered the priority of this task from High to Medium.Oct 1 2024, 3:26 PM
dr0ptp4kt set the point value for this task to 3.

First step is investigating to understand if we are missing events (which happens) or if there is a bug in the updater code.

Change #1104714 had a related patch set uploaded (by DCausse; author: DCausse):

[wikidata/query/rdf@master] Properly remove forms and senses when delete lexemes

https://gerrit.wikimedia.org/r/1104714

Change #1104714 merged by jenkins-bot:

[wikidata/query/rdf@master] Properly remove forms and senses when deleting lexemes

https://gerrit.wikimedia.org/r/1104714

Change #1112251 had a related patch set uploaded (by DCausse; author: DCausse):

[wikidata/query/deploy@master] deploy version 0.3.152

https://gerrit.wikimedia.org/r/1112251

Change #1112251 merged by Ryan Kemper:

[wikidata/query/deploy@master] deploy version 0.3.155

https://gerrit.wikimedia.org/r/1112251

Change #1133071 had a related patch set uploaded (by DCausse; author: DCausse):

[wikidata/query/deploy@master] Revert "deploy version 0.3.155"

https://gerrit.wikimedia.org/r/1133071

Change #1133071 merged by DCausse:

[wikidata/query/deploy@master] Revert "deploy version 0.3.155"

https://gerrit.wikimedia.org/r/1133071

Mentioned in SAL (#wikimedia-operations) [2025-04-01T08:46:05Z] <dcausse@deploy1003> Started deploy [wdqs/wdqs@354b5ac]: revert T326311, deletion query way too slow

Mentioned in SAL (#wikimedia-operations) [2025-04-01T08:58:21Z] <dcausse@deploy1003> Finished deploy [wdqs/wdqs@354b5ac]: revert T326311, deletion query way too slow (duration: 12m 15s)

Change #1133134 had a related patch set uploaded (by DCausse; author: DCausse):

[wikidata/query/rdf@master] Fix delete entity query

https://gerrit.wikimedia.org/r/1133134

I deployed test snapshot of the updater on wdqs2025 (test machine).
For the following query:

select * {
  VALUES ?f {wd:L327769-F1 wd:L326498-F1 wd:L327288-F1 wd:L324701-F1
             wd:L321884-F1 wd:L320004-F1 wd:L316324-F1 wd:L318083-F1
             wd:L327769-S1 wd:L326498-S1 wd:L327288-S1 wd:L324701-S1
             wd:L321884-S1 wd:L320004-S1 wd:L316324-S1 wd:L318083-S1}
  ?f ?p ?o .
} LIMIT 1000

It does properly return no results as opposed to query.wikidata.org which does not have yet the fix deployed.

Change #1133134 merged by jenkins-bot:

[wikidata/query/rdf@master] Fix delete entity query

https://gerrit.wikimedia.org/r/1133134

Change #1134710 had a related patch set uploaded (by DCausse; author: DCausse):

[wikidata/query/deploy@master] deploy version 0.3.156

https://gerrit.wikimedia.org/r/1134710

Change #1134710 merged by DCausse:

[wikidata/query/deploy@master] deploy version 0.3.156

https://gerrit.wikimedia.org/r/1134710

Mentioned in SAL (#wikimedia-operations) [2025-04-15T09:43:22Z] <dcausse@deploy1003> Started deploy [wdqs/wdqs@fe88851]: version 0.3.156 (T326311)

Mentioned in SAL (#wikimedia-operations) [2025-04-15T09:57:54Z] <dcausse@deploy1003> Finished deploy [wdqs/wdqs@fe88851]: version 0.3.156 (T326311) (duration: 14m 31s)