Page MenuHomePhabricator

Outdated data still present in WCQS a month after statement update
Closed, ResolvedPublic5 Estimated Story Points

Description

In this query , there are files mistakenly filed as made in January 1010.

I fixed the data on one of these files (sdc:M8979671) on October 8th, over a month ago. However, running the query again still returns that file.

Event Timeline

Restricted Application added a subscriber: Aklapper. · View Herald Transcript

As of today December 16th, M8979671 is still present in that query’s results.

CBogen set the point value for this task to 2.Jan 4 2021, 4:50 PM
CBogen subscribed.

This is likely a problem with the dumps; the search team will investigate to be sure and then pass it on to the right team if so.

The dumps have been investigated and the data is correct there, the problem is definitely on the search team's side. It could be with the reload scripts or nginx. It was possible to see the incorrect data on the wcqs server, so the problem is most likely with nginx.

On the wcqs beta host, curl -d "query=select * { sdc:M8979671 wdt:P571 ?o . }" localhost/bigdata/namespace/wcq/sparql returns an item with the correct data and curl -d "query=select * { sdc:M8979671 wdt:P571 ?o . }" localhost:9999/bigdata/namespace/wcq/sparql returns the 1010 date

CBogen changed the point value for this task from 2 to 3.Jan 11 2021, 4:50 PM
CBogen changed the point value for this task from 3 to 5.

Change 655639 had a related patch set uploaded (by ZPapierski; owner: ZPapierski):
[operations/puppet@production] Fix /sparql rewrite and alias rules

https://gerrit.wikimedia.org/r/655639

On the wcqs beta host, curl -d "query=select * { sdc:M8979671 wdt:P571 ?o . }" localhost/bigdata/namespace/wcq/sparql returns an item with the correct data and curl -d "query=select * { sdc:M8979671 wdt:P571 ?o . }" localhost:9999/bigdata/namespace/wcq/sparql returns the 1010 date

This is actually a correct behavior - first address resolves to a correct namespace, second simply points to an old namespace that should be used anymore (but unfortunately is). Issue is with the rewrite chaing in nginx configuration.

Change 655639 merged by Ryan Kemper:
[operations/puppet@production] Fix /sparql rewrite and alias rules

https://gerrit.wikimedia.org/r/655639

Issue was fixed manually for now - I'll do a proper deployment once I get some time to deal with recent microsite GUI changes.