885309 - Duplicate questions in the search index

Status: RESOLVED FIXED

(support.mozilla.org :: Search, defect, P1)

Description

[Ricky Rosario [:rrosario, :r1cky]]

Something is causing questions to appear in the index multiple times.

See:
https://support.mozilla.org/en-US/forums/contributors/709276

An example is:
https://support.mozilla.org/search?q=&asked_by=dick888&answered_by=&w=2&a=1&sortby=1

Although, that might go away as soon as we reindex.

Comment 1

[Ricky Rosario [:rrosario, :r1cky]]

Working on this now.

Comment 2

[Will Kahn-Greene [:willkg] ET needinfo? me]

I wonder if we're indexing a question by different document ids in different places. Maybe the update_questions_vote_task (or whatever it's called) is using a different document id than we're using in the indexing methods in QuestionMappingType (or whatever it's called)?

Comment 3

[Ricky Rosario [:rrosario, :r1cky]]

Putting into the next sprint as I haven't had a chance to dig too deep here. Unassigning myself in case somebody else can get to it first.

Making it a 2pt as a initial guess. I looked a little and it didn't look like a quick thing to troubleshoot.

Comment 4

[Will Kahn-Greene [:willkg] ET needinfo? me]

Grabbing this one to look into.

Comment 5

[Will Kahn-Greene [:willkg] ET needinfo? me]

https://github.com/mozilla/kitsune/pull/1454

Comment 6

[Will Kahn-Greene [:willkg] ET needinfo? me]

Fix landed in master in https://github.com/mozilla/kitsune/commit/fecf18a

This will prevent new duplicate questions, but we need to wipe the index of the old ones. I'll do that later today.

Comment 7

[Will Kahn-Greene [:willkg] ET needinfo? me]

Pushed out a two-step reindexing and the problem questions are now gone.

I did a grep across the codebase for other instances of document_id and didn't see any.

I no longer see duplicate questions in the urls provided in the description.

I'm pretty sure fecf18a should fix this issue.