hewiki's categorylinks shown as not empty though it is; purging does not help
Closed, ResolvedPublic
Actions

Assigned To

Authored By

	eranroz
	Oct 16 2015, 4:29 AM

Description

The category used for speed deletion in Hebrew Wikipedia is shown as not empty although it is empty for few days, and purging the category page doesn't change it.

Category used for speed deletion in Hebrew Wikipedia called: "ויקיפדיה: למחיקה מהירה" https://he.wikipedia.org/wiki/%D7%A7%D7%98%D7%92%D7%95%D7%A8%D7%99%D7%94:%D7%95%D7%99%D7%A7%D7%99%D7%A4%D7%93%D7%99%D7%94:_%D7%9C%D7%9E%D7%97%D7%99%D7%A7%D7%94_%D7%9E%D7%94%D7%99%D7%A8%D7%94
If you purge the above page it is shown as empty
but {{PAGESINCATEGORY:ויקיפדיה: למחיקה מהירה|files}} shows that are 4 files in this category

Based on database query

select * from categorylinks where cl_to like '%למחיקה%'

the files are:
https://he.wikipedia.org/wiki/%D7%A7%D7%95%D7%91%D7%A5:Sg_logo.PNG (deleted on October 9)
https://he.wikipedia.org/wiki/%D7%A7%D7%95%D7%91%D7%A5:KKS_Lech_Pozna%C5%84.gif (deleted on October 14)

Restoring, modifying and re-deleting the above files doesn't help

Related Objects

Mentioned Here: T115586: deleteLinks jobs reliably fail due to pageId = 0

Event Timeline

eranroz created this task.Oct 16 2015, 4:29 AM

eranroz raised the priority of this task from to Needs Triage.

eranroz updated the task description. (Show Details)

eranroz added a project: acl*sre-team.

eranroz added subscribers: eranroz, IKhitron.

Restricted Application added subscribers: Matanya, Aklapper. · View Herald TranscriptOct 16 2015, 4:29 AM

Legoktm added projects: WMF-General-or-Unknown, DBA.Oct 16 2015, 4:31 AM

Legoktm set Security to None.

Aklapper renamed this task from Database corruption in hewiki (categorylinks) to hewiki's categorylinks shown as not empty though it is; purging does not help.Oct 16 2015, 11:11 AM

Referring to LinksDeletionUpdate.php - are there cleanup triggers configurated on hewiki? are they enabled?

It's empty now. And purging it wouldn't help ever; you'd need to purge every file shown in it

Categorylinks isn't empty.
{{PAGESINCATEGORY:ויקיפדיה: למחיקה מהירה}} return 2
select * from categorylinks where cl_to like '%למחיקה%';

The page associated with those categorylinks are deleted - so it looks like there are no members, but there are.

The hewiki master shows 7

mysql:wikiadmin@db1062 [hewiki]> select * from categorylinks where cl_to like '%למחיקה%'
    -> ;
+---------+-------------------------------------------+--------------------------------------------+---------------------+-------------------+--------------+---------+
| cl_from | cl_to                                     | cl_sortkey                                 | cl_timestamp        | cl_sortkey_prefix | cl_collation | cl_type |
+---------+-------------------------------------------+--------------------------------------------+---------------------+-------------------+--------------+---------+
|  408236 | ויקיפדיה:_למחיקה_מהירה                    | GFATOTAL.GIF                               | 2015-10-16 09:40:25 |                   | uppercase    | file    |
|  510006 | ויקיפדיה:_למחיקה_מהירה                    | KKS LECH POZNAŃ.GIF                        | 2015-10-14 17:35:39 |                   | uppercase    | file    |
|  695697 | ויקיפדיה:_למחיקה_מהירה                    | SG LOGO.PNG                                | 2015-10-08 19:24:25 |                   | uppercase    | file    |
|  933942 | ויקיפדיה:_למחיקה_מהירה                    | CSKABASKET.JPG                             | 2015-10-16 07:38:49 |                   | uppercase    | file    |
| 1201221 | ויקיפדיה:_למחיקה_מהירה                    | הפועל ירושלים כדורסל.JPG                   | 2015-10-17 12:21:10 |                   | uppercase    | file    |
| 1319470 | ויקיפדיה:_למחיקה_מהירה                    | SG LOGO.PNG                                | 2015-10-16 04:19:34 |                   | uppercase    | file    |
| 1319471 | ויקיפדיה:_למחיקה_מהירה                    | KKS LECH POZNAŃ.GIF                        | 2015-10-16 04:22:06 |                   | uppercase    | file    |
+---------+-------------------------------------------+--------------------------------------------+---------------------+-------------------+--------------+---------+
7 rows in set (1.31 sec)

mysql:wikiadmin@db1062 [hewiki]>

Most or all of teh cl_from are deleted pages. e.g:

select * from categorylinks left join page on cl_from=page_id where cl_to like '%למחיקה%';

Probably a duplicate of T115586

What is needed from ops here if anything?

added MediaWiki-JobQueue, removed operations. per comment from Aaron this should be a duplicate

Please cleanup categorylinks that MW should have deleted e.g:

delete from categorylinks where cl_from not in (select page_id from page)

(CAUTION! I didn't test it)

Once we clean this up we can follow up:

Either it will not occur anymore and this was due to one of the outages that happened recently and that's it (hopefully)
It will come back and we will be sure it is a software bug or missing triggers

In T115682#1736507, @eranroz wrote:
Please cleanup categorylinks that MW should have deleted e.g:
delete from categorylinks where cl_from not in (select page_id from page)
(CAUTION! I didn't test it)

Once we clean this up we can follow up:

Either it will not occur anymore and this was due to one of the outages that happened recently and that's it (hopefully)

It will come back and we will be sure it is a software bug or missing triggers

Shouldn't be needed. When the refreshLinks.php maintenance script is run on the wiki, it will do it. Noting it's on a regular cron for all wikis

https://github.com/wikimedia/operations-puppet/blob/812f280d16acfe3083259e8dfa7ce12ebf71da87/modules/mediawiki/manifests/maintenance/refreshlinks.pp#L28-L29

7th November

(Noting this would likely be a problem on all wikis)

jcrespo moved this task from Triage to Blocked external/Not db team on the DBA board.Nov 4 2015, 7:19 PM

Restricted Application added a subscriber: StudiesWorld. · View Herald TranscriptNov 4 2015, 7:19 PM

Nemo_bis removed a project: WMF-General-or-Unknown.Dec 5 2015, 2:19 PM

Dzahn unsubscribed.May 27 2016, 10:33 PM

eranroz closed this task as Resolved.May 20 2017, 1:18 PM

eranroz claimed this task.

hewiki's categorylinks shown as not empty though it is; purging does not helpClosed, ResolvedPublicActions

Description

Related Objects

Event Timeline

hewiki's categorylinks shown as not empty though it is; purging does not help
Closed, ResolvedPublic
Actions