Page MenuHomePhabricator

"A file disappeared" thread on Commons:VP
Closed, ResolvedPublic

Description

Author: test5555

Description:
Some users uploads don't seem to be visible or disappeared. See URL for discussion.

Permanent link: http://commons.wikimedia.org/w/index.php?title=Commons:Village_pump&oldid=43288238#A_file_disappeared


Version: unspecified
Severity: major
URL: http://commons.wikimedia.org/wiki/Commons:Village_pump#A_file_disappeared

Details

Reference
bz24978

Related Objects

StatusSubtypeAssignedTask
DeclinedNone
ResolvedNone

Event Timeline

bzimport raised the priority of this task from to High.Nov 21 2014, 11:09 PM
bzimport set Reference to bz24978.
bzimport added a subscriber: Unknown Object (MLST).

These are image table entries disappearing. The article pages and actual files in upload.wikimedia.org remain there.

This is just another instance of bug 20744.

With YOGArtisticGymnastics-BishanSportsHall-Singapore-20100816-02.jpg we don't have evidence that the image record was created (although per the uploader, it seems so).

Itämetro-Östmetron.svg on the other hand had it, having even a screenshot about that [1]. And we know another interesting fact: it was a zombie file (ie. affected by bug 15430).

The cojuelo likes removing image entries the 16 of each month?

1-http://1.bp.blogspot.com/_QVjd811Hpm8/THDt1OuK9EI/AAAAAAAADm8/snSKrgrTlMk/s1600/jatko+it%C3%A4%C3%A4n.JPG

There was some trouble over the past week with the servers of course. I believe ms4 had quite some issues at one point, so perhaps it is related to that.

AFAIK ms4 is only storage. We don't have problems with the files, but with the db. There seems to have been a master switch, but it's not logged in the admin log.

I was the uploader of File:YOGArtisticGymnastics-BishanSportsHall-Singapore-20100816-02.jpg. Just wanted to confirm that it uploaded correctly together with the other files in "Category:Gymnastics at the 2010 Summer Youth Olympics", but then I noticed a few days ago that the file had "disappeared".

Abigor wrote:

*** Bug 20744 has been marked as a duplicate of this bug. ***

yeah, blame the infrastructure!!!!!!!!!!!!!!1111 SERVERS ARE ALWAYS WRONG.

all the other files were correct transaction, this one was logged as 'patrol' event, and didn't have 'image' row insertion.

so, the image didn't "disappear", it was never there.

SET TIMESTAMP=1282492172;
BEGIN;
SET INSERT_ID=11251250;
INSERT /* Article::insertOn Jacklee */ IGNORE INTO page (page_id,page_namespace,page_title,page_counter,page_restrictions,page_is_redirect,page_is_new,page_random,page_touched,page_latest,page_len) VALUES (NULL,'6','YOGArtisticGymnastics-BishanSportsHall-Singapore-20100816-02.jpg','0','','0','1','0.199751853441','20100822154932','0','0');

SET INSERT_ID=42296044;
INSERT /* Revision::insertOn Jacklee */ INTO text (old_id,old_text,old_flags) VALUES (NULL,'DB://cluster22/22407194','utf-8,gzip,external');

SET INSERT_ID=42997008;
INSERT /* Revision::insertOn Jacklee */ INTO revision (rev_id,rev_page,rev_text_id,rev_comment,rev_minor_edit,rev_user,rev_user_text,rev_timestamp,rev_deleted,rev_len,rev_parent_id) VALUES (NULL,'11251250','42296044','Added information','0','99051','Jacklee','20100822154932','0','608','0');

UPDATE /* Article::updateRevisionOn Jacklee */ page SET page_latest = '42997008',page_touched = '20100822154932',page_is_new = '1',page_is_redirect = '0',page_len = '608' WHERE page_id = '11251250' AND page_latest = '0';

SET INSERT_ID=44199374;
INSERT /* RecentChange::save Jacklee */ INTO recentchanges (rc_timestamp,rc_cur_time,rc_namespace,rc_title,rc_type,rc_minor,rc_cur_id,rc_user,rc_user_text,rc_comment,rc_this_oldid,rc_last_oldid,rc_bot,rc_moved_to_ns,rc_moved_to_title,rc_ip,rc_patrolled,rc_new,rc_old_len,rc_new_len,rc_deleted,rc_logid,rc_log_type,rc_log_action,rc_params,rc_id) VALUES ('20100822154932','20100822154932','6','YOGArtisticGymnastics-BishanSportsHall-Singapore-20100816-02.jpg','1','0','11251250','99051','Jacklee','Added information','42997008','0','0','0','','...','1','1','0','608','0','0',NULL,'','',NULL);

INSERT /* efUpdateCheckUserData Jacklee */ INTO cu_changes ...

SET INSERT_ID=2184367;
INSERT /* Job::insert Jacklee */ INTO job (job_id,job_cmd,job_namespace,job_title,job_params) VALUES (NULL,'enotifNotify','6','YOGArtisticGymnastics-BishanSportsHall-Singapore-20100816-02.jpg','a:7:{s:6:\"editor\";s:7:\"Jacklee\";s:8:\"editorID\";i:99051;s:9:\"timestamp\";s:14:\"20100822154932\";s:7:\"summary\";s:17:\"Added information\";s:9:\"minorEdit\";i:0;s:5:\"oldid\";i:0;s:8:\"watchers\";a:0:{}}');

SET TIMESTAMP=1282492172;
INSERT /* LogPage::saveContent Jacklee */ INTO logging (log_id,log_type,log_action,log_timestamp,log_user,log_user_text,log_namespace,log_title,log_page,log_comment,log_params) VALUES (NULL,'patrol','patrol','20100822154932','99051','Jacklee','6','YOGArtisticGymnastics-BishanSportsHall-Singapore-20100816-02.jpg','0','','42997008\n0\n1');

UPDATE /* User::incEditCount Jacklee */ user SET user_editcount=user_editcount+1 WHERE user_id = '99051';

UPDATE /* require/MediaWiki::performRequestForTitle/MediaWiki::performAction/EditPage::submit/EditPage::edit/EditPage::attemptSave/EditPage::internalAttemptSave/Article::insertNewArticle/Article::doEdit/User::incEditCount/User::invalidateCache/wfGetAllCallers Jacklee */ user SET user_touched = '20100822154937' WHERE user_id = '99051';

COMMIT;

I don't think there's much magic here, some image had failed upload - transaction was interrupted somewhere in the mediawikiland, and later a page was created on top with other data - and it has different information from what was provided to other uploads by the user.

So, file did not disappear, it was never there.

That's page creation. What about image insertion for Itämetro-Östmetron.svg?

Note that in http://1.bp.blogspot.com/_QVjd811Hpm8/THDt1OuK9EI/AAAAAAAADm8/snSKrgrTlMk/s1600/jatko+it%C3%A4%C3%A4n.JPG that file was *shown* in wiki page, so despite being a zombie file it would have had an entry in image table.

Just placing the bar a bit higher ;)

when was the file uploaded? only 'patrol' entry (and later, delete) in logging for that one.

I guess the problem here is that data gets placed into memcached without a valid transaction %)

(In reply to comment #10)

@Jacklee, how did you upload those files?

I used the basic upload page. It looked fine at the time when I uploaded the file, then I happened to notice last week that the thumbnail was no longer showing up. Anyway, it's clear that I did upload the file as it appears at http://upload.wikimedia.org/wikipedia/commons/e/ef/YOGArtisticGymnastics-BishanSportsHall-Singapore-20100816-02.jpg.

Fixed in r72021.

Jacklee, a sysadmin can import it.