Target deletion during page move fails
Open, Unbreak Now!Public

Description

Failed to delete page-move revision: <OK, collected 1 error(s) on the way, no value set>

Error

Request ID: W--3DgpAMFAAAIRg54sAAAAD

message
Failed to delete page-move revision: <OK, collected 1 error(s) on the way, no value set>
+------+---------------------------+------------------------------------------+
|    1 | cannotdelete              | Kosice                                   |
+------+---------------------------+------------------------------------------+
trace
#0 /srv/mediawiki/php-1.33.0-wmf.6/includes/MovePage.php(276): MovePage->moveToInternal(User, Title, string, boolean, array)
#1 /srv/mediawiki/php-1.33.0-wmf.6/includes/specials/SpecialMovepage.php(608): MovePage->move(User, string, boolean)
#2 /srv/mediawiki/php-1.33.0-wmf.6/includes/specials/SpecialMovepage.php(128): MovePageForm->doSubmit()
#3 /srv/mediawiki/php-1.33.0-wmf.6/includes/specialpage/SpecialPage.php(569): MovePageForm->execute(NULL)
#4 /srv/mediawiki/php-1.33.0-wmf.6/includes/specialpage/SpecialPageFactory.php(558): SpecialPage->run(NULL)
#5 /srv/mediawiki/php-1.33.0-wmf.6/includes/MediaWiki.php(288): MediaWiki\Special\SpecialPageFactory->executePath(Title, RequestContext)
#6 /srv/mediawiki/php-1.33.0-wmf.6/includes/MediaWiki.php(862): MediaWiki->performRequest()
#7 /srv/mediawiki/php-1.33.0-wmf.6/includes/MediaWiki.php(517): MediaWiki->main()
#8 /srv/mediawiki/php-1.33.0-wmf.6/index.php(42): MediaWiki->run()
#9 /srv/mediawiki/w/index.php(3): include(string)
#10 {main}

Impact

Notes

Found 9 occurences when deploying 1.33.0-wmf.6 to all wikis.

Happens at least on page renaming.

Related Objects

There are a very large number of changes, so older changes are hidden. Show Older Changes
Restricted Application added projects: Commons, Multimedia. · View Herald TranscriptFri, Jan 11, 1:17 PM

Looks like something started triggering it again. Did that bot re-start or something?

archive/exception.log-20190105.gz:0
archive/exception.log-20190106.gz:0
archive/exception.log-20190107.gz:0
archive/exception.log-20190108.gz:0
archive/exception.log-20190109.gz:4
archive/exception.log-20190110.gz:24
archive/exception.log-20190111:208
exception.log:114

I don't think this has been noted explicitly above, but manually deleting the target page (at least for me) let it go ahead just fine earlier today.

Sakretsu added a subscriber: Sakretsu.EditedFri, Jan 11, 4:41 PM

Between yesterday and today, it happened to me twice on it.wiki, and both times I avoided the error as Amorymeltzer described.

greg lowered the priority of this task from Unbreak Now! to High.Fri, Jan 11, 11:31 PM

Probably downgrade it from UBN, since it seems it mostly stopped happening:

Zache added a subscriber: Zache.Sat, Jan 12, 2:47 PM

One reported case from fiwiki when moving "Year of the Dog...Again" to "Year of the Dog... Again" with removing the target page.

I tested it and got error:
[XDn7cwpAMEkAADB3118AAACD] 2019-01-12 14:36:36: Fatal exception of type MWException

Ejs-80 added a subscriber: Ejs-80.Sat, Jan 12, 3:26 PM
Jules78120 added a subscriber: Jules78120.EditedSat, Jan 12, 4:06 PM

On fr-wiki, same error since (at least) yesterday:

[XDm4IgpAAD8AAAlTErkAAACP] 2019-01-12 09:49:23: Erreur fatale de type « MWException »

Please see this page in Portuguese Wikipedia. Ordinarily, when we try to move a page to other title, and this title already exists (a history of three revisions, e.g.), eliminators/administrators have the option of delete that page ('Yes, delete page'), as in other wikis. But, there is a bug that when we mark to delete page, it is displayed the following error "[XDomdgpAADgAAJv80TAAAABT] 2019-01-12 17:40:07: Fatal exception of type MWException". Could you help us? Regards, Pórokhov.

MWException from line 497 of /srv/mediawiki/php-1.33.0-wmf.12/includes/MovePage.php: Failed to delete page-move revision: <OK, collected 1 error(s) on the way, no value set>
+------+---------------------------+------------------------------------------+
|    1 | cannotdelete              | Usuário:Pórokhov/Testes234             |
+------+---------------------------+------------------------------------------+
 {"exception_id":"XDomdgpAADgAAJv80TAAAABT","exception_url":"/w/index.php?title=Especial:Mover_p%C3%A1gina&action=submit","caught_by":"mwe_handler"} 
[Exception MWException] (/srv/mediawiki/php-1.33.0-wmf.12/includes/MovePage.php:497) Failed to delete page-move revision: <OK, collected 1 error(s) on the way, no value set>
+------+---------------------------+------------------------------------------+
|    1 | cannotdelete              | Usuário:Pórokhov/Testes234             |
+------+---------------------------+------------------------------------------+

  #0 /srv/mediawiki/php-1.33.0-wmf.12/includes/MovePage.php(263): MovePage->moveToInternal(User, Title, string, boolean, array)
  #1 /srv/mediawiki/php-1.33.0-wmf.12/includes/specials/SpecialMovepage.php(608): MovePage->move(User, string, boolean)
  #2 /srv/mediawiki/php-1.33.0-wmf.12/includes/specials/SpecialMovepage.php(128): MovePageForm->doSubmit()
  #3 /srv/mediawiki/php-1.33.0-wmf.12/includes/specialpage/SpecialPage.php(569): MovePageForm->execute(NULL)
  #4 /srv/mediawiki/php-1.33.0-wmf.12/includes/specialpage/SpecialPageFactory.php(558): SpecialPage->run(NULL)
  #5 /srv/mediawiki/php-1.33.0-wmf.12/includes/MediaWiki.php(288): MediaWiki\Special\SpecialPageFactory->executePath(Title, RequestContext)
  #6 /srv/mediawiki/php-1.33.0-wmf.12/includes/MediaWiki.php(862): MediaWiki->performRequest()
  #7 /srv/mediawiki/php-1.33.0-wmf.12/includes/MediaWiki.php(517): MediaWiki->main()
  #8 /srv/mediawiki/php-1.33.0-wmf.12/index.php(42): MediaWiki->run()
  #9 /srv/mediawiki/w/index.php(3): include(string)
  #10 {main}
He7d3r added a subscriber: He7d3r.Sat, Jan 12, 6:55 PM

For the record, this happened 127 times in the last 24 hours. Not a lot, but not a little.

JOAN added a subscriber: JOAN.Sun, Jan 13, 4:36 PM
Wbm1058 added a comment.EditedMon, Jan 14, 11:16 PM

OK, I just got this error again: Internal error

[XD0XBwpAICMAAFWVm0cAAABH] 2019-01-14 23:11:04: Fatal exception of type "MWException"

Trying to move https://en.wikipedia.org/wiki/MOS_Technology_6582 to https://en.wikipedia.org/wiki/MOS_Technology_6581

There is no urgency to complete this move, so I will leave it as-is for now. This should give the developers an example which they may easily duplicate in real time, so you can trace it down to the root cause while it is actively failing.

Huji added a comment.Tue, Jan 15, 3:52 PM

The move I reported in T213819 is also non-urgent, so I will leave it as is for now, in case any developer needs to try it again.

I tried to test this on my wiki, and it succeeded 3-4 times in a row. I wonder if this could be due to some race condition... Also, the stream of errors is pretty steady, and is of ~100 errors a day. I guess the only reason for this not to be UBN! is that a simple workaround (delete first, then move) exists.

Huji added a comment.Tue, Jan 15, 5:08 PM

I think this should be UBN, per Collaboration/Team/Processes#Definition of Unbreak Now (which specifies that "Any fatal" is UBN). I had marked T213819 as UBN, for that reason.

Project management#Priority levels is not as verbose, but it does specify that "High" means "Someone is working or planning to work on this task soon", which, considering this task was opened in November and is still unassigned, I think is not completely appropriate.

But these technicalities aside, I think UBN is there to capture the attention of a wider group of people; this task seems to have attracted the attention of the right people, so I am fine leaving it as High.

Probably downgrade it from UBN, since it seems it mostly stopped happening:

It was downgraded from UBN after it mostly stopped happening. Well, it is now "mostly happening" again, so somebody should take ownership of it now, before it randomly goes back into hibernation.

Do we need to wait for a volunteer to step up? Is that how the WMF would handle it if some part of the fundraising system broke and failed when readers clicked the "donate" button?

Sorry, I don't have a good handle on how tasks are actually assigned.

I tried to test this on my wiki, and it succeeded 3-4 times in a row. I wonder if this could be due to some race condition...

It mysteriously mostly stopped happening after Dec 3, and then started again Jan 10. When it was occurring earlier, sometimes simply retrying the move would work. It was hypothesized at the time that mass page moves by TheSandBot were somehow causing problems, but as far as I know no one actually confirmed that.

Trying to move https://en.wikipedia.org/wiki/MOS_Technology_6582 to https://en.wikipedia.org/wiki/MOS_Technology_6581

There is no urgency to complete this move, so I will leave it as-is for now. This should give the developers an example which they may easily duplicate in real time, so you can trace it down to the root cause while it is actively failing.

Tried it just now, several times, with a hack in place to prevent the move from succeeding should the bug not be triggered. Unfortunately, the bug did not trigger any of the times I tried it.

Internal error

[XD4cSApAMFUAAC@xcKIAAAAK] 2019-01-15 17:45:44: Fatal exception of type "MWException"

Still fails for me. Feel free to try it and let the move go through. You can say in the edit summary it was "per request of Wbm1058"

Change 484516 had a related patch set uploaded (by Anomie; owner: Anomie):
[mediawiki/core@master] Add temporary logging for T210739

https://gerrit.wikimedia.org/r/484516

Aklapper raised the priority of this task from High to Unbreak Now!.Tue, Jan 15, 6:14 PM

It now does not seem anymore that this has mostly stopped, hence reverting the Priority change from a few days ago.

Anomie, thanks for "taking ownership" of this. Just noting that when it has failed for me, retrying has never been successful. I just retried another move several times, and it failed each time. I'll stop doing that now as it is probably just pointless spamming of your logs. Hopefully your temporary logging will find the culprit.

Huji added a comment.Tue, Jan 15, 7:26 PM

Sadly, the example I had kept on fawiki is no more; I was forced to delete the target and proceed with the move, because of a user complaint. :(

Change 484704 had a related patch set uploaded (by Anomie; owner: Anomie):
[mediawiki/core@wmf/1.33.0-wmf.12] Add temporary logging for T210739

https://gerrit.wikimedia.org/r/484704

Porokhov removed a subscriber: Porokhov.Wed, Jan 16, 4:47 PM

Change 484704 merged by jenkins-bot:
[mediawiki/core@wmf/1.33.0-wmf.12] Add temporary logging for T210739

https://gerrit.wikimedia.org/r/484704

Mentioned in SAL (#wikimedia-operations) [2019-01-16T17:07:15Z] <anomie@deploy1001> Synchronized php-1.33.0-wmf.12/includes/page/WikiPage.php: Add temporary logging for T210739 (duration: 00m 53s)

Change 484825 had a related patch set uploaded (by Gergő Tisza; owner: Gergő Tisza):
[mediawiki/core@master] Fix error reporting in MovePage

https://gerrit.wikimedia.org/r/484825

Change 484825 merged by jenkins-bot:
[mediawiki/core@master] Fix error reporting in MovePage

https://gerrit.wikimedia.org/r/484825

Change 484845 had a related patch set uploaded (by Jforrester; owner: Gergő Tisza):
[mediawiki/core@wmf/1.33.0-wmf.13] Fix error reporting in MovePage

https://gerrit.wikimedia.org/r/484845

All the log entries associated with this task are showing that WikiPage->getId() returned 0, which is not what I expected. Time for some different logging.

Change 484845 abandoned by Jforrester:
Fix error reporting in MovePage

https://gerrit.wikimedia.org/r/484845

Since a solution doesn't seem to be near, I guess we should at least inform communities of the problem. Many people already know, but not everybody.

I restored https://en.wikipedia.org/wiki/MediaWiki:Delete_and_move_confirm which I'd deleted when this went into hibernation. That's probably sufficient; I'm not sure a notice in Tech News is necessary.

Do we need to wait until next Thursday to install the second attempt at catching the culprit in the logs, or can that be done sooner?

I restored https://en.wikipedia.org/wiki/MediaWiki:Delete_and_move_confirm which I'd deleted when this went into hibernation. That's probably sufficient; I'm not sure a notice in Tech News is necessary.

The problem hits all wikis, not only English Wikipedia.

Right, that occured to me moments after I sent my last message. So, in the Tech News, please do advise sysops to add a notice to their wiki's [[MediaWiki:Delete and move confirm]] page.

RuyP added a subscriber: RuyP.Sat, Jan 19, 12:51 AM
Teles added a subscriber: Teles.Sat, Jan 19, 3:38 AM

Ukwiki is affected too, so I would support the mention in the Tech news for other wikis.

Daimona added a subscriber: Johan.Sat, Jan 19, 2:27 PM

The whole MediaWiki sofware is affected... Unfortunately T/N/04 is already translated and we'll probably have to announce it in /05 (which will be published on Jan 28th). CC'ing @Johan to be sure. The message could be:

Because of a known bug, moving a page over a target which already exists (thus overwriting it) can sometimes fail with an error message. We're actively working on understanding and trying to resolve the issue. You can check the state of the problem [[:phab:T210739|on Phabricator]].

Meno25 added a subscriber: Meno25.Sat, Jan 19, 2:33 PM
Huji added a comment.Sat, Jan 19, 5:17 PM

I would expand it as:

Because of a known bug, moving a page over a target which already exists (thus overwriting it) can sometimes fail with an error message. We're actively working on understanding and trying to resolve the issue. You can check the state of the problem [[:phab:T210739|on Phabricator]]. In the meantime, sysops can work around this issue, whenever it occurs, by deleting the target page first, and then proceeding with the move as usual.