Is the "SkinMinerva" now (after MW 1.30) separate out of Extension:MobileFrontend? I got same problem after upgrade to Mediawiki 1.30
Jan 4 2018
Jan 1 2018
Dec 31 2017
Cool! Thank you all & happy new year!
Dec 29 2017
Dec 24 2017
Solved by add 'wiki' => 'wiki-db-name-used-with-wgLBFactoryConf', to $wgForeignFileRepos = array(
Nov 21 2017
Same error happened again during MW 1.28 to 1.29 upgrade.
Jul 4 2017
I got similar error report in MW 1.28.2.
"PHP message: PHP Warning: Cannot modify header information - headers already sent by (output started at /var/www/wiki/includes/actions/RawAction.php:115) in /var/www/wiki/includes/WebResponse.php on line 45
May 8 2017
I don't have such problem before MW1.28...
Apr 10 2017
After some trial, we found that jobs been claimed but not run is reproduce-able when: php runJobs.php exit unusual, such as ssh connection lost, background php process killed due to high load, and etc.
Apr 9 2017
Apr 4 2017
When will this bug get fix? Using Redis instead of mysql to store jobs now.
However, We can't afford a large memory, which made jobs stored in Redis been truncated when reaching memory cap (2,000 to 3,000 jobs).
Apr 3 2017
Is this bug reintroduced into MW 1.28.0 with Abuse Filter version https://phabricator.wikimedia.org/r/revision/mediawiki/extensions/AbuseFilter.git;c3be1ce87f62d3e721af6d11720ad4c1e353ff9c ?
Mar 28 2017
I would suggest a systematic explanation about the conversion function documented in mediawiki.org or at least in the file's code comments.
Feb 15 2017
Feb 13 2017
BTW, it would be much easier to debug if ResourceLoaderImageModule could wrote some info to bug log when SVG rasterized process failed.
Some version of PHP7's GD (PHP: 7.0.15-1~dotdeb+8.1 (fpm-fcgi) for me) will crash with SVG rasterized under high load, use imagemagick instead solve problem.
The getDBkey() on null is reproduce-able if purposely messed up the SVG rasterized process (No matter using PHP's GD or imagemagick) (e.g. delete the convert file of imagemagick)
Feb 12 2017
What eventually work for us, is to switch SVG process function from PHP7's GD to imagemagick. SVG conversion by resourceloader failed with GD and seems resulting the getDBkey() on null error. Once we got the SVG conversion work with resourceloader, there are no more getDBkey() error reported.
Feb 11 2017
We have corrected all Map-values problem but still receive this error in php error log.
Hi @Jdlrobson, So I got sentry.io installed on machine and it shows me error url.
All right, it is T124196 . So for a temporary fix, I got all the job stored in Redis instead of mysql.
Feb 10 2017
I got nothing in mysql error.log.
Almost all cirrusSearchElasticaWrite type job failed due to this error.
Feb 8 2017
I will try....
Unfortunately I can't provide you more info now, because the PHP error log does not contain URL, I can't match the request from log in nginx with php.
Feb 7 2017
Feb 6 2017
I can't visit the link Jdlrobson provided, It require authentication.
Feb 4 2017
Using Echo-REL1_28-f55bdd9.tar.gz from https://www.mediawiki.org/wiki/Special:ExtensionDistributor?extdistname=Echo&extdistversion=REL1_28
Problem confirmed and solved (at least for us).
Feb 3 2017
Feb 2 2017
Feb 1 2017
There were 17369 sessions in uAPC (or DB or redis) within 10 mins. None of them get deleted. It just keep generating new session each page.
We tried $wgSessionCacheType to CACHE_ACCEL & CACHE_DB & redis, none of them solve problem.
Looks more like T149573 , where several servers keep changing session because each of them have a individual session set by wfMemcKey.
Jan 31 2017
Jan 14 2017
Dec 27 2016
Jan 29 2016
Yes and thank you Krinkle,
Jan 7 2016
Jan 3 2016
Jan 1 2016
I guess percent of any "String, int, bool, float, etc." will cause Fatal error.
Dec 31 2015
Nov 27 2015
Finally solved by delete everything in /wiki/vendor folder, then run composer install in /wiki folder.
I am using
Problem caused by extension EventLogging. download 1.26.0 stable version solve it. (Note: master version of this extension does not work for me)
Nov 26 2015
Nov 9 2015
Here is how we (MoegirlPeida http://moegirl.org/Mainpage ) solve the https problem over varnish & Mediawiki.
Aug 17 2015
We enable https and force http to ssl too. We got varnish listening port 8080 for normal data and 6082 for "management". which port should be used for $wgSquidServers?
For varnish 4.0, which port should I use? 80? 6082?
Aug 6 2015
Is it possible to support <img> tag, display image?
Aug 4 2015
Aug 3 2015
We (Moegirlpedia) use several different database. I am not sure how much huiji.wiki used.
Problem is user can't confirm their email.
It looks like caused by Timestamp. Engineer from huiji.wiki found that
in file User.php line 3672,
comment this line: //'user_touched' => $dbw->timestamp($oldTouched)
Looks quite similar. Our problem is "Invalid confirmation code" too. We believe this is caused by timestamp.
The time data didn't write into database. Then it will show "Invalid confirmation code"
Aug 1 2015
Jul 29 2015
Jun 10 2015
May 31 2015
A sorry. I forgot.
May 25 2015
MW 1.24.1 with the current master version SwiftMailer 5/25/2015, does not work with HHVM 3.6
Mar 8 2015
Jan 12 2015
Yes, tried this link: http://en.moegirl.org/Sandbox