MediaWiki version: n/a
Uncaught Error: Cannot use object of type MediaWiki\Logger\Monolog\LegacyHandler as array in /srv/mediawiki/php-1.36.0-wmf.20/includes/exception/MWExceptionRenderer.php:109
Just a few of them logged in.
It is most probably related to the monolog update T242751
#0 /srv/mediawiki/php-1.36.0-wmf.20/includes/exception/MWExceptionRenderer.php(136): MWExceptionRenderer::useOutputPage(Object(WMFTimeoutException)) #1 /srv/mediawiki/php-1.36.0-wmf.20/includes/exception/MWExceptionRenderer.php(65): MWExceptionRenderer::reportHTML(Object(WMFTimeoutException)) #2 /srv/mediawiki/php-1.36.0-wmf.20/includes/exception/MWExceptionHandler.php(106): MWExceptionRenderer::output(Object(WMFTimeoutException), 2) #3 /srv/mediawiki/php-1.36.0-wmf.20/includes/exception/MWExceptionHandler.php(185): MWExceptionHandler::report(Object(WMFTimeoutException)) #4 /srv/mediawiki/php-1.36.0-wmf.20/includes/exception/MWExceptionHandler.php(156): MWExceptionHandler::handleException(Object(WMFTimeoutException), 'mwe_handler') #5 [internal function]: MWExceptionHandler::handleUncaughtException(Object(WMFTimeoutException))
Why is this a Parsoid issue? Nothing in the stack posted suggests it is... It's a MediaWiki-REST-API call with an uncaught exception which MW core than doesn't handle..
{"messageTranslations":{"de":"Der angeforderte relative Pfad (/de.wikipedia.org/v3/page/pagebundle/Liste_der_Orte_mit_Stolpersteinen/206202978) stimmte mit keinem bekannten Handler überein","en":"The requested relative path (/de.wikipedia.org/v3/page/pagebundle/Liste_der_Orte_mit_Stolpersteinen/206202978) did not match any known handler"},"httpCode":404,"httpReason":"Not Found"}
Both were ~6-7 minutes after the deploy (not saying the deploy is necessarily responsible). Not reproducible with the request query...
Happy to look into it further if we get some more evidence/examples/similar
Both were on wtp1026
Stack and error contain no code from either Parsoid nor REST API.
Seems indeed that there's something about Monolog 2 or core's exception handler that stopped working correctly. Or if it's a one off on wtp1026, maybe opcache corruption?
I filed it against Parsoid since the error was on a wtp* server and I haven't looked further. Anyway it was a single occurrence, not worth pursuing further.