Page MenuHomePhabricator

Use swift directly from shellbox-media
Open, Needs TriagePublic

Description

We added support to shellbox to access files directly on swift. This was primarily done to improve the performance of video transcoding.

We however never made use of this for PdfHandler, so it seems that here we are still regularly POST'ing 300Mb to 1Gb pdf files instead.
This seems inefficient. It probably makes sense to use direct swift access for pdf/djvu and maybe even for pagedtiffhandler. For this we use:

$sourceFile->getRepo()->addShellboxInputFile()

instead of $cmd->inputFileFromFile()

See also: T292322

Event Timeline

Change #1310606 had a related patch set uploaded (by TheDJ; author: TheDJ):

[mediawiki/extensions/PdfHandler@master] Shellbox: fetch PDFs by storage url for metadata extraction

https://gerrit.wikimedia.org/r/1310606

Change #1310606 merged by jenkins-bot:

[mediawiki/extensions/PdfHandler@master] Shellbox: fetch PDFs by storage url for metadata extraction

https://gerrit.wikimedia.org/r/1310606