The script is not batched, but using Special:PagesWithProp/score shows that en.wp only has 1100 pages with scores on them, commons not even 50 and en.wikisource.org about 2000 pages, so that shouldn't be a problem.

I will note that that doesn't directly mean there'll be only 1100 ly files in swift :)

Ok, so I've cleaned up (ie deleted) literally millions of old files, everything before 20210101000000.

So now we have this left:

reedy@mwmaint1002:~$ mwscript extensions/Score/maintenance/GetLYFiles.php --wiki=enwiki --date=20210101000000
Total files (all extensions): 251527
53321 ly files created on or after 20210101000000

Can confirm. thx

Wondering how we have millions of old files though...

$cacheOptions = [
				'code' => $code,
				'lang' => $options['lang'],
				'note-language' => $options['note-language'],
				'raw'  => $options['raw'],
				'ExtVersion' => self::CACHE_VERSION,
				'LyVersion' => self::getLilypondVersion(),
			];

Hmm, that creates a very unstable sha versioning i guess. We should probably have some sort of cleanup routine for this in place ?

In T293756#7453460, @TheDJ wrote:

Wondering how we have millions of old files though...

Every half created version that's previewed etc results in files on disk...

Hmm, that creates a very unstable sha versioning i guess. We should probably have some sort of cleanup routine for this in place ?

Indeed. I created the script for it last year when we had the lilypond issues (which is what I ran, in stages, before doing this). But haven't wired it into a regular cleanup cron/systemd timer (yet).

See also: T257093: Add cache key information to metadata json

In T257093#7311136, @fgiunchedi wrote:

Not strictly in scope for this task but I thought I'd mention it, score is a single container AFAICT (global-data-score-render) with 3.8M files and 300G (~78k average file size). If performance becomes a problem we can also consider sharding the container into multiple ones like big wikis e.g. global-data-score-render.aa (for comparison commons thumbs containers have average ~128k file size, at ~7M files and ~800G per container). In that case the two letter sharding comes handy to match media storage.