[meta] Improve reports and cache performance investigation features

FWIW, I once created a CLI tool around this https://github.com/fgm/drupal_redis_stats

Thanks. The item count is already part of the report and has been for a long time. (additionally to render cache variations, read/write statistics, cache tag invalidations and more)

What I'm considering per most recent addition is the size as well, but only for the ChainedFast bins. I had a look at your implementation but honestly don't understand enough Go to parse that, so you're welcome to describe how exactly you calculate the size part, so I could implement that in PHP, i would probably just go with a strlen of the main data string (note that this supports compression, so would probably need to check the non-compressed version as apcu backend doesn't compress)

Actually, I'm just asking Redis for the size, as you can see on https://github.com/fgm/drupal_redis_stats/blob/main/stats/stats.go#L28

Ah, I see. with compression, I think that's not really what I'm after. The goal of "ChainedFast related checks: check apcu uptime/expunges, maybe estimate fast chained bin sizes vs configured apcu shm size." would be to tell you if e.g. config + discovery + bootstrap uncompressed cache sizes are above or close to apcu size as you will likely run into frequent memory resets. Less interested in the actual size of those items in Redis.

Just an idea: would it make sense to include compression (with a threshold) in the APCu handler ?

Possibly, see 📌 Make igbinary the default serializer if available, it saves 50% time on unserialize and memory footprint Active . Would need to use a serializer to make that pluggable.