Anyone familiar with why large-ish (10s of MB, so ...
# developmentf
fresh-mechanic-68429
05/13/2025, 2:43 AMAnyone familiar with why large-ish (10s of MB, so stored outside of lmdb) files might be slow when pulling from a remote cache? My first thought is either locking (insided sharedlmdb?) or queueing on the tokio threads (behind other work on the blocking threadpool)
I've got ~10 files of this size that need to be pulled. A small cli app using reapi can pull them in 10s of milliseconds but pants is showing ~25 seconds. The flamegraph hasn't been too illuminating yet
fresh-mechanic-68429
05/13/2025, 2:46 AMThats not the whole flamegraph, but the relevant portion, the relevant code
ByteStore::load_monomorphicfresh-mechanic-68429
05/13/2025, 2:46 AMimage.png
b
broad-processor-92400
05/13/2025, 5:43 AMHmm, good question.
Can you share the demo CLI? And/or highlight what if anything it is missing compared to the Pants code?
3 Views