Repository navigation
Prefault mmap'd datasets before timed upload - #184
Open
qdrant-cloud-bot wants to merge 1 commit into
Open
qdrant-cloud-bot wants to merge 1 commit into
qdrant-cloud-bot wants to merge 1 commit into
Conversation
Cold page faults of vectors.npy on FUSE/NTFS data disks were dominating upload_time (3–6× spikes). Fault every mapped page during reader open so the timed upsert loop measures ingest, not dataset I/O.
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Summary
vectors.npy(and fbin/h5 sidecars) were included in the timed upsert loop, soupload_timespiked 3–6× when the page cache was cold — especially on Azure/mntFUSE/NTFS data disks used by bench-platform..npy/ tar bundles, fbin, and h5 sidecars via a sharedprefault_mmaphelper.Context
Observed on Dataset index & search / upload_time_sec: ~65/73 dbpedia runs at 4.6–5.5s, 8 runs at 14–28s, then back to baseline. Search RPS stayed flat. Same image digest could be fast one night and slow the next.
Test plan
cargo test prefaultcargo test reads_vectors_of_a_valid_filecargo fmt --check/cargo clippy --all-targets -- -D warningsqdrant/bfb:devpublishes this change, confirm nightly dbpediaupload_time_secloses the 14–28s spike cluster