perf(ingester): invalidate head postings cache per-label#7678
Draft
kylestang wants to merge 1 commit into
Draft
Conversation
…er-metric The head expanded postings cache previously keyed a per-metric "seed" into the cache key, so creating or deleting any series for a metric invalidated every cached entry for that metric name, even entries whose result could not have changed. Replace the seed with a per-(tenant, label) write counter. On series create/delete each of the series' label counts is incremented. On creation, a cache entry snapshots the counts of its matched labels and stays valid as long as at least one tracked count is unchanged. Creating or deleting a series which matches all of the tracked labels expires the cache entries. Matchers that can match an absent label (e.g. foo!="x", foo=~".*") are excluded from the snapshot since a matching series need not carry the label, so they cannot vouch for an entry. Name-only queries fall back to the old whole-metric behaviour via the __name__ count. The counter array is shared across tenants and namespaced by userId; it is uint32 (8MB) and read under the existing stripe locks. Adds a reason="stale" value to the cache miss metric to distinguish count invalidation from TTL expiry. Signed-off-by: Kyle Stang <kylestng@amazon.com>
kylestang
force-pushed
the
kyle/label-counts
branch
from
ff63b21 to
fe2da16
Compare
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
The head expanded postings cache previously keyed a per-metric "seed" into the cache key, so creating or deleting any series for a metric invalidated every cached entry for that metric name, even entries whose result could not have changed.
Replace the seed with a per-(tenant, label) write counter. On series create/delete each of the series' label counts is incremented. On creation, a cache entry snapshots the counts of its matched labels and stays valid as long as at least one tracked count is unchanged. Creating or deleting a series which matches all of the tracked labels expires the cache entries.
Matchers that can match an absent label (e.g. foo!="x", foo=~".*") are excluded from the snapshot since a matching series need not carry the label, so they cannot vouch for an entry. Name-only queries fall back to the old whole-metric behaviour via the name count.
The counter array is shared across tenants and namespaced by userId; it is uint32 (8MB) and read under the existing stripe locks. Adds a reason="stale" value to the cache miss metric to distinguish count invalidation from TTL expiry.
What this PR does:
Which issue(s) this PR fixes:
Fixes #
Checklist
CHANGELOG.mdupdated - the order of entries should be[CHANGE],[FEATURE],[ENHANCEMENT],[BUGFIX]docs/configuration/v1-guarantees.mdupdated if this PR introduces experimental flags