[szibis]

POST metadata reads could cache a body-filtered answer under a URL-only key. A later POST for a different app could then return the previous app's labels. This change keys effective form parameters, preserves body precedence and fences the affected old namespaces across RAM, disk, peers and the outer compatibility cache. It also serves GET/POST /loki/api/v1/label directly through the existing labels handler.

The read-scope/reporting changes from #948 are merged. This PR is rebased onto main at 4c5b769f96f1630dd69854449dd0e0f1872b1859; its diff contains only the alias, POST cache correction, regression tests and supporting evidence/docs.

Historical main versus PR evidence

Retained results, method and limits, harness and isolated compose. Historical measured main runtime: 28a5a2f6994240d85d254b5daa23dbc04ecfd891; candidate runtime: c320190c3b2cda0137b36518491e406bbc8835d3. This capture predates #947. The branch is now rebased onto main at 4c5b769 (including v2.54.0 and #948) and includes its S2 ownership change. These rows retain their original measured identities and are not measurements of the rebased head. Reference: Loki 3.7.9 and VictoriaLogs 1.52.0, the same dual-written TestSetup_IngestLogs fixture and a fixed one-hour selector window.

Signal	Historical main runtime	Historical PR runtime	Meaning
Singular labels alias, GET and POST	307 redirect	200, equal to plural and Loki	Closes the modern alias route gap
Canonical filtered POST immediately after matching GET	4 VL select requests	0	Equivalent forms reuse the existing metadata answer
POST for a second, missing app	Returned previous app's labels	Empty, like its GET control and Loki	Reproduces and fixes incorrect selector reuse
Backend cost for that new selector	0 VL selects, wrong answer	3 VL selects, correct answer	Correctness requires a new read
Alias after canonical priming	0 selects, redirect only	0 selects, correct labels	The labels metadata cache is shared; outer GET cache remains path-specific
Upgrade reuse of affected old keys	Ambiguous entries eligible	New namespaces fence them	Initial refill cost; old records follow existing expiry/eviction
Read inventory after scope correction	24/27 routed, 18 differential, 11 declared proven	25/27 routed, 19 differential, 12 declared proven	One alias; no new backend-native algorithm

The counter rows are one dedicated sequence, including prior endpoint-test priming. They are not a cold latency benchmark, per-tier hit ratios or physical disk operations. The empty-selector raw response differs in an existing way: Loki omits data; the proxy returns data: [].

Resource / state	Scope and result
Topology and RF	One main and one PR proxy with separate cache volumes; shared fixture; peers disabled, RF not applicable
Configured budgets, both builds	Primary L1 256 MiB/10,000 entries; additional GET cache 10%; logical SSD cache 512 MiB; Go memory target 2 GiB, not a process/container cap
CPU, RSS, allocations, retained occupancy, evictions, encoded disk traffic, physical IOPS, VL-server CPU/RSS	Not measured in this endpoint correctness proof; no resource-savings claim
Layer attribution	Labels are Loki-facing metadata; VL-facing cache reuse and independent tier hit/miss rates were not measured
p50/p95, shifted windows, refresh, disk restart, peer loss, mixed-version fleet and pressure	Not measured here; exact repeats do not prove those states

Tenant/auth and backend scoping, existing freshness/completeness rules and configured capacities are preserved. POST skips the URL-only key memo and copies the parsed form for normalization. The old-key fence applies to upgraded binaries; it is not a distributed deletion-invalidation protocol. cache/generation_numbers remains unimplemented. index/shards stays a counted optional planner-only gap; it does not block ordinary pinned Grafana/Drilldown queries or peer-cache lookup.

Verification

• Pre-rebase go test ./... -count=1: passed, including the proxy package (310 s). After rebasing onto perf: release oversized S2 cache backing buffers #947, focused race tests including POST/alias/tenant/compatibility regressions, stats extents and S2 disk-queue/restart tests passed (26.9 s). Full CI reruns on the updated head.
• Focused -race regressions: passed for POST form identities, selectors/windows, alias routing, old namespaces, compatibility cache and tenant/auth isolation. A malformed partial-form case also passes.
• HTTP upgrade regressions seed prior RAM, SSD and outer GET-cache answers and require fresh backend labels. Shared peer/stale key fencing is checked structurally by the versioned identities; this is not a live L3 migration claim.
• Live TestCompat_LabelsSingularAlias: passed against the committed candidate image and Loki, including nonempty direct GET/POST parity and a second empty selector. It runs in the existing TestCompat_.* core shard.
• Existing CI-script suite: 411 tests passed, including read-scope and generated-page regression tests. Production Docusaurus build passed. Conformance/registry and changelog gates passed; compose configuration validates.
• Independent code and proof review: no unresolved findings. Pre-rebase pinned Gitleaks scan: 80 locally available commits, no leaks; full remote-history scanning remains CI's job.

The retained harness is manually reproducible; it adds no CI workload solely for the PR comment. Existing product compatibility, visual smoke and regression workflows provide the CI evidence.

The README and four curated API-status blocks are regenerated and checked by the existing conformance gate. This tree shows 25/27 routed reads, 19/27 with differential evidence and 12/27 declared proven; the parent scope/report PR shows 24/27, 18/27 and 11/27. The singular alias leaves the missing-route list only in this product PR. Known Issues records the POST collision in the parent and the tested correction here. Optional deletion remains outside the read score; generation_numbers remains unimplemented and index/shards optional.

Release audit after v2.54.0: the released changelog is unchanged byte-for-byte; the POST/alias and security notes remain under Unreleased. Runtime source is unchanged by the docs/release rebases. The latest local pinned Gitleaks run before the release rebase scanned 101 available commits with no leaks; remote full-history checks rerun on the updated head. The report string-construction quality finding is fixed in the shared parent using explicit concatenation, with unchanged rendered output.

Conflict-resolution verification (2026-11-01): rebased only the five product commits onto merged #948. The resulting file tree is byte-for-byte identical to the previous PR head. Fresh focused POST/alias race regressions passed (36 tests); conformance and changelog gates passed; all five rebased commits have valid signatures. CI reruns on the updated head.

[github-actions]

⚪ Performance A/B: no change beyond noise

56 shape×range — 56 ⚪ same · results vs Loki: 49 same, 0 new difference(s), 7 pre-existing

Loki profiles compared (like for like):

• aligned: not run (no build supports -align-queries-with-step, so every aligned row would be not like-for-like)
• loki-defaults: 56 shape×range; proxy flag base off, pr off

Every shape is within noise of the base build (±30% and 100 ms).

56 unchanged shape×range (within noise)

	shape	range	Loki profile	base p50	PR p50	change	Loki p50	PR ÷ Loki	result vs Loki
⚪	drilldown service landing	1h	defaults	7 ms	6 ms	⚪ -14%	86 ms	0.07×	✅ same
⚪	instant ungrouped rate	instant	defaults	9 ms	11 ms	⚪ +22%	70 ms	0.16×	✅ same
⚪	json rate by level (no filter, no drop)	1h	defaults	400	400	⚪ +0%	400	—	✅ same
⚪	json volume no filters	1h	defaults	57 ms	54 ms	⚪ -5%	341 ms	0.16×	✅ same
⚪	logs plain	1h	defaults	53 ms	51 ms	⚪ -4%	435 ms	0.12×	✅ same
⚪	logs | json	1h	defaults	76 ms	75 ms	⚪ -1%	668 ms	0.11×	✅ same
⚪	stream-label count by app	1h	defaults	7 ms	7 ms	⚪ +0%	99 ms	0.07×	✅ same
⚪	topk rate by app	1h	defaults	9 ms	9 ms	⚪ +0%	134 ms	0.07×	➖ differs (as base)
⚪	label values env ending now	1h	defaults	9 ms	8 ms	⚪ -11%	20 ms	0.40×	✅ same
⚪	label values env ending two days ago (control)	1h	defaults	2 ms	1 ms	⚪ -50%	2 ms	0.50×	➖ differs (as base)
⚪	labels ending now	1h	defaults	6 ms	6 ms	⚪ +0%	2 ms	3.00×	✅ same
⚪	labels ending two days ago (control)	1h	defaults	2 ms	1 ms	⚪ -50%	2 ms	0.50×	➖ differs (as base)
⚪	E1 json duration_ms sum by service	1h	defaults	6 ms	6 ms	⚪ +0%	14 ms	0.43×	✅ same
⚪	E2 json duration_ms p95 by service	1h	defaults	9 ms	9 ms	⚪ +0%	12 ms	0.75×	✅ same
⚪	E3 logfmt latency_ms rate, production	1h	defaults	10 ms	9 ms	⚪ -10%	88 ms	0.10×	➖ differs (as base)
⚪	E4 logfmt size_bytes max by service	1h	defaults	6 ms	5 ms	⚪ -17%	11 ms	0.45×	✅ same
⚪	E6 logfmt ttl unit string (Loki 400)	1h	defaults	400	400	⚪ -9%	400	—	✅ same
⚪	E7 json hexadecimal trace_id (Loki 400)	1h	defaults	400	400	⚪ -15%	400	—	✅ same
⚪	E8 json duration_ms sum by service, namespace prod	1h	defaults	8 ms	7 ms	⚪ -12%	67 ms	0.10×	✅ same
⚪	E9 logfmt latency_ms sum by service, namespace prod	1h	defaults	8 ms	8 ms	⚪ +0%	64 ms	0.12×	✅ same
⚪	D1 logfmt ttl duration_seconds rate by service	1h	defaults	12 ms	11 ms	⚪ -8%	9 ms	1.22×	✅ same
⚪	D2 logfmt ttl duration rate by service	1h	defaults	1 ms	1 ms	⚪ +0%	7 ms	0.14×	✅ same
⚪	G1 logfmt amount sum_over_time, no filter	1h	defaults	4 ms	3 ms	⚪ -25%	12 ms	0.25×	✅ same
⚪	G10 logfmt latency_ms sum_over_time, production	1h	defaults	9 ms	8 ms	⚪ -11%	88 ms	0.09×	➖ differs (as base)
⚪	G2 json duration_ms sum_over_time, no filter	1h	defaults	5 ms	4 ms	⚪ -20%	25 ms	0.16×	✅ same
⚪	G3 logfmt amount sum_over_time, error filter	1h	defaults	4 ms	3 ms	⚪ -25%	10 ms	0.30×	✅ same
⚪	G4 json duration_ms sum_over_time, error filter	1h	defaults	5 ms	4 ms	⚪ -20%	22 ms	0.18×	✅ same
⚪	G5 logfmt amount sum by rate	1h	defaults	4 ms	4 ms	⚪ +0%	9 ms	0.44×	✅ same
⚪	G6 json duration_ms sum by rate	1h	defaults	5 ms	4 ms	⚪ -20%	22 ms	0.18×	✅ same
⚪	G7 duration() of a unit string	1h	defaults	11 ms	10 ms	⚪ -9%	9 ms	1.11×	✅ same
⚪	G8 bytes() of a number	1h	defaults	8 ms	8 ms	⚪ +0%	8 ms	1.00×	✅ same
⚪	G9 json duration_ms max_over_time	1h	defaults	7 ms	5 ms	⚪ -29%	85 ms	0.06×	✅ same
⚪	W1 sum without pod, max_over_time	1h	defaults	5 ms	5 ms	⚪ +0%	14 ms	0.36×	✅ same
⚪	W2 sum without pod, sum_over_time	1h	defaults	5 ms	5 ms	⚪ +0%	12 ms	0.42×	✅ same
⚪	W3 sum without pod, rate	1h	defaults	7 ms	5 ms	⚪ -29%	12 ms	0.42×	✅ same
⚪	W4 max without pod, max_over_time	1h	defaults	2 ms	2 ms	⚪ +0%	12 ms	0.17×	✅ same
⚪	W7 topk 5 of sum without pod	1h	defaults	2 ms	2 ms	⚪ +0%	12 ms	0.17×	✅ same
⚪	W8 sum without pod, count_over_time (production)	1h	defaults	15 ms	14 ms	⚪ -7%	42 ms	0.33×	✅ same
⚪	U1 logfmt amount rate	1h	defaults	5 ms	5 ms	⚪ +0%	12 ms	0.42×	✅ same
⚪	U2 json duration_ms rate	1h	defaults	6 ms	5 ms	⚪ -17%	21 ms	0.24×	✅ same
⚪	U3 json duration_ms rate, error filter	1h	defaults	6 ms	6 ms	⚪ +0%	20 ms	0.30×	✅ same
⚪	U4 logfmt latency_ms rate, production	1h	defaults	9 ms	9 ms	⚪ +0%	90 ms	0.10×	➖ differs (as base)
⚪	U8 logfmt amount rate by service (grouped)	1h	defaults	6 ms	5 ms	⚪ -17%	10 ms	0.50×	✅ same
⚪	U9 logfmt amount rate_counter (folded per series)	1h	defaults	6 ms	6 ms	⚪ +0%	23 ms	0.26×	✅ same
⚪	U1 json duration_ms sum by service	1h	defaults	6 ms	6 ms	⚪ +0%	15 ms	0.40×	✅ same
⚪	U12 json duration_ms over every service	1h	defaults	8 ms	8 ms	⚪ +0%	42 ms	0.19×	✅ same
⚪	U13 logfmt latency_ms over every service	1h	defaults	18 ms	17 ms	⚪ -6%	112 ms	0.15×	➖ differs (as base)
⚪	U14 duration() of a unit string	1h	defaults	13 ms	13 ms	⚪ +0%	6 ms	2.17×	✅ same
⚪	U2 json duration_ms max by service	1h	defaults	6 ms	6 ms	⚪ +0%	26 ms	0.23×	✅ same
⚪	U3 json duration_ms sliding sum	1h	defaults	12 ms	9 ms	⚪ -25%	45 ms	0.20×	✅ same
⚪	U4 logfmt size_bytes sum by service	1h	defaults	7 ms	6 ms	⚪ -14%	7 ms	0.86×	✅ same
⚪	U5 logfmt ttl unit string (no sample)	1h	defaults	6 ms	5 ms	⚪ -17%	5 ms	1.00×	✅ same
⚪	U6 missing label, no parser	1h	defaults	10 ms	10 ms	⚪ +0%	9 ms	1.11×	✅ same
⚪	U7 missing label, ungrouped sum	1h	defaults	5 ms	5 ms	⚪ +0%	9 ms	0.56×	✅ same
⚪	U8 rate of an unwrapped label	1h	defaults	6 ms	6 ms	⚪ +0%	12 ms	0.50×	✅ same
⚪	U9 bytes() conversion	1h	defaults	10 ms	10 ms	⚪ +0%	9 ms	1.11×	✅ same

Cold (first-run) timings — what a user opening a panel sees

shape	range	Loki profile	base cold	PR cold	Loki cold	PR ÷ Loki cold
drilldown service landing	1h	defaults	21 ms	19 ms	390 ms	0.05×
instant ungrouped rate	instant	defaults	11 ms	12 ms	74 ms	0.16×
json rate by level (no filter, no drop)	1h	defaults	400	400	400	—
json volume no filters	1h	defaults	164 ms	165 ms	1.24 s	0.13×
logs plain	1h	defaults	61 ms	49 ms	449 ms	0.11×
logs | json	1h	defaults	115 ms	115 ms	731 ms	0.16×
stream-label count by app	1h	defaults	17 ms	16 ms	394 ms	0.04×
topk rate by app	1h	defaults	15 ms	17 ms	415 ms	0.04×
label values env ending now	1h	defaults	57 ms	56 ms	21 ms †	—
label values env ending two days ago (control)	1h	defaults	4 ms	4 ms	2 ms †	—
labels ending now	1h	defaults	20 ms	18 ms	3 ms †	—
labels ending two days ago (control)	1h	defaults	3 ms	3 ms	2 ms †	—
E1 json duration_ms sum by service	1h	defaults	63 ms	49 ms	130 ms	0.38×
E2 json duration_ms p95 by service	1h	defaults	108 ms	124 ms	144 ms	0.86×
E3 logfmt latency_ms rate, production	1h	defaults	302 ms	289 ms	662 ms	0.44×
E4 logfmt size_bytes max by service	1h	defaults	37 ms	27 ms	117 ms	0.23×
E6 logfmt ttl unit string (Loki 400)	1h	defaults	400	400	400	—
E7 json hexadecimal trace_id (Loki 400)	1h	defaults	400	400	400	—
E8 json duration_ms sum by service, namespace prod	1h	defaults	167 ms	183 ms	418 ms	0.44×
E9 logfmt latency_ms sum by service, namespace prod	1h	defaults	186 ms	189 ms	473 ms	0.40×
D1 logfmt ttl duration_seconds rate by service	1h	defaults	44 ms	44 ms	34 ms	1.29×
D2 logfmt ttl duration rate by service	1h	defaults	1 ms	1 ms	31 ms	0.03×
G1 logfmt amount sum_over_time, no filter	1h	defaults	30 ms	24 ms	50 ms	0.48×
G10 logfmt latency_ms sum_over_time, production	1h	defaults	279 ms	280 ms	767 ms	0.37×
G2 json duration_ms sum_over_time, no filter	1h	defaults	41 ms	39 ms	166 ms	0.23×
G3 logfmt amount sum_over_time, error filter	1h	defaults	26 ms	24 ms	52 ms	0.46×
G4 json duration_ms sum_over_time, error filter	1h	defaults	56 ms	58 ms	166 ms ↺	0.35×
G5 logfmt amount sum by rate	1h	defaults	15 ms	14 ms	45 ms ↺	0.31×
G6 json duration_ms sum by rate	1h	defaults	51 ms	52 ms	170 ms	0.31×
G7 duration() of a unit string	1h	defaults	41 ms	39 ms	41 ms	0.95×
G8 bytes() of a number	1h	defaults	40 ms	37 ms	38 ms	0.97×
G9 json duration_ms max_over_time	1h	defaults	59 ms	52 ms	1.12 s	0.05×
W1 sum without pod, max_over_time	1h	defaults	43 ms	35 ms	64 ms	0.55×
W2 sum without pod, sum_over_time	1h	defaults	38 ms	37 ms	66 ms	0.56×
W3 sum without pod, rate	1h	defaults	56 ms	44 ms	60 ms	0.73×
W4 max without pod, max_over_time	1h	defaults	20 ms	16 ms	68 ms	0.24×
W7 topk 5 of sum without pod	1h	defaults	17 ms	18 ms	66 ms	0.27×
W8 sum without pod, count_over_time (production)	1h	defaults	87 ms	88 ms	524 ms	0.17×
U1 logfmt amount rate	1h	defaults	28 ms	22 ms	37 ms	0.59×
U2 json duration_ms rate	1h	defaults	50 ms	47 ms	96 ms	0.49×
U3 json duration_ms rate, error filter	1h	defaults	47 ms	47 ms	106 ms	0.44×
U4 logfmt latency_ms rate, production	1h	defaults	199 ms	196 ms	757 ms	0.26×
U8 logfmt amount rate by service (grouped)	1h	defaults	25 ms	24 ms	45 ms	0.53×
U9 logfmt amount rate_counter (folded per series)	1h	defaults	29 ms	30 ms	184 ms †	—
U1 json duration_ms sum by service	1h	defaults	58 ms	50 ms	163 ms	0.31×
U12 json duration_ms over every service	1h	defaults	149 ms	138 ms	575 ms	0.24×
U13 logfmt latency_ms over every service	1h	defaults	284 ms	342 ms	931 ms	0.37×
U14 duration() of a unit string	1h	defaults	64 ms	50 ms	42 ms	1.19×
U2 json duration_ms max by service	1h	defaults	53 ms	57 ms	537 ms	0.11×
U3 json duration_ms sliding sum	1h	defaults	6 ms	6 ms	166 ms	0.04×
U4 logfmt size_bytes sum by service	1h	defaults	41 ms	45 ms	38 ms	1.18×
U5 logfmt ttl unit string (no sample)	1h	defaults	33 ms	26 ms	33 ms †	—
U6 missing label, no parser	1h	defaults	30 ms	30 ms	67 ms †	—
U7 missing label, ungrouped sum	1h	defaults	10 ms	10 ms	68 ms †	—
U8 rate of an unwrapped label	1h	defaults	51 ms	51 ms	166 ms	0.31×
U9 bytes() conversion	1h	defaults	37 ms	37 ms	39 ms	0.95×

† Loki's results cache answered this cold run (its own counters, or identical answers for windows a minute apart): not a computed time, so it is in no ratio.

↺ Loki cold taken from the first set that ran the same query; Loki's cache answered this one.

Why these shapes

• internal/proxy/proxy.go → 44: unwrap-conversion-error/E1 json duration_ms sum by service, unwrap-conversion-error/E2 json duration_ms p95 by service, unwrap-conversion-error/E3 logfmt latency_ms rate, production, unwrap-conversion-error/E4 logfmt size_bytes max by service, unwrap-conversion-error/E6 logfmt ttl unit string (Loki 400), unwrap-conversion-error/E7 json hexadecimal trace_id (Loki 400), unwrap-conversion-error/E8 json duration_ms sum by service, namespace prod, unwrap-conversion-error/E9 logfmt latency_ms sum by service, namespace prod …
• control smoke → 8: control/drilldown service landing, control/instant ungrouped rate, control/json rate by level (no filter, no drop), control/json volume no filters, control/logs plain, control/logs | json, control/stream-label count by app, control/topk rate by app
• internal/proxy/proxy.go (endpoint handler) → 8: control/drilldown service landing, control/instant ungrouped rate, control/json rate by level (no filter, no drop), control/json volume no filters, control/logs plain, control/logs | json, control/stream-label count by app, control/topk rate by app
• internal/proxy/cache_keys.go → 4: metadata-freshness/label values env ending now, metadata-freshness/label values env ending two days ago (control), metadata-freshness/labels ending now, metadata-freshness/labels ending two days ago (control)

_{base 4c5b769f96 → PR 51a5dc7fae · 4 runs (first is cold), warm p50 · ranges 1h · noise ±30% and 100 ms · VictoriaLogs restarts 0→0 · 3.0 min · phases stack_up 39s, seed_and_build 42s, loki-defaults:loki_profile 0s, loki-defaults:proxies 2s, loki-defaults:matrix 89s. Same runner, one fresh stack, targets interleaved per request. How to read and reproduce: bench/ab/README.md.}

[github-actions]

✅ Visual smoke: passed

Base 4c5b769f96 vs PR 87728646b2, Grafana through the proxy builds and Loki on one fresh stack; run.
Core set: 9 capture(s) at 1h. Detailed set: 0 entries (nothing visual touched). 9 capture(s), 15m/1h/6h, 24h and 7d stay out of PR runs.

page	range	set	base = PR (data)	vs Loki (PR)	pixel diff	result
dd-fields	1h	core	✅ identical	identical (4 explained, 1 history-dependent)	0.11%	✅
dd-landing	1h	core	✅ identical	identical (3 explained)	0.14%	✅
dd-service-logs	1h	core	✅ identical	identical (3 explained, 1 history-dependent)	0.11%	✅
explore-logs-plain	1h	core	✅ identical	identical (1 explained)	0.07%	✅
explore-metric-plain	1h	core	✅ identical	identical (1 explained)	0.04%	✅
explore-sum-by-over-binary	1h	core	✅ identical	identical (1 explained)	0.04%	✅
explore-topk-by-per-step	1h	core	✅ identical	identical (1 explained)	0.04%	✅
explore-unwrap-sum	1h	core	✅ identical	identical (1 explained)	0.04%	✅
lvl-explore-rate-errors-by-service	1h	core	✅ identical	1 difference(s), 1 on base too; 1 explained	0.04%	✅

Gate: a base-vs-PR data difference passes when the PR is closer to Loki, and fails when the PR diverges from Loki or Loki cannot decide it (at a range Loki does not hold, a difference passes when the same page proved that query to answer like Loki at a shorter range; otherwise label the pull request visual-change-expected to accept an intended change). Also failing: a panel empty on the PR but not on the base, an error banner, panel error, error answer or new "No data" panel on the PR, a PR side that never settled, a value difference that did not reproduce on the recapture. A delayed first-pass request and pixel differences only warn.

Core set montages (9): base | PR | Loki

dd-fields-1h

dd-landing-1h

dd-service-logs-1h

explore-logs-plain-1h

explore-metric-plain-1h

explore-sum-by-over-binary-1h

explore-topk-by-per-step-1h

explore-unwrap-sum-1h

lvl-explore-rate-errors-by-service-1h

Published inline montages: pr-visuals/pr-949. Full montage set: artifact.

[github-actions]

PR Quality Report

Compared with the PR base commit; measurements below are scoped to their linked run.

Base 4c5b769 → PR 8772864; tested merge 51a5dc7.

Focus signals:

• Package statement coverage decreased 0.08 percentage points: github.com/ReliablyObserve/Loki-VL-proxy/internal/cache.
• visual: error / success; Evidence could not be read safely; see the original CI run.. run
• Exact case comparison: 36 new passing (9 top-level, 27 subtests); github.com/ReliablyObserve/Loki-VL-proxy/internal/proxy TestCanonicalReadCacheKeyPOSTFormMatchesGET, github.com/ReliablyObserve/Loki-VL-proxy/internal/proxy TestCanonicalReadCacheKeyPOSTFormMatchesGET/detected_field_values, github.com/ReliablyObserve/Loki-VL-proxy/internal/proxy TestCanonicalReadCacheKeyPOSTFormMatchesGET/detected_fields.
• Declared proven endpoints increased: 11 → 12; declared inventory trend, not runtime proof.
• Declared API inventory: 25/27 read endpoints implemented, 19 linked to Loki comparison tests, 12 declared proven; 67 open compatibility gaps. These are registry counts, not a runtime pass percentage.
• A/B selected workload: 56 same; 0 new and 7 pre-existing Loki answer differences. run
• Above 2× Loki: 0 new, 0 worsened, 0 known, 0 cleared, 0 locked, 0 lost below (warning only; green check does not clear gaps). run
• Existing CI test evidence: 21202 passed, 0 failed, 72 skipped; scope: internal packages (existing CI artifact). run
• Broader API/Grafana execution: 17 scopes with retained test outcomes; 0 contain failing cases, 4 contain skips and 0 report flaky/incomplete cases. Candidate-only proof, not a full compatibility percentage.

Main-versus-PR whole-process CPU, peak RSS, physical SSD throughput/IOPS and per-layer evictions are not measured here. Synthetic allocations and fleet cache occupancy do not substitute for them.

Evidence sources and revision status

Source	State	Result	Link	Notes
quality	completed	success	run
ci	completed	success	run
perf	completed	success	run
visual	error	success	run	Evidence could not be read safely; see the original CI run.
drilldown	completed	success	run
loki	completed	success	run
security	completed	success	run
vl	completed	success	run

Pending, stale, unavailable and skipped sources contribute no measurements.

Coverage and tests

Signal	Base	PR	Delta
Test count	21362	21398	36
Coverage	89.6%	89.6%	0.0% (stable)

Test count includes passing Go test/subtest events, not independent scenarios.

Current CI: 21202 passed, 0 failed, 72 skipped, 2984 top level passed, 18218 subtests passed, 11 packages.
Result scope: internal packages (existing CI artifact).
Cache-package coverage: 88.8932% (candidate only).

API contract coverage and regression changes

Exact Go case outcomes from the existing quality test run. Top-level tests and subtests are separate populations; package + full test name defines identity.

Observed case outcomes	Base	PR
top-level pass	3104	3113
top-level fail	0	0
top-level skip	2	2
subtests pass	18258	18285
subtests fail	0	0
subtests skip	63	63

Showing 6/6 rows.

Change	Package	Full test name	Population	Base	PR
new passing	github.com/ReliablyObserve/Loki-VL-proxy/internal/proxy	TestCanonicalReadCacheKeyPOSTFormMatchesGET	top-level	not present	pass
new passing	github.com/ReliablyObserve/Loki-VL-proxy/internal/proxy	TestCanonicalReadCacheKeyPOSTFormMatchesGET/detected_field_values	subtests	not present	pass
new passing	github.com/ReliablyObserve/Loki-VL-proxy/internal/proxy	TestCanonicalReadCacheKeyPOSTFormMatchesGET/detected_fields	subtests	not present	pass
new passing	github.com/ReliablyObserve/Loki-VL-proxy/internal/proxy	TestCanonicalReadCacheKeyPOSTFormMatchesGET/detected_labels	subtests	not present	pass
new passing	github.com/ReliablyObserve/Loki-VL-proxy/internal/proxy	TestCanonicalReadCacheKeyPOSTFormMatchesGET/index_stats	subtests	not present	pass
new passing	github.com/ReliablyObserve/Loki-VL-proxy/internal/proxy	TestCanonicalReadCacheKeyPOSTFormMatchesGET/label_values	subtests	not present	pass
new passing	github.com/ReliablyObserve/Loki-VL-proxy/internal/proxy	TestCanonicalReadCacheKeyPOSTFormMatchesGET/labels	subtests	not present	pass
new passing	github.com/ReliablyObserve/Loki-VL-proxy/internal/proxy	TestCanonicalReadCacheKeyPOSTFormMatchesGET/volume	subtests	not present	pass
new passing	github.com/ReliablyObserve/Loki-VL-proxy/internal/proxy	TestCanonicalReadCacheKeyPOSTFormMatchesGET/volume_range	subtests	not present	pass
new passing	github.com/ReliablyObserve/Loki-VL-proxy/internal/proxy	TestCanonicalReadCacheKeyPOSTFormPrecedenceAndImmutable	top-level	not present	pass
new passing	github.com/ReliablyObserve/Loki-VL-proxy/internal/proxy	TestCanonicalReadCacheKeyPOSTFormSeparatesSelectorsAndWindows	top-level	not present	pass
new passing	github.com/ReliablyObserve/Loki-VL-proxy/internal/proxy	TestCanonicalReadCacheKeyPOSTMigrationRejectsPriorNamespace	top-level	not present	pass
new passing	github.com/ReliablyObserve/Loki-VL-proxy/internal/proxy	TestCanonicalReadCacheKeyPOSTMigrationRejectsPriorNamespace/detected_field_values	subtests	not present	pass
new passing	github.com/ReliablyObserve/Loki-VL-proxy/internal/proxy	TestCanonicalReadCacheKeyPOSTMigrationRejectsPriorNamespace/detected_fields	subtests	not present	pass
new passing	github.com/ReliablyObserve/Loki-VL-proxy/internal/proxy	TestCanonicalReadCacheKeyPOSTMigrationRejectsPriorNamespace/detected_labels	subtests	not present	pass
new passing	github.com/ReliablyObserve/Loki-VL-proxy/internal/proxy	TestCanonicalReadCacheKeyPOSTMigrationRejectsPriorNamespace/index_stats	subtests	not present	pass
new passing	github.com/ReliablyObserve/Loki-VL-proxy/internal/proxy	TestCanonicalReadCacheKeyPOSTMigrationRejectsPriorNamespace/label_inventory	subtests	not present	pass
new passing	github.com/ReliablyObserve/Loki-VL-proxy/internal/proxy	TestCanonicalReadCacheKeyPOSTMigrationRejectsPriorNamespace/label_values	subtests	not present	pass
new passing	github.com/ReliablyObserve/Loki-VL-proxy/internal/proxy	TestCanonicalReadCacheKeyPOSTMigrationRejectsPriorNamespace/labels	subtests	not present	pass
new passing	github.com/ReliablyObserve/Loki-VL-proxy/internal/proxy	TestCanonicalReadCacheKeyPOSTMigrationRejectsPriorNamespace/volume	subtests	not present	pass

Showing 20/36 rows.

Per-package covered/total instrumented statements; missing packages or snapshots have no coverage delta. Coverage changes also reflect changes in the statement denominator.

Package	Base covered/total	PR covered/total	Change
github.com/ReliablyObserve/Loki-VL-proxy/internal/cache	2276/2557 (89.01%)	2274/2557 (88.93%)	-0.08 pp
github.com/ReliablyObserve/Loki-VL-proxy/cmd/configdoc	0/26 (0.00%)	0/26 (0.00%)	+0.00 pp
github.com/ReliablyObserve/Loki-VL-proxy/cmd/healthcheck	0/9 (0.00%)	0/9 (0.00%)	+0.00 pp
github.com/ReliablyObserve/Loki-VL-proxy/cmd/proxy	891/993 (89.73%)	891/993 (89.73%)	+0.00 pp
github.com/ReliablyObserve/Loki-VL-proxy/cmd/rules-migrate	37/42 (88.10%)	37/42 (88.10%)	+0.00 pp
github.com/ReliablyObserve/Loki-VL-proxy/cmd/validate_check	0/19 (0.00%)	0/19 (0.00%)	+0.00 pp
github.com/ReliablyObserve/Loki-VL-proxy/internal/config	206/232 (88.79%)	206/232 (88.79%)	+0.00 pp
github.com/ReliablyObserve/Loki-VL-proxy/internal/logql	1786/1951 (91.54%)	1786/1951 (91.54%)	+0.00 pp
github.com/ReliablyObserve/Loki-VL-proxy/internal/logsql	1134/1415 (80.14%)	1134/1415 (80.14%)	+0.00 pp
github.com/ReliablyObserve/Loki-VL-proxy/internal/memlimit	30/49 (61.22%)	30/49 (61.22%)	+0.00 pp
github.com/ReliablyObserve/Loki-VL-proxy/internal/metrics	2174/2300 (94.52%)	2174/2300 (94.52%)	+0.00 pp
github.com/ReliablyObserve/Loki-VL-proxy/internal/middleware	415/454 (91.41%)	415/454 (91.41%)	+0.00 pp
github.com/ReliablyObserve/Loki-VL-proxy/internal/observability	128/141 (90.78%)	128/141 (90.78%)	+0.00 pp
github.com/ReliablyObserve/Loki-VL-proxy/internal/rulesmigrate	83/89 (93.26%)	83/89 (93.26%)	+0.00 pp
github.com/ReliablyObserve/Loki-VL-proxy/internal/translator	3052/3285 (92.91%)	3052/3285 (92.91%)	+0.00 pp
github.com/ReliablyObserve/Loki-VL-proxy/internal/proxy	24688/27599 (89.45%)	24697/27606 (89.46%)	+0.01 pp

Showing 16/16 rows.

Declared conformance inventory, not full runtime compatibility. Registry states describe the checked-in contract and proof links; these counts do not certify that every endpoint, version or workload ran successfully.

Loki surface version: v3.7.7 → v3.7.7.

Read API scope: write endpoints are excluded and deletion management is reported separately. POST query transports remain reads. Mixed rule endpoints retain their read methods after registry exclusions. Deprecated read aliases are excluded by owner scope decision; this changes the denominator without proving a compatibility gain.

Declared endpoint inventory	Base (read API)	PR (read API)	Change
Read endpoints: declared total	27	27	+0
Read endpoints: declared implemented	24	25	+1
Read endpoints: linked differential tests	18	19	+1
Read endpoints: declared proven	11	12	+1

Showing 4/4 rows.

Outside the read endpoint score	Base	PR	Change
Write-only endpoints excluded	2	2	+0
Deprecated read aliases excluded	5	5	+0
Deletion management: declared total	1	1	+0
Deletion management: declared implemented	1	1	+0
Deletion management: linked differential tests	0	0	+0
Deletion management: declared proven	0	0	+0

Showing 6/6 rows.

Declared track states	Base	PR	Change
behaviours: gap	3	3	+0
behaviours: not-applicable	0	0	+0
behaviours: partial	17	17	+0
behaviours: proven	6	6	+0
behaviours: unknown	0	0	+0
behaviours: waived	0	0	+0
translations: gap	0	0	+0
translations: not-applicable	0	0	+0
translations: partial	2	2	+0
translations: proven	0	0	+0
translations: unknown	0	0	+0
translations: waived	0	0	+0

Showing 12/12 rows.

Declared gaps / error layer	Base	PR	Change
Declared gaps: open	67	67	+0
Declared gaps: fixed	57	57	+0
Declared gaps: documented	12	12	+0
Declared gaps: owner-kept	6	6	+0
Declared gaps: upstream	2	2	+0
Known error-layer leak cells	57	57	+0
Known error-layer leak links	74	74	+0

Showing 7/7 rows.

Declared item	Transition	Base state	PR state
endpoint:loki_api_v1_label	changed	gap	proven

Showing 1/1 rows.

Selected smoke component checks against the candidate backend only. Both quality snapshot invocations share that backend image, so these checks provide no runtime comparison between revisions. Use the real workload A/B evidence for separately built revisions. This is a small contract sample, not the full declared inventory or version matrix.

Track	Component	PR passing
drilldown	detected_fields	11/11 (100.00%)
drilldown	label_values	1/1 (100.00%)
drilldown	level_volume	2/2 (100.00%)
drilldown	patterns	1/1 (100.00%)
drilldown	service_logs	1/1 (100.00%)
drilldown	service_selection	1/1 (100.00%)
loki	label_values	2/2 (100.00%)
loki	labels	4/4 (100.00%)
loki	metrics	2/2 (100.00%)
loki	otel	1/1 (100.00%)
loki	query_range	1/1 (100.00%)
loki	series	1/1 (100.00%)
vl	detected_fields	4/4 (100.00%)
vl	field_values	3/3 (100.00%)
vl	index_stats	1/1 (100.00%)
vl	stream_translation	1/1 (100.00%)
vl	synthetic_labels	1/1 (100.00%)
vl	volume_range	1/1 (100.00%)

Showing 18/18 rows.

Broader API and Grafana execution scopes

Existing CI job logs, verified against the tested revision. Candidate only: these scopes have no paired base run here. Go counts include top-level cases and subtests; repeated identities count once and a failed repeat stays failed. Playwright invocations remain separate. Do not sum profiles as independent API coverage. Failed jobs can contain only partial observed populations.

CI source / scope	Evidence	Passing	Failing	Skipped	Go pass parent / child	Flaky / interrupted / not run	Logged stack / note
ci / e2e-compat (core)	available	4450	0	87	191 / 4259	not recorded	PROXY_IMAGE=loki-vl-proxy:e2e-ci
ci / e2e-compat (drilldown)	available	95	0	0	28 / 67	not recorded	PROXY_IMAGE=loki-vl-proxy:e2e-ci
ci / e2e-compat (otel-edge)	available	136	0	0	74 / 62	not recorded	PROXY_IMAGE=loki-vl-proxy:e2e-ci
ci / e2e-compat (semantics)	available	928	0	3	49 / 879	not recorded	PROXY_IMAGE=loki-vl-proxy:e2e-ci
ci / e2e-compat (tail-multitenancy)	available	45	0	0	36 / 9	not recorded	PROXY_IMAGE=loki-vl-proxy:e2e-ci
ci / e2e-fleet	available	56	0	0	9 / 47	not recorded	PROXY_IMAGE=loki-vl-proxy:fleet-ci
ci / e2e-ui (comprehensive-drilldown-mt) / invocation 1	available	22	0	0	—	0 / 0 / 0	PROXY_IMAGE=loki-vl-proxy:e2e-ui
ci / e2e-ui (drilldown-core-tail) / invocation 1	available	25	0	0	—	0 / 0 / 0	PROXY_IMAGE=loki-vl-proxy:e2e-ui
ci / e2e-ui (explore-core) / invocation 1	available	9	0	0	—	0 / 0 / 0	PROXY_IMAGE=loki-vl-proxy:e2e-ui
ci / e2e-ui (ops-mt) / invocation 1	available	15	0	1	—	0 / 0 / 0	PROXY_IMAGE=loki-vl-proxy:e2e-ui
ci / e2e-ui (regression-datasource) / invocation 1	available	41	0	10	—	0 / 0 / 0	PROXY_IMAGE=loki-vl-proxy:e2e-ui
ci / e2e-ui (regression-datasource) / invocation 2	available	3	0	0	—	0 / 0 / 0	PROXY_IMAGE=loki-vl-proxy:e2e-ui
drilldown / drilldown-grafana-pr-matrix (12.4.12, previous_smoke, true, ^(TestGrafanaDatasourceCatalogAndHeal...	available	49	0	0	3 / 46	not recorded	LOKI_IMAGE=grafana/loki:3.7.9, VICTORIALOGS_IMAGE=victoriametrics/victoria-logs:v1.52.0, GRAFANA_IMAGE=grafana/grafana:12.4.12, PROXY_IMAGE=loki-vl-proxy:compat-drilldown
drilldown / drilldown-grafana-pr-matrix (13.2.3, current_smoke, true, ^(TestGrafanaDatasourceCatalogAndHealth...	available	49	0	0	3 / 46	not recorded	LOKI_IMAGE=grafana/loki:3.7.9, VICTORIALOGS_IMAGE=victoriametrics/victoria-logs:v1.52.0, GRAFANA_IMAGE=grafana/grafana:13.2.3, PROXY_IMAGE=loki-vl-proxy:compat-drilldown
drilldown / drilldown-pinned-runtime	available	1	0	0	1 / 0	not recorded	LOKI_IMAGE=grafana/loki:3.7.9, VICTORIALOGS_IMAGE=victoriametrics/victoria-logs:v1.52.0, GRAFANA_IMAGE=grafana/grafana:13.2.3, PROXY_IMAGE=loki-vl-proxy:compat-drilldown
loki / loki-pinned	available	79	0	0	15 / 64	not recorded	LOKI_IMAGE=grafana/loki:3.7.9, VICTORIALOGS_IMAGE=victoriametrics/victoria-logs:v1.52.0, GRAFANA_IMAGE=grafana/grafana:13.2.3, PROXY_IMAGE=loki-vl-proxy:compat-loki
vl / vl-pinned	available	3	0	0	3 / 0	not recorded	LOKI_IMAGE=grafana/loki:3.7.9, VICTORIALOGS_IMAGE=victoriametrics/victoria-logs:v1.52.0, GRAFANA_IMAGE=grafana/grafana:13.2.3, PROXY_IMAGE=loki-vl-proxy:compat-vl

Showing 17/17 scopes; source links are above.

Real workload A/B: base, PR and Loki

Selected shape × range × Loki profile rows; not the complete workload baseline.

Profile / shape / range	Base cold / warm p50	PR cold / warm p50 / p95	Loki cold / warm p50	Result vs Loki
loki-defaults / logs plain / 1h	61 ms / 53 ms	49 ms / 51 ms / 79 ms	449 ms / 435 ms	same; same
loki-defaults / logs | json / 1h	115 ms / 76 ms	115 ms / 75 ms / 91 ms	731 ms / 668 ms	same; same
loki-defaults / json volume no filters / 1h	164 ms / 57 ms	165 ms / 54 ms / 96 ms	1243 ms / 341 ms	same; same
loki-defaults / stream-label count by app / 1h	17 ms / 7 ms	16 ms / 7 ms / 8 ms	394 ms / 99 ms	same; same
loki-defaults / topk rate by app / 1h	15 ms / 9 ms	17 ms / 9 ms / 9 ms	415 ms / 134 ms	differs; same
loki-defaults / drilldown service landing / 1h	21 ms / 7 ms	19 ms / 6 ms / 7 ms	390 ms / 86 ms	same; same
loki-defaults / json rate by level (no filter, no drop) / 1h	HTTP 400	HTTP 400	HTTP 400	same; same
loki-defaults / instant ungrouped rate / instant	11 ms / 9 ms	12 ms / 11 ms / 11 ms	74 ms / 70 ms	same; same
loki-defaults / labels ending now / 1h	20 ms / 6 ms	18 ms / 6 ms / 7 ms	3 ms / 2 ms	same; same; stale answer: base, pr, loki; not a latency reference
loki-defaults / label values env ending now / 1h	57 ms / 9 ms	56 ms / 8 ms / 12 ms	21 ms / 20 ms	same; same; stale answer: base, pr, loki; not a latency reference
loki-defaults / labels ending two days ago (control) / 1h	3 ms / 2 ms	3 ms / 1 ms / 1 ms	2 ms / 2 ms	differs; same; stale answer: base, pr, loki; not a latency reference
loki-defaults / label values env ending two days ago (control) / 1h	4 ms / 2 ms	4 ms / 1 ms / 2 ms	2 ms / 2 ms	differs; same; stale answer: base, pr, loki; not a latency reference
loki-defaults / E1 json duration_ms sum by service / 1h	63 ms / 6 ms	49 ms / 6 ms / 9 ms	130 ms / 14 ms	same; same
loki-defaults / E2 json duration_ms p95 by service / 1h	108 ms / 9 ms	124 ms / 9 ms / 9 ms	144 ms / 12 ms	same; same
loki-defaults / E3 logfmt latency_ms rate, production / 1h	302 ms / 10 ms	289 ms / 9 ms / 10 ms	662 ms / 88 ms	differs; same
loki-defaults / E4 logfmt size_bytes max by service / 1h	37 ms / 6 ms	27 ms / 5 ms / 6 ms	117 ms / 11 ms	same; same
loki-defaults / E6 logfmt ttl unit string (Loki 400) / 1h	HTTP 400	HTTP 400	HTTP 400	same; same
loki-defaults / E7 json hexadecimal trace_id (Loki 400) / 1h	HTTP 400	HTTP 400	HTTP 400	same; same
loki-defaults / E8 json duration_ms sum by service, namespace prod / 1h	167 ms / 8 ms	183 ms / 7 ms / 7 ms	418 ms / 67 ms	same; same
loki-defaults / E9 logfmt latency_ms sum by service, namespace prod / 1h	186 ms / 8 ms	189 ms / 8 ms / 12 ms	473 ms / 64 ms	same; same

Showing 20/56 rows; full details remain in the linked A/B artifact/comment. † Loki's cache answered cold; not a computed cold comparison. Not-like-for-like rows are not speedup proof.

Profile / shape / range	VL sampled max CPU, base	VL sampled max CPU, PR	VL sampled max CPU, Loki pass
loki-defaults / logs plain / 1h	63.7%	63.7%	63.7%
loki-defaults / logs | json / 1h	63.7%	63.7%	72.1%
loki-defaults / json volume no filters / 1h	72.1%	72.1%	72.1%
loki-defaults / stream-label count by app / 1h	41.28%	41.28%	41.28%
loki-defaults / topk rate by app / 1h	41.28%	41.28%	41.28%
loki-defaults / drilldown service landing / 1h	17.55%	17.55%	17.55%
loki-defaults / json rate by level (no filter, no drop) / 1h	20.36%	20.36%	20.36%
loki-defaults / instant ungrouped rate / instant	20.36%	20.36%	20.36%
loki-defaults / labels ending now / 1h	39.82%	39.82%	39.82%
loki-defaults / label values env ending now / 1h	39.82%	39.82%	39.82%
loki-defaults / labels ending two days ago (control) / 1h	2.42%	2.42%	2.42%
loki-defaults / label values env ending two days ago (control) / 1h	2.31%	2.31%	2.31%

Docker CPU percentages sampled within ±1.5s of each request, maximum across that row's runs. Overlapping work and background activity can contribute; these are not query-attributable CPU seconds. This summary's zero may also mean missing samples.

Distributed cache: candidate only

Current PR candidate only. RF1/RF2 are different configurations, not main/PR comparison. First/repeat populations follow each fleet phase; first is not always cold VL.

RF / phase / layer / replica	First p50 / p95 ms	Repeat p50 / p95 ms	VL calls	Peer hits / errors
1 / cross_peer_read / stats_extent / proxy-a	14.452 / 14.452	0.444 / 0.532	1	0 / 0
1 / cross_peer_read / stats_extent / proxy-b	4.212 / 4.212	0.466 / 0.478	0	3 / 0
1 / cross_peer_read / stats_extent / proxy-c	3.973 / 3.973	0.448 / 0.516	0	4 / 0
1 / cross_peer_read / query_steps / proxy-a	15.052 / 24.079	0.549 / 0.825	19	0 / 0
1 / cross_peer_read / query_steps / proxy-b	2.997 / 3.295	0.555 / 0.93	2	18 / 0
1 / cross_peer_read / query_steps / proxy-c	1.939 / 2.733	0.535 / 0.719	2	17 / 0
1 / cross_peer_read / patterns / proxy-a	17.016 / 24.857	0.491 / 0.617	24	0 / 0
1 / cross_peer_read / patterns / proxy-b	4.114 / 4.915	0.624 / 0.895	0	1 / 0
1 / cross_peer_read / patterns / proxy-c	4.636 / 5.315	0.524 / 0.574	0	2 / 0
1 / cross_peer_read / query_range_window / proxy-a	15.6 / 26.655	0.745 / 0.957	32	0 / 0
1 / cross_peer_read / query_range_window / proxy-b	15.874 / 27.185	0.766 / 0.838	19	13 / 0
1 / cross_peer_read / query_range_window / proxy-c	16.307 / 19.342	0.639 / 0.802	11	21 / 0
1 / cross_peer_read / metadata / proxy-a	7.421 / 45.091	0.298 / 0.391	104	0 / 0
1 / cross_peer_read / metadata / proxy-b	0.474 / 1.273	0.313 / 0.394	0	4 / 0
1 / cross_peer_read / metadata / proxy-c	0.707 / 0.8	0.306 / 0.389	0	10 / 0
1 / memory_lost_disk_kept / stats_extent / proxy-b	3.989 / 3.989	0.461 / 0.499	0	3 / 0

Showing 16/216 layer rows; full report in CI artifact.

RF / phase / replica	Phase seconds / samples	CPU seconds	Peak RSS MiB	End Loki-layer logical RAM / SSD bytes
1 / cross_peer_read / proxy-a	20.3982 / 41	1.75322	198.543	184065 / 69632
1 / cross_peer_read / proxy-b	20.3982 / 41	1.52392	189.82	133658 / 69632
1 / cross_peer_read / proxy-c	20.3982 / 41	1.54667	186.051	135898 / 77824
1 / memory_lost_disk_kept / proxy-a	7.36311 / 15	0.490045	198.891	169979 / 69632
1 / memory_lost_disk_kept / proxy-b	7.36311 / 15	0.633418	189.836	133047 / 69632
1 / memory_lost_disk_kept / proxy-c	7.36311 / 15	0.456926	186.379	136244 / 77824
1 / disk_lost_refill_from_peers / proxy-a	15.8176 / 32	0.791922	198.949	169978 / 69632
1 / disk_lost_refill_from_peers / proxy-b	15.8176 / 32	0.966384	115.441	157575 / 65536
1 / disk_lost_refill_from_peers / proxy-c	15.8176 / 32	0.730624	188.926	136244 / 77824
1 / owner_down_served_by_others / proxy-a	21.3248 / 14	0.65531	202.402	256077 / 81920
1 / owner_down_served_by_others / proxy-b	21.3248 / 43	1.60384	197.527	288612 / 102400
1 / owner_down_served_by_others / proxy-c	21.3248 / 43	1.48302	200.586	189733 / 94208

CPU seconds are sampled process-counter deltas within a phase, including background activity. RSS peaks belong to each sampled replica, not a simultaneous fleet-wide peak. Occupancy uses Loki-facing l1_memory/l2_disk counters; VL-facing cache occupancy is unmeasured here. Logical cache bytes are not physical disk I/O or process memory. Resources have no main-versus-PR delta.

RF / phase / layer / replica	Requests	Disk / window hits	Peer misses / backup hits	Fleet VL / background VL calls	Observed VL routes
1 / all_lost_rebuild_equals_loki / metadata / proxy-a	12	0 / 0	118 / 0	104 / 0	/select/logsql/field_names=22, /select/logsql/field_values=6, /select/logsql/query=20, /select/logsql/stream_field_names=20
1 / all_lost_rebuild_equals_loki / patterns / proxy-a	2	0 / 0	2 / 0	24 / 0	/select/logsql/query=24
1 / all_lost_rebuild_equals_loki / patterns / proxy-b	2	0 / 0	0 / 0	0 / 0	—
1 / all_lost_rebuild_equals_loki / patterns / proxy-c	2	0 / 0	0 / 0	0 / 0	—
1 / all_lost_rebuild_equals_loki / query_range_window / proxy-a	2	0 / 0	18 / 0	32 / 0	/select/logsql/hits=16, /select/logsql/query=16
1 / all_lost_rebuild_equals_loki / query_steps / proxy-a	2	0 / 0	39 / 0	19 / 0	/select/logsql/query=1, /select/logsql/stats_query_range=2, /select/logsql/stream_field_names=16
1 / all_lost_rebuild_equals_loki / query_steps / proxy-b	2	0 / 0	1 / 0	2 / 0	/select/logsql/query=1, /select/logsql/stream_field_names=1
1 / all_lost_rebuild_equals_loki / query_steps / proxy-c	2	0 / 0	2 / 0	2 / 0	/select/logsql/query=1, /select/logsql/stream_field_names=1
1 / all_lost_rebuild_equals_loki / stats_extent / proxy-a	1	0 / 0	3 / 0	1 / 0	/select/logsql/stats_query_range=1
1 / all_lost_rebuild_equals_loki / stats_extent / proxy-b	1	0 / 0	0 / 0	0 / 0	—
1 / all_lost_rebuild_equals_loki / stats_extent / proxy-c	1	0 / 0	0 / 0	0 / 0	—
1 / disk_lost_refill_from_peers / metadata / proxy-b	36	0 / 0	6 / 0	31 / 0	/select/logsql/field_names=8, /select/logsql/field_values=6, /select/logsql/query=8, /select/logsql/stream_field_names=2
1 / disk_lost_refill_from_peers / patterns / proxy-b	6	0 / 0	0 / 0	0 / 0	—
1 / disk_lost_refill_from_peers / query_range_window / proxy-b	6	0 / 9	2 / 0	11 / 0	/select/logsql/hits=4, /select/logsql/query=7
1 / disk_lost_refill_from_peers / query_steps / proxy-a	2	0 / 0	0 / 0	0 / 0	—
1 / disk_lost_refill_from_peers / query_steps / proxy-b	8	0 / 0	2 / 0	12 / 0	/select/logsql/query=5, /select/logsql/stats_query_range=2, /select/logsql/stream_field_names=5

Request-counter sums across each shown phase/layer/replica, including first and repeat requests. Showing 16/216 groups, disk-loss/restart and owner-loss phases first. Fleet VL calls exclude background calls; background is listed separately. These counters are not physical disk reads or throughput. Full routes and inventory-unit attribution remain in the artifact.

RF / phase / backend container	Phase seconds / samples	CPU avg / max	Peak container memory MiB
1 / cross_peer_read / fleet-loki	20.3982 / 5	0.67% / 1.2%	50.68
1 / cross_peer_read / fleet-vl	20.3982 / 5	3.228% / 6.81%	15.2
1 / memory_lost_disk_kept / fleet-loki	7.36311 / 2	0.525% / 0.54%	51.95
1 / memory_lost_disk_kept / fleet-vl	7.36311 / 2	3.475% / 3.62%	11.77
1 / disk_lost_refill_from_peers / fleet-loki	15.8176 / 4	0.5675% / 0.62%	54.76
1 / disk_lost_refill_from_peers / fleet-vl	15.8176 / 4	1.4625% / 3.86%	12.64
1 / owner_down_served_by_others / fleet-loki	21.3248 / 5	0.848% / 2.1%	57.81
1 / owner_down_served_by_others / fleet-vl	21.3248 / 5	3.17% / 6.47%	14.47
1 / all_lost_rebuild_equals_loki / fleet-loki	16.9519 / 4	0.5475% / 0.58%	58.14
1 / all_lost_rebuild_equals_loki / fleet-vl	16.9519 / 4	2.705% / 6.83%	16.86
1 / near_now_label_inventory / fleet-loki	33.3444 / 8	0.5875% / 0.74%	88.37
1 / near_now_label_inventory / fleet-vl	33.3444 / 8	2.79% / 7.39%	12.15
1 / edge_disk_full / fleet-loki	30.191 / 8	0.5675% / 0.68%	93.01
1 / edge_disk_full / fleet-vl	30.191 / 8	1.75375% / 5.61%	17.13
1 / edge_owner_paused_timeout_bound / fleet-loki	27.9954 / 7	0.572857% / 0.68%	97.54
1 / edge_owner_paused_timeout_bound / fleet-vl	27.9954 / 7	1.75714% / 3.92%	17.39

Existing docker stats samples from the candidate fleet's Loki/VL containers; CPU is Docker percentage, not CPU seconds or a main/PR delta. Memory is sampled container usage, not proxy RSS.

Synthetic handler microbenchmarks

Synthetic handler elapsed time and allocations; not process CPU, RSS or physical disk I/O. Repeated-sample medians.

Signal	Base	PR	Delta
QueryRange cache-hit elapsed	3564 ns/op	3531 ns/op	-0.9% (stable)
QueryRange cache-hit allocated bytes	1256 B/op	1256 B/op	0.0% (stable)
QueryRange cache-hit allocations	28 allocs/op	28 allocs/op	0.0% (stable)
QueryRange cache-bypass elapsed	4243 ns/op	4222 ns/op	-0.5% (stable)
QueryRange cache-bypass allocated bytes	1325 B/op	1321 B/op	-0.3% (stable)
QueryRange cache-bypass allocations	28 allocs/op	28 allocs/op	0.0% (stable)
Labels cache-hit elapsed	949.2 ns/op	940.1 ns/op	-1.0% (stable)
Labels cache-hit allocated bytes	48 B/op	48 B/op	0.0% (stable)
Labels cache-hit allocations	3 allocs/op	3 allocs/op	0.0% (stable)
Labels cache-bypass elapsed	1571 ns/op	1544 ns/op	-1.7% (stable)
Labels cache-bypass allocated bytes	400 B/op	400 B/op	0.0% (stable)
Labels cache-bypass allocations	14 allocs/op	14 allocs/op	0.0% (stable)

Labels follow the existing gate percentage, absolute and low-baseline guards, including pooled-path allowances.

CI regression checks

Source / check	State	Result
ci / bench	completed	success
ci / conformance	completed	success
ci / docker	completed	success
ci / e2e-compat	completed	success
ci / e2e-compat (core)	completed	success
ci / e2e-compat (drilldown)	completed	success
ci / e2e-compat (otel-edge)	completed	success
ci / e2e-compat (semantics)	completed	success
ci / e2e-compat (tail-multitenancy)	completed	success
ci / e2e-fleet	completed	success
ci / e2e-ui (comprehensive-drilldown-mt)	completed	success
ci / e2e-ui (drilldown-core-tail)	completed	success
ci / e2e-ui (explore-core)	completed	success
ci / e2e-ui (ops-mt)	completed	success
ci / e2e-ui (regression-datasource)	completed	success
ci / fuzz-smoke	completed	success
ci / helm	completed	success
ci / image	completed	success
ci / lint	completed	success
ci / memory-leak-tests	completed	success
ci / race-stress	completed	success
ci / race-stress-shard (1)	completed	success
ci / race-stress-shard (2)	completed	success
ci / race-stress-shard (3)	completed	success
ci / race-stress-shard (other)	completed	success
ci / scope / scope	completed	success
ci / stress-tests	completed	success
ci / test	completed	success
ci / tuple-contract	completed	success
ci / tuple-smoke	completed	success
drilldown / drilldown-contract-matrix	completed	skipped
drilldown / drilldown-grafana-pr-matrix (12.4.12, previous_smoke, true, ^(TestGrafanaDatasourceCatalogAndHeal...	completed	success
drilldown / drilldown-grafana-pr-matrix (13.2.3, current_smoke, true, ^(TestGrafanaDatasourceCatalogAndHealth...	completed	success
drilldown / drilldown-grafana-runtime-matrix	completed	skipped
drilldown / drilldown-matrix-source	completed	success
drilldown / drilldown-pinned-runtime	completed	success
drilldown / scope / scope	completed	success
loki / loki-matrix	completed	skipped
loki / loki-matrix-source	completed	success
loki / loki-pinned	completed	success
loki / scope / scope	completed	success
perf / perf-smoke	completed	success
perf / scope / scope	completed	success
quality / report	completed	success
quality / scope / scope	completed	success
security / Security / runtime	completed	success
security / Security / static	completed	success
security / scope / scope	completed	success
vl / scope / scope	completed	success
vl / vl-matrix	completed	skipped
vl / vl-matrix-source	completed	success
vl / vl-pinned	completed	success

Showing 52/52 checks; failed and pending checks first. Full job lists remain in the linked runs.

Required checks decide merge safety. This report summarizes selected evidence; rendering starts no tests or heavy workloads.