This is an automated email from the ASF dual-hosted git repository.

hanahmily pushed a commit to branch vectorized-query
in repository https://gitbox.apache.org/repos/asf/skywalking-banyandb.git

commit a0dc29e9595fb5966edabe1a204c59ce9884ea57
Author: Hongtao Gao <[email protected]>
AuthorDate: Wed May 13 03:28:23 2026 +0000

    feat(measure): flip vectorized query path to enabled by default (G5e)
    
    Enable VectorizedConfig.Enabled by default following the G5d soak (see
    docs/soak/g5d-phase-d-summary.md and commit 48c0f76b — 48h vec-on run,
    final_parity_pass=true, 576/576 replay-and-diff cycles pass, zero
    MemoryTracker exhaustion). The criterion-4 goroutine-drift miss is
    root-caused upstream to apache/skywalking#13874 (bluge index writer
    pool lifecycle at segment rotation), independent of the vec path.
    
    Wire format (*measurev1.InternalDataPoint) byte-identical. Rollback via
    --measure-vectorized-enabled=false on the standalone or data-node
    command line; the row path resumes on restart.
    
    Verification:
    - go build ./...                                              PASS
    - go test ./banyand/measure/... -short -race                  PASS
    - go test ./pkg/query/vectorized/... -race                    PASS
    - RUN_BENCH_GATES=1 TestBenchGates_PerWorkload (W1..W5 + MB)  PASS (27.8s)
    - vectorized parity integration suite (488 specs)             PASS (90.1s)
    - make -s lint                                                clean
---
 CHANGES.md                             | 1 +
 pkg/query/vectorized/measure/config.go | 9 +++++----
 2 files changed, 6 insertions(+), 4 deletions(-)

diff --git a/CHANGES.md b/CHANGES.md
index ebbdb96b0..b28a6ed4a 100644
--- a/CHANGES.md
+++ b/CHANGES.md
@@ -6,6 +6,7 @@ Release Notes.
 
 ### Features
 
+- Vectorized measure query path is now enabled by default. The columnar 
pipeline replaces per-row protobuf serialization in `NewMIterator`, cutting 
allocations and ns/op for scan-heavy measure queries; gRPC wire format 
(`*measurev1.InternalDataPoint`) is byte-identical. Rollback: pass 
`--measure-vectorized-enabled=false` on the standalone or data-node command 
line and restart; the row path resumes immediately.
 - Organize access logs under a dedicated "accesslog" subdirectory to improve 
log organization and separation from other application data.
 - Collect BanyanDB data on e2e test failure for CI debugging.
 - Add log query e2e test.
diff --git a/pkg/query/vectorized/measure/config.go 
b/pkg/query/vectorized/measure/config.go
index 83d2eed7a..b505c1a83 100644
--- a/pkg/query/vectorized/measure/config.go
+++ b/pkg/query/vectorized/measure/config.go
@@ -26,11 +26,12 @@ type VectorizedConfig struct {
        Enabled        bool
 }
 
-// DefaultConfig returns the v1 default — disabled, 1024-row batches, 256 MiB
-// per-query memory budget. v1 ships with Enabled false; the rollout PR flips
-// the default after soak and bench gates pass.
+// DefaultConfig returns the v1 default — enabled, 1024-row batches, 256 MiB
+// per-query memory budget. v1 ships with Enabled true post-soak/bench-gate
+// rollout (G5e). To roll back, pass --measure-vectorized-enabled=false on
+// the standalone or data-node command line and restart.
 func DefaultConfig() VectorizedConfig {
-       return VectorizedConfig{Enabled: false, BatchSize: 1024, 
QueryMemoryMiB: 256}
+       return VectorizedConfig{Enabled: true, BatchSize: 1024, QueryMemoryMiB: 
256}
 }
 
 // Validate rejects nonsense configurations.

Reply via email to