Reduce histogram metric cardinality for baseline resync

Convert snapshot histogram metrics from full bucket distributions to
sum/count publishing mode

Changes:
- Convert 4 snapshot histograms to publish_as_sum_count:
  - snp_dnr_blob_process_latency (donor)
  - snp_dnr_batch_process_latency (donor)
  - snp_rcvr_blob_process_time (receiver)
  - snp_rcvr_batch_process_time (receiver)

- snapshot batch e2e remains as histogram
  - snp_dnr_batch_e2e_latency (donor)

- Update 4 GC histogram comments for accuracy:
  - PercentileBuckets: 128 → 10 buckets (10% increments)
  - LinearUpto64Buckets: Clarify 17 buckets covering 0-64s

Signed-off-by: Xiaoxi Chen <xiaoxchen@ebay.com>

Author: xiaoxichen

conanfile.py

@@ -10,7 +10,7 @@
 
 class HomeObjectConan(ConanFile):
     name = "homeobject"
-    version = "4.1.0"
+    version = "4.1.1"
 
     homepage = "https://github.com/eBay/HomeObject"
     description = "Blob Store built on HomeStore"
@@ -49,7 +49,7 @@ def build_requirements(self):
         self.test_requires("gtest/1.17.0")
 
     def requirements(self):
-        self.requires("sisl/[^13.0]@oss/master", transitive_headers=True)
+        self.requires("sisl/[^13.2.3]@oss/master", transitive_headers=True)
         self.requires("homestore/[^7.4]@oss/master")
         self.requires("iomgr/[^12.0]@oss/master")
 

src/lib/homestore_backend/gc_manager.hpp

@@ -137,16 +137,16 @@ class GCManager {
 
                 // gc task level histogram metrics
                 REGISTER_HISTOGRAM(reclaim_ratio_gc, "the ratio of reclaimed blks to total blks in a gc task",
-                                   HistogramBucketsType(PercentileBuckets)); // 0% to 100% in 128 buckets
+                                   HistogramBucketsType(PercentileBuckets)); // 0% to 100% in 10 buckets (10% increments)
                 REGISTER_HISTOGRAM(
                     gc_time_duration_s_gc, "how long a successful gc task takes by second",
-                    HistogramBucketsType(LinearUpto64Buckets)); // gc task is expected to finish within 1 minutes
+                    HistogramBucketsType(LinearUpto64Buckets)); // 17 buckets covering 0-64 seconds in 4s increments
 
                 REGISTER_HISTOGRAM(reclaim_ratio_egc, "the ratio of reclaimed blks to total blks in an egc task",
-                                   HistogramBucketsType(PercentileBuckets)); // 0% to 100% in 128 buckets
+                                   HistogramBucketsType(PercentileBuckets)); // 0% to 100% in 10 buckets (10% increments)
                 REGISTER_HISTOGRAM(
                     gc_time_duration_s_egc, "how long a successful egc task takes by second",
-                    HistogramBucketsType(LinearUpto64Buckets)); // gc task is expected to finish within 1 minutes
+                    HistogramBucketsType(LinearUpto64Buckets)); // 17 buckets covering 0-64 seconds in 4s increments
 
                 register_me_to_farm();
                 attach_gather_cb(std::bind(&pdev_gc_metrics::on_gather, this));

src/lib/homestore_backend/hs_homeobject.hpp

@@ -576,10 +576,12 @@ class HSHomeObject : public HomeObjectImpl {
                 REGISTER_COUNTER(snp_dnr_error_count, "Error times when reading blobs in baseline resync");
                 REGISTER_HISTOGRAM(snp_dnr_blob_process_latency,
                                    "Time cost(us) of successfully process a blob in baseline resync",
-                                   HistogramBucketsType(LowResolutionLatecyBuckets));
+                                   HistogramBucketsType(LowResolutionLatecyBuckets),
+                                   _publish_as::publish_as_sum_count);
                 REGISTER_HISTOGRAM(snp_dnr_batch_process_latency,
                                    "Time cost(ms) of successfully process a batch in baseline resync",
-                                   HistogramBucketsType(LowResolutionLatecyBuckets));
+                                   HistogramBucketsType(LowResolutionLatecyBuckets),
+                                   _publish_as::publish_as_sum_count);
                 REGISTER_HISTOGRAM(snp_dnr_batch_e2e_latency,
                                    "Time cost(ms) of a batch end-to-end round trip in baseline resync",
                                    HistogramBucketsType(LowResolutionLatecyBuckets));
@@ -682,10 +684,12 @@ class HSHomeObject : public HomeObjectImpl {
                 REGISTER_GAUGE(snp_rcvr_error_count, "Error count in baseline resync");
                 REGISTER_HISTOGRAM(snp_rcvr_blob_process_time,
                                    "Time cost(us) of successfully process a blob in baseline resync",
-                                   HistogramBucketsType(LowResolutionLatecyBuckets));
+                                   HistogramBucketsType(LowResolutionLatecyBuckets),
+                                   _publish_as::publish_as_sum_count);
                 REGISTER_HISTOGRAM(snp_rcvr_batch_process_time,
                                    "Time cost(ms) of successfully process a batch in baseline resync",
-                                   HistogramBucketsType(LowResolutionLatecyBuckets));
+                                   HistogramBucketsType(LowResolutionLatecyBuckets),
+                                   _publish_as::publish_as_sum_count);
 
                 attach_gather_cb(std::bind(&ReceiverSnapshotMetrics::on_gather, this));
                 register_me_to_farm();