Merge pull request #64842 from Unalian/jemalloc_as_metric

[feature] Add an asynchronous metric jemalloc.profile.active to show whether sampling is currently active

docs/en/operations/system-tables/asynchronous_metrics.md

@@ -639,6 +639,10 @@ An internal metric of the low-level memory allocator (jemalloc). See https://jem
 
 An internal metric of the low-level memory allocator (jemalloc). See https://jemalloc.net/jemalloc.3.html
 
+### jemalloc.prof.active
+
+An internal metric of the low-level memory allocator (jemalloc). See https://jemalloc.net/jemalloc.3.html
+
 **See Also**
 
 - [Monitoring](../../operations/monitoring.md) — Base concepts of ClickHouse monitoring.

src/Common/AsynchronousMetrics.cpp

@@ -415,6 +415,15 @@ Value saveAllArenasMetric(AsynchronousMetricValues & values,
         fmt::format("jemalloc.arenas.all.{}", metric_name));
 }
 
+template<typename Value>
+Value saveJemallocProf(AsynchronousMetricValues & values,
+    const std::string & metric_name)
+{
+    return saveJemallocMetricImpl<Value>(values,
+        fmt::format("prof.{}", metric_name),
+        fmt::format("jemalloc.prof.{}", metric_name));
+}
+
 }
 #endif
 
@@ -607,6 +616,7 @@ void AsynchronousMetrics::update(TimePoint update_time, bool force_update)
     saveJemallocMetric<size_t>(new_values, "background_thread.num_threads");
     saveJemallocMetric<uint64_t>(new_values, "background_thread.num_runs");
     saveJemallocMetric<uint64_t>(new_values, "background_thread.run_intervals");
+    saveJemallocProf<size_t>(new_values, "active");
     saveAllArenasMetric<size_t>(new_values, "pactive");
     [[maybe_unused]] size_t je_malloc_pdirty = saveAllArenasMetric<size_t>(new_values, "pdirty");
     [[maybe_unused]] size_t je_malloc_pmuzzy = saveAllArenasMetric<size_t>(new_values, "pmuzzy");

tests/integration/test_asynchronous_metric_jemalloc_profile_active/configs/asynchronous_metrics_update_period_s.xml

@@ -0,0 +1,3 @@
+<clickhouse>
+    <asynchronous_metrics_update_period_s>1</asynchronous_metrics_update_period_s>
+</clickhouse>

tests/integration/test_asynchronous_metric_jemalloc_profile_active/test.py

@@ -0,0 +1,73 @@
+import time
+
+import pytest
+from helpers.cluster import ClickHouseCluster
+
+cluster = ClickHouseCluster(__file__)
+node1 = cluster.add_instance(
+    "node1",
+    main_configs=["configs/asynchronous_metrics_update_period_s.xml"],
+    env_variables={"MALLOC_CONF": "background_thread:true,prof:true"},
+)
+
+
+@pytest.fixture(scope="module")
+def started_cluster():
+    try:
+        cluster.start()
+        yield cluster
+
+    finally:
+        cluster.shutdown()
+
+
+# asynchronous metrics are updated once every 60s by default. To make the test run faster, the setting
+# asynchronous_metric_update_period_s is being set to 1s so that the metrics are populated faster and
+# are available for querying during the test.
+def test_asynchronous_metric_jemalloc_profile_active(started_cluster):
+    # default open
+    if node1.is_built_with_sanitizer():
+        pytest.skip("Disabled for sanitizers")
+
+    res_o = node1.query(
+        "SELECT * FROM system.asynchronous_metrics WHERE metric ILIKE '%jemalloc.prof.active%' FORMAT Vertical;"
+    )
+    assert (
+        res_o
+        == """Row 1:
+──────
+metric:      jemalloc.prof.active
+value:       1
+description: An internal metric of the low-level memory allocator (jemalloc). See https://jemalloc.net/jemalloc.3.html
+"""
+    )
+    # disable
+    node1.query("SYSTEM JEMALLOC DISABLE PROFILE")
+    time.sleep(5)
+    res_t = node1.query(
+        "SELECT * FROM system.asynchronous_metrics WHERE metric ILIKE '%jemalloc.prof.active%' FORMAT Vertical;"
+    )
+    assert (
+        res_t
+        == """Row 1:
+──────
+metric:      jemalloc.prof.active
+value:       0
+description: An internal metric of the low-level memory allocator (jemalloc). See https://jemalloc.net/jemalloc.3.html
+"""
+    )
+    # enable
+    node1.query("SYSTEM JEMALLOC ENABLE PROFILE")
+    time.sleep(5)
+    res_f = node1.query(
+        "SELECT * FROM system.asynchronous_metrics WHERE metric ILIKE '%jemalloc.prof.active%' FORMAT Vertical;"
+    )
+    assert (
+        res_f
+        == """Row 1:
+──────
+metric:      jemalloc.prof.active
+value:       1
+description: An internal metric of the low-level memory allocator (jemalloc). See https://jemalloc.net/jemalloc.3.html
+"""
+    )