Probabilistic tracking for new tenants.

src/backend/distributed/shared_library_init.c

@@ -2402,7 +2402,6 @@ RegisterCitusConfigVariables(void)
 		PGC_POSTMASTER,
 		GUC_STANDARD,
 		NULL, NULL, NULL);
-
 	DefineCustomEnumVariable(
 		"citus.stat_tenants_log_level",
 		gettext_noop("Sets the level of citus_stat_tenants log messages"),
@@ -2424,6 +2423,17 @@ RegisterCitusConfigVariables(void)
 		GUC_STANDARD,
 		NULL, NULL, NULL);
 
+
+	DefineCustomIntVariable(
+		"citus.stat_tenants_sample_rate_for_new_tenants",
+		gettext_noop("Sampling rate for new tenants in citus_stat_tenants."),
+		NULL,
+		&StatTenantsSampleRateForNewTenants,
+		100, 1, 100,
+		PGC_USERSET,
+		GUC_STANDARD,
+		NULL, NULL, NULL);
+
 	DefineCustomEnumVariable(
 		"citus.stat_tenants_track",
 		gettext_noop("Enables/Disables the stats collection for citus_stat_tenants."),

src/backend/distributed/utils/citus_stat_tenants.c

@@ -77,7 +77,7 @@ int StatTenantsLogLevel = CITUS_LOG_LEVEL_OFF;
 int StatTenantsPeriod = (time_t) 60;
 int StatTenantsLimit = 100;
 int StatTenantsTrack = STAT_TENANTS_TRACK_NONE;
-
+int StatTenantsSampleRateForNewTenants = 100;
 
 PG_FUNCTION_INFO_V1(citus_stat_tenants_local);
 PG_FUNCTION_INFO_V1(citus_stat_tenants_local_reset);
@@ -177,8 +177,6 @@ citus_stat_tenants_local(PG_FUNCTION_ARGS)
 }
 
 
-#include "miscadmin.h"
-
 /*
  * citus_stat_tenants_local_reset resets monitor for tenant statistics
  * on the local node.
@@ -186,9 +184,6 @@ citus_stat_tenants_local(PG_FUNCTION_ARGS)
 Datum
 citus_stat_tenants_local_reset(PG_FUNCTION_ARGS)
 {
-	/* ereport(NOTICE, (errmsg("MyProcPid: %d", MyProcPid))); */
-	/*sleep(10); */
-
 	MultiTenantMonitor *monitor = GetMultiTenantMonitor();
 
 	/* if monitor is not created yet, there is nothing to reset */
@@ -268,6 +263,21 @@ AttributeTask(char *tenantId, int colocationId, CmdType commandType)
 		return;
 	}
 
+	TenantStatsHashKey key = { 0 };
+	FillTenantStatsHashKey(&key, tenantId, colocationId);
+
+	MultiTenantMonitor *monitor = GetMultiTenantMonitor();
+	bool found = false;
+	hash_search(monitor->tenants, &key, HASH_FIND, NULL);
+
+	int randomValue = rand() % 100;
+
+	/* If the tenant is not found in the hash table, we will track the query with a probability of StatTenantsSampleRateForNewTenants.*/
+	if (!(!found && randomValue < StatTenantsSampleRateForNewTenants))
+	{
+		return;
+	}
+
 	AttributeToColocationGroupId = colocationId;
 	strncpy_s(AttributeToTenant, MAX_TENANT_ATTRIBUTE_LENGTH, tenantId,
 			  MAX_TENANT_ATTRIBUTE_LENGTH - 1);

src/include/distributed/utils/citus_stat_tenants.h

@@ -121,5 +121,6 @@ extern int StatTenantsLogLevel;
 extern int StatTenantsPeriod;
 extern int StatTenantsLimit;
 extern int StatTenantsTrack;
+extern int StatTenantsSampleRateForNewTenants;
 
 #endif /*CITUS_ATTRIBUTE_H */