fix: F004 exclude empty tables from recommendations

In case when a table is empty (has 0 rows) its heap bloat is estimated as 100%. In some cases, such a table might be included in "F004 Recommendations" and considered as of top priority (P1). This leads to messages containing the wrong statement. This behavior was changed: from now on, empty tables are excluded from consideration.
CI: a new test was added to check this case automatically.

Author: dmius

.ci/test_db_dump.sql

@@ -58,6 +58,10 @@ delete from bloated where i % 2 = 0;
 
 -- F004
 create table t_with_bloat as select i from generate_series(1, 1000000) _(i);
+-- create table without data
+create table t_f004_empty as select i from generate_series(1, 1000000) _(i);
+alter table t_f004_empty set (autovacuum_enabled = 'off');
+analyze t_f004_empty;
 
 -- h002 Supports fk
 create table t_red_fk_1 as select id::int8 from generate_series(0, 1000000) _(id);
@@ -117,3 +121,7 @@ CREATE TABLE test_schema."orders_A"
 
 INSERT INTO test_schema."orders_A"(cnt) select id from generate_series(0, 100) _(id);
 SELECT setval('test_schema."orders_A_id_seq"'::regclass, 300000000, false);
+
+-- t_f004 continue
+delete from t_f004_empty;
+analyze t_f004_empty;

.gitlab-ci.yml

@@ -55,6 +55,8 @@ variables:
     - data_dir=$(cat ./artifacts/test/nodes.json | jq -r '.last_check | .dir') && l001_indexes_size=$(cat ./artifacts/test/json_reports/$data_dir/L001_table_sizes.json | jq '.results .postgres .data .tables_data_total .indexes_size_bytes_sum') && echo $l001_indexes_size
     - diff=$((f005_indexes_size - l001_indexes_size)) && diff=${diff#-} && echo $diff && ratio=$((diff * 100 / l001_indexes_size)) && echo $ratio && ([[ $ratio -gt "5" ]]) && exit 303
     - echo "F005/L001 Total indexes size passed"
+    - ./checkup -h postgres --username test_user --project test --dbname dbname -e 1 --file ./resources/checks/F004_heap_bloat.sh
+    - data_dir=$(cat ./artifacts/test/nodes.json | jq -r '.last_check | .dir') && msg=$(grep "can be reduced 0.00 times" ./artifacts/test/md_reports/$data_dir/F004.md) && [[ ! -z "$msg" ]] && exit 304
     - ([[ "$CI_COMMIT_REF_NAME" != "master" ]]) && exit 0
     # Check small indexes
     - .ci/prepare_test_db.sh postgres

pghrep/src/checkup/f004/f004.go

@@ -20,7 +20,7 @@ const F004_GENERAL_INFO string = "F004_GENERAL_INFO"
 const WARNING_BLOAT_RATIO float32 = 40.0
 const CRITICAL_BLOAT_RATIO float32 = 90.0
 const CRITICAL_TOTAL_BLOAT_RATIO float32 = 20.0
-const MIN_INDEX_SIZE_TO_ANALYZE int64 = 1024 * 1024
+const MIN_TABLE_SIZE_TO_ANALYZE int64 = 1024 * 1024
 
 func appendTable(list []string, tableBloatData F004HeapBloat) []string {
 	return append(list, fmt.Sprintf(TABLE_DETAILS, tableBloatData.TableName,
@@ -48,11 +48,12 @@ func F004Process(report F004Report) checkup.ReportResult {
 		}
 		for _, table := range sortedTables {
 			heapBloatData := hostData.Data.HeapBloat[table]
-			if (heapBloatData.RealSizeBytes > MIN_INDEX_SIZE_TO_ANALYZE) && (heapBloatData.BloatRatioPercent >= WARNING_BLOAT_RATIO) &&
-				(heapBloatData.BloatRatioPercent < CRITICAL_BLOAT_RATIO) {
+			if (heapBloatData.RealSizeBytes > MIN_TABLE_SIZE_TO_ANALYZE) && (heapBloatData.BloatRatioPercent >= WARNING_BLOAT_RATIO) &&
+				(heapBloatData.BloatRatioPercent < CRITICAL_BLOAT_RATIO) && (heapBloatData.BloatRatioFactor > 0) {
 				warningTables = appendTable(warningTables, heapBloatData)
 			}
-			if heapBloatData.RealSizeBytes > MIN_INDEX_SIZE_TO_ANALYZE && heapBloatData.BloatRatioPercent >= CRITICAL_BLOAT_RATIO {
+			if (heapBloatData.RealSizeBytes > MIN_TABLE_SIZE_TO_ANALYZE) && (heapBloatData.BloatRatioPercent >= CRITICAL_BLOAT_RATIO) &&
+				(heapBloatData.BloatRatioFactor > 0) {
 				criticalTables = appendTable(criticalTables, heapBloatData)
 			}
 		}

pghrep/src/checkup/f005/f005.go

@@ -22,7 +22,7 @@ const F005_BLOAT_EXCESS_INFO string = "F005_BLOAT_EXCESS_INFO"
 const WARNING_BLOAT_RATIO float32 = 40.0
 const CRITICAL_BLOAT_RATIO float32 = 90.0
 const CRITICAL_TOTAL_BLOAT_RATIO float32 = 20.0
-const MIN_TABLE_SIZE_TO_ANALYZE int64 = 1024 * 1024
+const MIN_INDEX_SIZE_TO_ANALYZE int64 = 1024 * 1024
 
 func appendIndex(list []string, indexBloatData F005IndexBloat) []string {
 	return append(list, fmt.Sprintf(INDEX_DETAILS, indexBloatData.IndexName,
@@ -52,15 +52,17 @@ func F005Process(report F005Report) checkup.ReportResult {
 		}
 		for _, index := range sortedIndexes {
 			indexBloatData := hostData.Data.IndexBloat[index]
-			if totalBloatIsCritical && indexBloatData.RealSizeBytes > MIN_TABLE_SIZE_TO_ANALYZE && i < 5 {
+			if totalBloatIsCritical && indexBloatData.RealSizeBytes > MIN_INDEX_SIZE_TO_ANALYZE && i < 5 &&
+        (indexBloatData.BloatRatioFactor > 0) {
 				top5Indexes = appendIndex(top5Indexes, indexBloatData)
 				i++
 			}
-			if indexBloatData.RealSizeBytes > MIN_TABLE_SIZE_TO_ANALYZE && indexBloatData.BloatRatioPercent >= CRITICAL_BLOAT_RATIO {
+			if indexBloatData.RealSizeBytes > MIN_INDEX_SIZE_TO_ANALYZE && indexBloatData.BloatRatioPercent >= CRITICAL_BLOAT_RATIO &&
+        (indexBloatData.BloatRatioFactor > 0) {
 				criticalIndexes = appendIndex(criticalIndexes, indexBloatData)
 			}
-			if (indexBloatData.RealSizeBytes > MIN_TABLE_SIZE_TO_ANALYZE) && (indexBloatData.BloatRatioPercent >= WARNING_BLOAT_RATIO) &&
-				(indexBloatData.BloatRatioPercent < CRITICAL_BLOAT_RATIO) {
+			if (indexBloatData.RealSizeBytes > MIN_INDEX_SIZE_TO_ANALYZE) && (indexBloatData.BloatRatioPercent >= WARNING_BLOAT_RATIO) &&
+				(indexBloatData.BloatRatioPercent < CRITICAL_BLOAT_RATIO) && (indexBloatData.BloatRatioFactor > 0) {
 				warningIndexes = appendIndex(warningIndexes, indexBloatData)
 			}
 		}