Added max and min to histogram output. Standardized database and collection in query and collection definitions

Author: daniel-percona

.gitignore

@@ -1,5 +1,6 @@
 __pycache__*
 todo.txt
+demo.tape
 # Binaries
 # Ignore all binaries in the bin folder
 bin/

README.md

@@ -225,64 +225,8 @@ When executed, genMongoLoad performs the following steps:
     * Prints a detailed summary table at the end of the run.
 
 ### Sample Output
-```text
-./bin/genMongoLoad
-Enter Password for user 'root':
-
-  genMongoLoad v1.1.0-dirty
-  --------------------------------------------------
-  Database:     airline
-  Workers:      40 active
-  Duration:     10s
-  Report Freq:  1s
-
-  WORKLOAD DEFINITION
-  --------------------------------------------------
-  Batch Size:    1000
-  Mode:          Default Workload
-  Distribution:  Select (60%)  Update (20%)
-                 Insert (10%)  Delete (10%)
-                 Agg    (0%)
-
-  [INFO] Loaded 1 collection definition(s)
-  [INFO] Loaded 2 query templates(s)
-  [INFO] Skipping sharding for 'flights': Cluster is not sharded (Replica Set)
-  [INFO] Created 4 indexes on 'flights'
-  [INFO] Skipping data seeding (configured)
-
-> Starting Workload...
-
- TIME    | TOTAL OPS | SELECT | INSERT | UPDATE | DELETE |    AGG
- ---------------------------------------------------------------
- 00:01   |     8,300 |  5,004 |    798 |  1,650 |    848 |      0
- 00:02   |     8,048 |  4,736 |    773 |  1,694 |    845 |      0
- 00:03   |     8,168 |  4,728 |    824 |  1,737 |    879 |      0
- 00:04   |     8,182 |  4,893 |    817 |  1,695 |    777 |      0
- 00:05   |     8,504 |  5,047 |    843 |  1,724 |    890 |      0
- 00:06   |     8,776 |  5,271 |    851 |  1,757 |    897 |      0
- 00:07   |     8,546 |  5,145 |    880 |  1,699 |    822 |      0
- 00:08   |     8,365 |  4,945 |    828 |  1,753 |    839 |      0
- 00:09   |     8,733 |  5,208 |    907 |  1,716 |    902 |      0
- 00:10   |     6,084 |  3,718 |    551 |  1,236 |    579 |      0
-
-> Workload Finished.
-
-  SUMMARY
-  --------------------------------------------------
-  Runtime:    10.00s
-  Total Ops:  81,746
-  Avg Rate:   8,174 ops/sec
-
-  LATENCY DISTRIBUTION (ms)
-  --------------------------------------------------
-  TYPE             AVG          P95          P99
-  ----             ---          ---          ---
-  SELECT       1.24 ms      4.00 ms      9.00 ms
-  INSERT      12.11 ms     66.00 ms     73.00 ms
-  UPDATE       9.71 ms     65.00 ms     71.00 ms
-  DELETE       9.60 ms     65.00 ms     72.00 ms
-  AGG                -            -            -
-```
+
+![genMongoLoad Demo](./genMongoLoadDemo.gif)
 
 ---
 
@@ -296,8 +240,8 @@ To run your own workload against your own schema:
     ```json
     [
       {
-        "databaseName": "ecommerce",
-        "collectionName": "orders",
+        "database": "ecommerce",
+        "collection": "orders",
         "fields": {
           "_id": { "type": "objectid" },
           "customer_name": { "type": "string", "provider": "first_name" },

genMongoLoadDemo.gif

@@ -32,8 +32,8 @@ type ShardConfig struct {
 }
 
 type CollectionDefinition struct {
-	DatabaseName string                     `json:"databaseName"`
-	Name         string                     `json:"collectionName"`
+	DatabaseName string                     `json:"database"`
+	Name         string                     `json:"collection"`
 	Fields       map[string]CollectionField `json:"fields"`
 	Indexes      []IndexDefinition          `json:"indexes,omitempty"`
 	ShardConfig  *ShardConfig               `json:"shardConfig,omitempty"`

internal/config/collections.go

@@ -25,13 +25,23 @@ type LatencyHistogram struct {
 	Overflow int64
 	Count    int64
 	Sum      float64
+	Min      float64
+	Max      float64
 }
 
 func (h *LatencyHistogram) Record(ms float64) {
 	h.mu.Lock()
 	defer h.mu.Unlock()
 	h.Count++
 	h.Sum += ms
+
+	if ms < h.Min {
+		h.Min = ms
+	}
+	if ms > h.Max {
+		h.Max = ms
+	}
+
 	bucket := int(math.Round(ms))
 	if bucket < 0 {
 		bucket = 0
@@ -88,15 +98,16 @@ type Collector struct {
 
 func NewCollector() *Collector {
 	return &Collector{
-		FindHist:   &LatencyHistogram{},
-		InsertHist: &LatencyHistogram{},
-		UpdateHist: &LatencyHistogram{},
-		DeleteHist: &LatencyHistogram{},
-		AggHist:    &LatencyHistogram{},
+		FindHist:   &LatencyHistogram{Min: math.MaxFloat64},
+		InsertHist: &LatencyHistogram{Min: math.MaxFloat64},
+		UpdateHist: &LatencyHistogram{Min: math.MaxFloat64},
+		DeleteHist: &LatencyHistogram{Min: math.MaxFloat64},
+		AggHist:    &LatencyHistogram{Min: math.MaxFloat64},
 		startTime:  time.Now(),
 	}
 }
 
+// CHANGED: Second argument is now []config.CollectionDefinition instead of dbName string
 func PrintConfiguration(appCfg *config.AppConfig, collections []config.CollectionDefinition, version string) {
 	fmt.Println()
 	fmt.Printf("  %s\n", logger.CyanString("genMongoLoad %s", version))
@@ -139,28 +150,14 @@ func PrintConfiguration(appCfg *config.AppConfig, collections []config.Collectio
 	w.Flush()
 	fmt.Println()
 
-	// Logic to extract unique DBs and Namespaces
-	uniqueDBs := make(map[string]bool)
+	// Logic to print Namespaces instead of just Database
 	var namespaces []string
 	for _, col := range collections {
-		uniqueDBs[col.DatabaseName] = true
 		namespaces = append(namespaces, fmt.Sprintf("%s.%s", col.DatabaseName, col.Name))
 	}
 
-	var dbs []string
-	for db := range uniqueDBs {
-		dbs = append(dbs, db)
-	}
-	sort.Strings(dbs)
-	sort.Strings(namespaces)
-
 	w = tabwriter.NewWriter(os.Stdout, 0, 0, 2, ' ', 0)
-
-	// Display Namespaces
-	if len(namespaces) > 0 {
-		fmt.Fprintf(w, "  Namespaces:\t%s\n", strings.Join(namespaces, ", "))
-	}
-
+	fmt.Fprintf(w, "  Namespaces:\t%s\n", strings.Join(namespaces, ", ")) // Changed from Database
 	fmt.Fprintf(w, "  Workers:\t%d active\n", appCfg.Concurrency)
 	fmt.Fprintf(w, "  Duration:\t%s\n", appCfg.Duration)
 	fmt.Fprintf(w, "  Report Freq:\t%ds\n", appCfg.StatusRefreshRateSec)
@@ -303,11 +300,11 @@ func (c *Collector) PrintFinalSummary(duration time.Duration) {
 	fmt.Println(logger.BoldString("  LATENCY DISTRIBUTION (ms)"))
 	fmt.Println(logger.CyanString("  --------------------------------------------------"))
 
-	const tableLayout = "  %-7s   %10s   %10s   %10s"
-	headerStr := fmt.Sprintf(tableLayout, "TYPE", "AVG", "P95", "P99")
+	const tableLayout = "  %-7s   %10s   %10s   %10s   %10s   %10s"
+	headerStr := fmt.Sprintf(tableLayout, "TYPE", "AVG", "MIN", "MAX", "P95", "P99")
 	fmt.Println(logger.BoldString(headerStr))
 
-	separatorStr := fmt.Sprintf(tableLayout, "----", "---", "---", "---")
+	separatorStr := fmt.Sprintf(tableLayout, "----", "---", "---", "---", "---", "---")
 	fmt.Println(logger.CyanString(separatorStr))
 
 	printLatencyRow(tableLayout, "SELECT", c.FindHist)
@@ -320,14 +317,15 @@ func (c *Collector) PrintFinalSummary(duration time.Duration) {
 
 func printLatencyRow(layout string, label string, h *LatencyHistogram) {
 	if h.Count == 0 {
-		fmt.Printf(layout+"\n", label, "-", "-", "-")
+		fmt.Printf(layout+"\n", label, "-", "-", "-", "-", "-")
 		return
 	}
 	avgMs := h.Sum / float64(h.Count)
 	p95Ms := h.GetPercentile(95.0)
 	p99Ms := h.GetPercentile(99.0)
-	fmt.Printf(layout+"\n", label, formatLatency(avgMs), formatLatency(p95Ms), formatLatency(p99Ms))
+	fmt.Printf(layout+"\n", label, formatLatency(avgMs), formatLatency(h.Min), formatLatency(h.Max), formatLatency(p95Ms), formatLatency(p99Ms))
 }
+
 func formatLatency(ms float64) string {
 	if ms < 1000.0 {
 		return fmt.Sprintf("%.2f ms", ms)
@@ -337,6 +335,7 @@ func formatLatency(ms float64) string {
 	}
 	return fmt.Sprintf("%.2f m", ms/60000.0)
 }
+
 func formatInt(n int64) string {
 	in := strconv.FormatInt(n, 10)
 	numOfDigits := len(in)

internal/stats/collector.go

@@ -1,7 +1,7 @@
 [
     {
-        "databaseName": "shop",
-        "collectionName": "orders",
+        "database": "shop",
+        "collection": "orders",
         "fields": {
             "_id": {
                 "type": "objectid"

resources/collections/complex_collection.json

@@ -1,7 +1,7 @@
 [
     {
-        "databaseName": "airline",
-        "collectionName": "flights",
+        "database": "airline",
+        "collection": "flights",
         "shardConfig": {
             "key": {
                 "flight_id": "hashed"

resources/collections/default.json

@@ -1,7 +1,7 @@
 [
     {
-        "databaseName": "generic",
-        "collectionName": "sample",
+        "database": "generic",
+        "collection": "sample",
         "shardConfig": {
             "key": {
                 "field_int32": "hashed"