metrics: enlarge upper limits of some duration metrics (#17451) (#17648)

* cherry pick #17451 to release-4.0 Signed-off-by: sre-bot <sre-bot@pingcap.com> * resolve conflicts Signed-off-by: Shuaipeng Yu <jackysp@gmail.com> Co-authored-by: Jack Yu <yusp@pingcap.com> Co-authored-by: Shuaipeng Yu <jackysp@gmail.com>
pingcap · Jun 6, 2020 · a0ad6f6 · a0ad6f6
1 parent cf7467a
commit a0ad6f6
Show file tree

Hide file tree

Showing 8 changed files with 26 additions and 26 deletions.
diff --git a/metrics/ddl.go b/metrics/ddl.go
@@ -31,7 +31,7 @@ var (
 			Subsystem: "ddl",
 			Name:      "handle_job_duration_seconds",
 			Help:      "Bucketed histogram of processing time (s) of handle jobs",
-			Buckets:   prometheus.ExponentialBuckets(0.01, 2, 22), // 10ms ~ 6hours
+			Buckets:   prometheus.ExponentialBuckets(0.01, 2, 24), // 10ms ~ 24hours
 		}, []string{LblType, LblResult})
 
 	BatchAddIdxHistogram = prometheus.NewHistogramVec(
@@ -40,7 +40,7 @@ var (
 			Subsystem: "ddl",
 			Name:      "batch_add_idx_duration_seconds",
 			Help:      "Bucketed histogram of processing time (s) of batch handle data",
-			Buckets:   prometheus.ExponentialBuckets(0.001, 2, 22), // 1ms ~ 0.5hours
+			Buckets:   prometheus.ExponentialBuckets(0.001, 2, 28), // 1ms ~ 1.5days
 		}, []string{LblType})
 
 	SyncerInit            = "init"
@@ -91,7 +91,7 @@ var (
 			Subsystem: "ddl",
 			Name:      "worker_operation_duration_seconds",
 			Help:      "Bucketed histogram of processing time (s) of ddl worker operations",
-			Buckets:   prometheus.ExponentialBuckets(0.001, 2, 22), // 1ms ~ 2048s
+			Buckets:   prometheus.ExponentialBuckets(0.001, 2, 28), // 1ms ~ 1.5days
 		}, []string{LblType, LblAction, LblResult})
 
 	CreateDDLInstance = "create_ddl_instance"

diff --git a/metrics/distsql.go b/metrics/distsql.go
@@ -25,7 +25,7 @@ var (
 			Subsystem: "distsql",
 			Name:      "handle_query_duration_seconds",
 			Help:      "Bucketed histogram of processing time (s) of handled queries.",
-			Buckets:   prometheus.ExponentialBuckets(0.0005, 2, 18), // 0.5ms ~ 64s
+			Buckets:   prometheus.ExponentialBuckets(0.0005, 2, 29), // 0.5ms ~ 1.5days
 		}, []string{LblType, LblSQLType})
 
 	DistSQLScanKeysPartialHistogram = prometheus.NewHistogram(

diff --git a/metrics/meta.go b/metrics/meta.go
@@ -28,7 +28,7 @@ var (
 			Subsystem: "autoid",
 			Name:      "operation_duration_seconds",
 			Help:      "Bucketed histogram of processing time (s) of handled autoid.",
-			Buckets:   prometheus.ExponentialBuckets(0.0005, 2, 22), // 0.5ms ~ 1048s
+			Buckets:   prometheus.ExponentialBuckets(0.0005, 2, 29), // 0.5ms ~ 1.5days
 		}, []string{LblType, LblResult})
 
 	GetSchemaDiff    = "get_schema_diff"
@@ -43,6 +43,6 @@ var (
 			Subsystem: "meta",
 			Name:      "operation_duration_seconds",
 			Help:      "Bucketed histogram of processing time (s) of tidb meta data operations.",
-			Buckets:   prometheus.ExponentialBuckets(0.0005, 2, 22), // 0.5ms ~ 1048s
+			Buckets:   prometheus.ExponentialBuckets(0.0005, 2, 29), // 0.5ms ~ 1.5days
 		}, []string{LblType, LblResult})
 )
diff --git a/metrics/oracles.go b/metrics/oracles.go
@@ -25,6 +25,6 @@ var (
 			Subsystem: "pdclient",
 			Name:      "ts_future_wait_seconds",
 			Help:      "Bucketed histogram of seconds cost for waiting timestamp future.",
-			Buckets:   prometheus.ExponentialBuckets(0.000005, 2, 24), // 5us ~ 40s
+			Buckets:   prometheus.ExponentialBuckets(0.000005, 2, 30), // 5us ~ 2560s
 		})
 )
diff --git a/metrics/server.go b/metrics/server.go
@@ -34,7 +34,7 @@ var (
 			Subsystem: "server",
 			Name:      "handle_query_duration_seconds",
 			Help:      "Bucketed histogram of processing time (s) of handled queries.",
-			Buckets:   prometheus.ExponentialBuckets(0.0005, 2, 22), // 0.5ms ~ 1048s
+			Buckets:   prometheus.ExponentialBuckets(0.0005, 2, 29), // 0.5ms ~ 1.5days
 		}, []string{LblSQLType})
 
 	QueryTotalCounter = prometheus.NewCounterVec(
@@ -128,7 +128,7 @@ var (
 			Subsystem: "server",
 			Name:      "get_token_duration_seconds",
 			Help:      "Duration (us) for getting token, it should be small until concurrency limit is reached.",
-			Buckets:   prometheus.ExponentialBuckets(1, 2, 26), // 1us ~ 33s
+			Buckets:   prometheus.ExponentialBuckets(1, 2, 30), // 1us ~ 528s
 		})
 
 	TotalQueryProcHistogram = prometheus.NewHistogram(
@@ -137,23 +137,23 @@ var (
 			Subsystem: "server",
 			Name:      "slow_query_process_duration_seconds",
 			Help:      "Bucketed histogram of processing time (s) of of slow queries.",
-			Buckets:   prometheus.ExponentialBuckets(0.001, 2, 22), // 1ms ~ 2048s
+			Buckets:   prometheus.ExponentialBuckets(0.001, 2, 28), // 1ms ~ 1.5days
 		})
 	TotalCopProcHistogram = prometheus.NewHistogram(
 		prometheus.HistogramOpts{
 			Namespace: "tidb",
 			Subsystem: "server",
 			Name:      "slow_query_cop_duration_seconds",
 			Help:      "Bucketed histogram of all cop processing time (s) of of slow queries.",
-			Buckets:   prometheus.ExponentialBuckets(0.001, 2, 22), // 1ms ~ 2048s
+			Buckets:   prometheus.ExponentialBuckets(0.001, 2, 28), // 1ms ~ 1.5days
 		})
 	TotalCopWaitHistogram = prometheus.NewHistogram(
 		prometheus.HistogramOpts{
 			Namespace: "tidb",
 			Subsystem: "server",
 			Name:      "slow_query_wait_duration_seconds",
 			Help:      "Bucketed histogram of all cop waiting time (s) of of slow queries.",
-			Buckets:   prometheus.ExponentialBuckets(0.001, 2, 22), // 1ms ~ 2048s
+			Buckets:   prometheus.ExponentialBuckets(0.001, 2, 28), // 1ms ~ 1.5days
 		})
 )
 

diff --git a/metrics/session.go b/metrics/session.go
@@ -23,7 +23,7 @@ var (
 			Subsystem: "session",
 			Name:      "parse_duration_seconds",
 			Help:      "Bucketed histogram of processing time (s) in parse SQL.",
-			Buckets:   prometheus.ExponentialBuckets(0.00004, 2, 22), // 40us ~ 84s
+			Buckets:   prometheus.ExponentialBuckets(0.00004, 2, 28), // 40us ~ 1.5h
 		}, []string{LblSQLType})
 	SessionExecuteCompileDuration = prometheus.NewHistogramVec(
 		prometheus.HistogramOpts{
@@ -32,15 +32,15 @@ var (
 			Name:      "compile_duration_seconds",
 			Help:      "Bucketed histogram of processing time (s) in query optimize.",
 			// Build plan may execute the statement, or allocate table ID, so it might take a long time.
-			Buckets: prometheus.ExponentialBuckets(0.00004, 2, 22), // 40us ~ 84s
+			Buckets: prometheus.ExponentialBuckets(0.00004, 2, 28), // 40us ~ 1.5h
 		}, []string{LblSQLType})
 	SessionExecuteRunDuration = prometheus.NewHistogramVec(
 		prometheus.HistogramOpts{
 			Namespace: "tidb",
 			Subsystem: "session",
 			Name:      "execute_duration_seconds",
 			Help:      "Bucketed histogram of processing time (s) in running executor.",
-			Buckets:   prometheus.ExponentialBuckets(0.0001, 2, 22), // 100us ~ 209s
+			Buckets:   prometheus.ExponentialBuckets(0.0001, 2, 30), // 100us ~ 15h
 		}, []string{LblSQLType})
 	SchemaLeaseErrorCounter = prometheus.NewCounterVec(
 		prometheus.CounterOpts{
@@ -88,7 +88,7 @@ var (
 			Subsystem: "session",
 			Name:      "transaction_duration_seconds",
 			Help:      "Bucketed histogram of a transaction execution duration, including retry.",
-			Buckets:   prometheus.ExponentialBuckets(0.001, 2, 20), // 1ms ~ 524s
+			Buckets:   prometheus.ExponentialBuckets(0.001, 2, 28), // 1ms ~ 1.5days
 		}, []string{LblSQLType, LblType})
 
 	StatementDeadlockDetectDuration = prometheus.NewHistogram(
@@ -97,7 +97,7 @@ var (
 			Subsystem: "session",
 			Name:      "statement_deadlock_detect_duration_seconds",
 			Help:      "Bucketed histogram of a statement deadlock detect duration.",
-			Buckets:   prometheus.ExponentialBuckets(0.001, 2, 20), // 1ms ~ 524s
+			Buckets:   prometheus.ExponentialBuckets(0.001, 2, 28), // 1ms ~ 1.5days
 		},
 	)
 

diff --git a/metrics/stats.go b/metrics/stats.go
@@ -25,7 +25,7 @@ var (
 			Subsystem: "statistics",
 			Name:      "auto_analyze_duration_seconds",
 			Help:      "Bucketed histogram of processing time (s) of auto analyze.",
-			Buckets:   prometheus.ExponentialBuckets(0.01, 2, 20), // 10ms ~ 1.5hours
+			Buckets:   prometheus.ExponentialBuckets(0.01, 2, 24), // 10ms ~ 24h
 		})
 
 	AutoAnalyzeCounter = prometheus.NewCounterVec(

diff --git a/metrics/tikvclient.go b/metrics/tikvclient.go
@@ -23,7 +23,7 @@ var (
 			Subsystem: "tikvclient",
 			Name:      "txn_cmd_duration_seconds",
 			Help:      "Bucketed histogram of processing time of txn cmds.",
-			Buckets:   prometheus.ExponentialBuckets(0.0005, 2, 20), // 0.5ms ~ 262s
+			Buckets:   prometheus.ExponentialBuckets(0.0005, 2, 29), // 0.5ms ~ 1.5days
 		}, []string{LblType})
 
 	TiKVBackoffHistogram = prometheus.NewHistogramVec(
@@ -32,7 +32,7 @@ var (
 			Subsystem: "tikvclient",
 			Name:      "backoff_seconds",
 			Help:      "total backoff seconds of a single backoffer.",
-			Buckets:   prometheus.ExponentialBuckets(0.0005, 2, 20), // 0.5ms ~ 262s
+			Buckets:   prometheus.ExponentialBuckets(0.0005, 2, 29), // 0.5ms ~ 1.5days
 		}, []string{LblType})
 
 	TiKVSendReqHistogram = prometheus.NewHistogramVec(
@@ -41,7 +41,7 @@ var (
 			Subsystem: "tikvclient",
 			Name:      "request_seconds",
 			Help:      "Bucketed histogram of sending request duration.",
-			Buckets:   prometheus.ExponentialBuckets(0.0005, 2, 20), // 0.5ms ~ 262s
+			Buckets:   prometheus.ExponentialBuckets(0.0005, 2, 29), // 0.5ms ~ 1.5days
 		}, []string{LblType, LblStore})
 
 	TiKVCoprocessorHistogram = prometheus.NewHistogram(
@@ -50,7 +50,7 @@ var (
 			Subsystem: "tikvclient",
 			Name:      "cop_duration_seconds",
 			Help:      "Run duration of a single coprocessor task, includes backoff time.",
-			Buckets:   prometheus.ExponentialBuckets(0.0005, 2, 20), // 0.5ms ~ 262s
+			Buckets:   prometheus.ExponentialBuckets(0.0005, 2, 29), // 0.5ms ~ 1.5days
 		})
 
 	TiKVLockResolverCounter = prometheus.NewCounterVec(
@@ -93,7 +93,7 @@ var (
 			Subsystem: "tikvclient",
 			Name:      "rawkv_cmd_seconds",
 			Help:      "Bucketed histogram of processing time of rawkv cmds.",
-			Buckets:   prometheus.ExponentialBuckets(0.0005, 2, 20), // 0.5ms ~ 262s
+			Buckets:   prometheus.ExponentialBuckets(0.0005, 2, 29), // 0.5ms ~ 1.5days
 		}, []string{LblType})
 
 	TiKVRawkvSizeHistogram = prometheus.NewHistogramVec(
@@ -102,7 +102,7 @@ var (
 			Subsystem: "tikvclient",
 			Name:      "rawkv_kv_size_bytes",
 			Help:      "Size of key/value to put, in bytes.",
-			Buckets:   prometheus.ExponentialBuckets(1, 2, 24), // 1Byte ~ 8MB
+			Buckets:   prometheus.ExponentialBuckets(1, 2, 30), // 1Byte ~ 512MB
 		}, []string{LblType})
 
 	TiKVTxnRegionsNumHistogram = prometheus.NewHistogramVec(
@@ -111,7 +111,7 @@ var (
 			Subsystem: "tikvclient",
 			Name:      "txn_regions_num",
 			Help:      "Number of regions in a transaction.",
-			Buckets:   prometheus.ExponentialBuckets(1, 2, 20), // 1 ~ 1M
+			Buckets:   prometheus.ExponentialBuckets(1, 2, 25), // 1 ~ 16M
 		}, []string{LblType})
 
 	TiKVLoadSafepointCounter = prometheus.NewCounterVec(
@@ -187,7 +187,7 @@ var (
 			Namespace: "tidb",
 			Subsystem: "tikvclient",
 			Name:      "batch_client_unavailable_seconds",
-			Buckets:   prometheus.ExponentialBuckets(0.001, 2, 20), // 1ms ~ 524s
+			Buckets:   prometheus.ExponentialBuckets(0.001, 2, 28), // 1ms ~ 1.5days
 			Help:      "batch client unavailable",
 		})