Merge #74377 #76837 #76986

74377: server: hot ranges api r=koorosh a=koorosh This change implements second version of hot ranges api that's required for UI to represent enhanced Hot Ranges view. It reuses former implementation but doesn't expose sensitive internal data. Instead, it provides flat structure of fields required for ui. New `HotRangeV2` endpoint is established to: - maintain backward compatibility - do not expose internal data as it's done in `HotRange` endpoint Release note: None Release justification: bug fixes and low-risk updates to new functionality 76837: ttljob: add row statistics r=rafiss a=otan This commit adds the ability for the TTL job to generate statistics on number of rows and number of expired rows on the table. This is off by default, controllable by the `ttl_automatic_stats_poll_interval` storage parameter syntax. Release justification: low risk high benefit changes to new functionality 76986: kvserver: disable assertion in SetPriorityID r=erikgrinaker a=tbg Merge-to-master is currently pretty red, and this has popped up once or twice. Disable the assertion (which is fine for tests, for now) while we investigate. Touches #75939 Release justification: temporarily de-flakes tests. Release note: None Co-authored-by: Andrii Vorobiov <[email protected]> Co-authored-by: Oliver Tan <[email protected]> Co-authored-by: Tobias Grieger <[email protected]>
cockroachdb · Mar 2, 2022 · bb54797 · bb54797
4 parents bd5def5 + af62e80 + 6cf5b12 + 86f51ef
commit bb54797
Show file tree

Hide file tree

Showing 27 changed files with 796 additions and 62 deletions.
diff --git a/docs/generated/http/full.md b/docs/generated/http/full.md
@@ -3192,6 +3192,8 @@ of ranges currently considered “hot” by the node(s).
 | Field | Type | Label | Description | Support status |
 | ----- | ---- | ----- | ----------- | -------------- |
 | node_id | [string](#cockroach.server.serverpb.HotRangesRequest-string) |  | NodeID indicates which node to query for a hot range report. It is possible to populate any node ID; if the node receiving the request is not the target node, it will forward the request to the target node.<br><br>If left empty, the request is forwarded to every node in the cluster. | [alpha](#support-status) |
+| page_size | [int32](#cockroach.server.serverpb.HotRangesRequest-int32) |  |  | [reserved](#support-status) |
+| page_token | [string](#cockroach.server.serverpb.HotRangesRequest-string) |  |  | [reserved](#support-status) |
 
 
 
@@ -3271,6 +3273,91 @@ target node(s) selected in a HotRangesRequest.
 | ----- | ---- | ----- | ----------- | -------------- |
 | desc | [cockroach.roachpb.RangeDescriptor](#cockroach.server.serverpb.HotRangesResponse-cockroach.roachpb.RangeDescriptor) |  | Desc is the descriptor of the range for which the report was produced.<br><br>TODO(knz): This field should be removed. See: https://github.com/cockroachdb/cockroach/issues/53212 | [reserved](#support-status) |
 | queries_per_second | [double](#cockroach.server.serverpb.HotRangesResponse-double) |  | QueriesPerSecond is the recent number of queries per second on this range. | [alpha](#support-status) |
+| leaseholder_node_id | [int32](#cockroach.server.serverpb.HotRangesResponse-int32) |  | LeaseholderNodeID indicates the Node ID that is the current leaseholder for the given range. | [reserved](#support-status) |
+
+
+
+
+
+
+## HotRangesV2
+
+`POST /_status/v2/hotranges`
+
+
+
+Support status: [reserved](#support-status)
+
+#### Request Parameters
+
+
+
+
+HotRangesRequest queries one or more cluster nodes for a list
+of ranges currently considered “hot” by the node(s).
+
+
+| Field | Type | Label | Description | Support status |
+| ----- | ---- | ----- | ----------- | -------------- |
+| node_id | [string](#cockroach.server.serverpb.HotRangesRequest-string) |  | NodeID indicates which node to query for a hot range report. It is possible to populate any node ID; if the node receiving the request is not the target node, it will forward the request to the target node.<br><br>If left empty, the request is forwarded to every node in the cluster. | [alpha](#support-status) |
+| page_size | [int32](#cockroach.server.serverpb.HotRangesRequest-int32) |  |  | [reserved](#support-status) |
+| page_token | [string](#cockroach.server.serverpb.HotRangesRequest-string) |  |  | [reserved](#support-status) |
+
+
+
+
+
+
+
+#### Response Parameters
+
+
+
+
+HotRangesResponseV2 is a response payload returned by `HotRangesV2` service.
+
+
+| Field | Type | Label | Description | Support status |
+| ----- | ---- | ----- | ----------- | -------------- |
+| ranges | [HotRangesResponseV2.HotRange](#cockroach.server.serverpb.HotRangesResponseV2-cockroach.server.serverpb.HotRangesResponseV2.HotRange) | repeated | Ranges contain list of hot ranges info that has highest number of QPS. | [reserved](#support-status) |
+| errors_by_node_id | [HotRangesResponseV2.ErrorsByNodeIdEntry](#cockroach.server.serverpb.HotRangesResponseV2-cockroach.server.serverpb.HotRangesResponseV2.ErrorsByNodeIdEntry) | repeated | errors contains any errors that occurred during fan-out calls to other nodes. | [reserved](#support-status) |
+| next_page_token | [string](#cockroach.server.serverpb.HotRangesResponseV2-string) |  | NextPageToken represents next pagination token to request next slice of data. | [reserved](#support-status) |
+
+
+
+
+
+
+<a name="cockroach.server.serverpb.HotRangesResponseV2-cockroach.server.serverpb.HotRangesResponseV2.HotRange"></a>
+#### HotRangesResponseV2.HotRange
+
+HotRange message describes a single hot range, ie its QPS, node ID it belongs to, etc.
+
+| Field | Type | Label | Description | Support status |
+| ----- | ---- | ----- | ----------- | -------------- |
+| range_id | [int32](#cockroach.server.serverpb.HotRangesResponseV2-int32) |  | range_id indicates Range ID that's identified as hot range. | [reserved](#support-status) |
+| node_id | [int32](#cockroach.server.serverpb.HotRangesResponseV2-int32) |  | node_id indicates the node that contains the current hot range. | [reserved](#support-status) |
+| qps | [double](#cockroach.server.serverpb.HotRangesResponseV2-double) |  | qps (queries per second) shows the amount of queries that interact with current range. | [reserved](#support-status) |
+| table_name | [string](#cockroach.server.serverpb.HotRangesResponseV2-string) |  | table_name indicates the SQL table that the range belongs to. | [reserved](#support-status) |
+| database_name | [string](#cockroach.server.serverpb.HotRangesResponseV2-string) |  | database_name indicates on database that has current hot range. | [reserved](#support-status) |
+| index_name | [string](#cockroach.server.serverpb.HotRangesResponseV2-string) |  | index_name indicates the index name for current range. | [reserved](#support-status) |
+| replica_node_ids | [int32](#cockroach.server.serverpb.HotRangesResponseV2-int32) | repeated | replica_node_ids specifies the list of node ids that contain replicas with current hot range. | [reserved](#support-status) |
+| leaseholder_node_id | [int32](#cockroach.server.serverpb.HotRangesResponseV2-int32) |  | leaseholder_node_id indicates the Node ID that is the current leaseholder for the given range. | [reserved](#support-status) |
+| schema_name | [string](#cockroach.server.serverpb.HotRangesResponseV2-string) |  | schema_name provides the name of schema (if exists) for table in current range. | [reserved](#support-status) |
+
+
+
+
+
+<a name="cockroach.server.serverpb.HotRangesResponseV2-cockroach.server.serverpb.HotRangesResponseV2.ErrorsByNodeIdEntry"></a>
+#### HotRangesResponseV2.ErrorsByNodeIdEntry
+
+
+
+| Field | Type | Label | Description | Support status |
+| ----- | ---- | ----- | ----------- | -------------- |
+| key | [int32](#cockroach.server.serverpb.HotRangesResponseV2-int32) |  |  |  |
+| value | [string](#cockroach.server.serverpb.HotRangesResponseV2-string) |  |  |  |
 
 
 

diff --git a/docs/generated/http/hotranges-other.md b/docs/generated/http/hotranges-other.md
@@ -62,5 +62,6 @@ Support status: [alpha](#support-status)
 | ----- | ---- | ----- | ----------- | -------------- |
 | desc | [cockroach.roachpb.RangeDescriptor](#cockroach.roachpb.RangeDescriptor) |  | Desc is the descriptor of the range for which the report was produced.<br><br>TODO(knz): This field should be removed. See: https://github.com/cockroachdb/cockroach/issues/53212 | [reserved](#support-status) |
 | queries_per_second | [double](#double) |  | QueriesPerSecond is the recent number of queries per second on this range. | [alpha](#support-status) |
+| leaseholder_node_id | [int32](#int32) |  | LeaseholderNodeID indicates the Node ID that is the current leaseholder for the given range. | [reserved](#support-status) |
 
 
diff --git a/docs/generated/http/hotranges-request.md b/docs/generated/http/hotranges-request.md
@@ -12,5 +12,7 @@ Support status: [alpha](#support-status)
 | Field | Type | Label | Description | Support status |
 | ----- | ---- | ----- | ----------- | -------------- |
 | node_id | [string](#string) |  | NodeID indicates which node to query for a hot range report. It is possible to populate any node ID; if the node receiving the request is not the target node, it will forward the request to the target node.<br><br>If left empty, the request is forwarded to every node in the cluster. | [alpha](#support-status) |
+| page_size | [int32](#int32) |  |  | [reserved](#support-status) |
+| page_token | [string](#string) |  |  | [reserved](#support-status) |
 
 
diff --git a/docs/generated/swagger/spec.json b/docs/generated/swagger/spec.json
@@ -1023,6 +1023,12 @@
       },
       "x-go-package": "github.com/cockroachdb/cockroach/pkg/util/metric"
     },
+    "RangeID": {
+      "type": "integer",
+      "format": "int64",
+      "title": "A RangeID is a unique ID associated to a Raft consensus group.",
+      "x-go-package": "github.com/cockroachdb/cockroach/pkg/roachpb"
+    },
     "RangeProblems": {
       "type": "object",
       "title": "RangeProblems describes issues reported by a range. For internal use only.",
@@ -1788,6 +1794,51 @@
       },
       "x-go-package": "github.com/cockroachdb/cockroach/pkg/server"
     },
+    "hotRangeInfo": {
+      "description": "(ie its range ID, QPS, table name, etc.).",
+      "type": "object",
+      "title": "Hot range details struct describes common information about hot range,",
+      "properties": {
+        "database_name": {
+          "type": "string",
+          "x-go-name": "DatabaseName"
+        },
+        "index_name": {
+          "type": "string",
+          "x-go-name": "IndexName"
+        },
+        "leaseholder_node_id": {
+          "$ref": "#/definitions/NodeID"
+        },
+        "node_id": {
+          "$ref": "#/definitions/NodeID"
+        },
+        "qps": {
+          "type": "number",
+          "format": "double",
+          "x-go-name": "QPS"
+        },
+        "range_id": {
+          "$ref": "#/definitions/RangeID"
+        },
+        "replica_node_ids": {
+          "type": "array",
+          "items": {
+            "$ref": "#/definitions/NodeID"
+          },
+          "x-go-name": "ReplicaNodeIDs"
+        },
+        "schema_name": {
+          "type": "string",
+          "x-go-name": "SchemaName"
+        },
+        "table_name": {
+          "type": "string",
+          "x-go-name": "TableName"
+        }
+      },
+      "x-go-package": "github.com/cockroachdb/cockroach/pkg/server"
+    },
     "hotRangesResponse": {
       "type": "object",
       "title": "Response struct for listHotRanges.",
@@ -1797,15 +1848,12 @@
           "type": "string",
           "x-go-name": "Next"
         },
-        "ranges_by_node_id": {
-          "type": "object",
-          "additionalProperties": {
-            "type": "array",
-            "items": {
-              "$ref": "#/definitions/rangeDescriptorInfo"
-            }
+        "ranges": {
+          "type": "array",
+          "items": {
+            "$ref": "#/definitions/hotRangeInfo"
           },
-          "x-go-name": "RangesByNodeID"
+          "x-go-name": "Ranges"
         },
         "response_error": {
           "type": "array",

diff --git a/pkg/kv/kvserver/scheduler.go b/pkg/kv/kvserver/scheduler.go
@@ -121,7 +121,10 @@ func (q *rangeIDQueue) Len() int {
 }
 
 func (q *rangeIDQueue) SetPriorityID(id roachpb.RangeID) {
-	if q.priorityID != 0 && q.priorityID != id {
+	if q.priorityID != 0 && q.priorityID != id &&
+		// This assertion is temporarily disabled, see:
+		// https://github.com/cockroachdb/cockroach/issues/75939
+		false {
 		panic(fmt.Sprintf(
 			"priority range ID already set: old=%d, new=%d",
 			q.priorityID, id))

diff --git a/pkg/server/BUILD.bazel b/pkg/server/BUILD.bazel
@@ -142,6 +142,7 @@ go_library(
         "//pkg/spanconfig/spanconfigsqltranslator",
         "//pkg/spanconfig/spanconfigsqlwatcher",
         "//pkg/sql",
+        "//pkg/sql/catalog",
         "//pkg/sql/catalog/bootstrap",
         "//pkg/sql/catalog/catalogkeys",
         "//pkg/sql/catalog/catconstants",

diff --git a/pkg/server/api_v2_ranges.go b/pkg/server/api_v2_ranges.go
@@ -14,7 +14,6 @@ import (
 	"context"
 	"fmt"
 	"net/http"
-	"sort"
 	"strconv"
 	"strings"
 
@@ -415,13 +414,29 @@ type responseError struct {
 //
 // swagger:model hotRangesResponse
 type hotRangesResponse struct {
-	RangesByNodeID map[string][]rangeDescriptorInfo `json:"ranges_by_node_id"`
-	Errors         []responseError                  `json:"response_error,omitempty"`
+	Ranges []hotRangeInfo  `json:"ranges"`
+	Errors []responseError `json:"response_error,omitempty"`
 	// Continuation token for the next paginated call. Use as the `start`
 	// parameter.
 	Next string `json:"next,omitempty"`
 }
 
+// Hot range details struct describes common information about hot range,
+// (ie its range ID, QPS, table name, etc.).
+//
+// swagger:model hotRangeInfo
+type hotRangeInfo struct {
+	RangeID           roachpb.RangeID  `json:"range_id"`
+	NodeID            roachpb.NodeID   `json:"node_id"`
+	QPS               float64          `json:"qps"`
+	LeaseholderNodeID roachpb.NodeID   `json:"leaseholder_node_id"`
+	TableName         string           `json:"table_name"`
+	DatabaseName      string           `json:"database_name"`
+	IndexName         string           `json:"index_name"`
+	SchemaName        string           `json:"schema_name"`
+	ReplicaNodeIDs    []roachpb.NodeID `json:"replica_node_ids"`
+}
+
 // swagger:operation GET /ranges/hot/ listHotRanges
 //
 // List hot ranges
@@ -464,9 +479,7 @@ func (a *apiV2Server) listHotRanges(w http.ResponseWriter, r *http.Request) {
 	nodeIDStr := r.URL.Query().Get("node_id")
 	limit, start := getRPCPaginationValues(r)
 
-	response := &hotRangesResponse{
-		RangesByNodeID: make(map[string][]rangeDescriptorInfo),
-	}
+	response := &hotRangesResponse{}
 	var requestedNodes []roachpb.NodeID
 	if len(nodeIDStr) > 0 {
 		requestedNodeID, _, err := a.status.parseNodeID(nodeIDStr)
@@ -484,32 +497,29 @@ func (a *apiV2Server) listHotRanges(w http.ResponseWriter, r *http.Request) {
 	remoteRequest := serverpb.HotRangesRequest{NodeID: "local"}
 	nodeFn := func(ctx context.Context, client interface{}, nodeID roachpb.NodeID) (interface{}, error) {
 		status := client.(serverpb.StatusClient)
-		resp, err := status.HotRanges(ctx, &remoteRequest)
+		resp, err := status.HotRangesV2(ctx, &remoteRequest)
 		if err != nil || resp == nil {
 			return nil, err
 		}
-		rangeDescriptorInfos := make([]rangeDescriptorInfo, 0)
-		for _, store := range resp.HotRangesByNodeID[nodeID].Stores {
-			for _, hotRange := range store.HotRanges {
-				var r rangeDescriptorInfo
-				r.init(&hotRange.Desc)
-				r.StoreID = int32(store.StoreID)
-				r.QueriesPerSecond = hotRange.QueriesPerSecond
-				rangeDescriptorInfos = append(rangeDescriptorInfos, r)
+
+		var hotRangeInfos = make([]hotRangeInfo, len(resp.Ranges))
+		for i, r := range resp.Ranges {
+			hotRangeInfos[i] = hotRangeInfo{
+				RangeID:           r.RangeID,
+				NodeID:            r.NodeID,
+				QPS:               r.QPS,
+				LeaseholderNodeID: r.LeaseholderNodeID,
+				TableName:         r.TableName,
+				DatabaseName:      r.DatabaseName,
+				IndexName:         r.IndexName,
+				ReplicaNodeIDs:    r.ReplicaNodeIds,
+				SchemaName:        r.SchemaName,
 			}
 		}
-		sort.Slice(rangeDescriptorInfos, func(i, j int) bool {
-			if rangeDescriptorInfos[i].StoreID == rangeDescriptorInfos[j].StoreID {
-				return rangeDescriptorInfos[i].RangeID < rangeDescriptorInfos[j].RangeID
-			}
-			return rangeDescriptorInfos[i].StoreID < rangeDescriptorInfos[j].StoreID
-		})
-		return rangeDescriptorInfos, nil
+		return hotRangeInfos, nil
 	}
 	responseFn := func(nodeID roachpb.NodeID, resp interface{}) {
-		if hotRangesResp, ok := resp.([]rangeDescriptorInfo); ok {
-			response.RangesByNodeID[nodeID.String()] = hotRangesResp
-		}
+		response.Ranges = append(response.Ranges, resp.([]hotRangeInfo)...)
 	}
 	errorFn := func(nodeID roachpb.NodeID, err error) {
 		response.Errors = append(response.Errors, responseError{

diff --git a/pkg/server/api_v2_ranges_test.go b/pkg/server/api_v2_ranges_test.go
@@ -46,25 +46,16 @@ func TestHotRangesV2(t *testing.T) {
 	require.NoError(t, json.NewDecoder(resp.Body).Decode(&hotRangesResp))
 	require.NoError(t, resp.Body.Close())
 
-	if len(hotRangesResp.RangesByNodeID) == 0 {
+	if len(hotRangesResp.Ranges) == 0 {
 		t.Fatalf("didn't get hot range responses from any nodes")
 	}
 	if len(hotRangesResp.Errors) > 0 {
 		t.Errorf("got an error in hot range response from n%d: %v",
 			hotRangesResp.Errors[0].NodeID, hotRangesResp.Errors[0].ErrorMessage)
 	}
-
-	for nodeID, nodeResp := range hotRangesResp.RangesByNodeID {
-		if len(nodeResp) == 0 {
-			t.Fatalf("didn't get hot range response from node n%s", nodeID)
-		}
-		// We don't check for ranges being sorted by QPS, as this hot ranges
-		// report does not use that as its sort key (for stability across multiple
-		// pagination calls).
-		for _, r := range nodeResp {
-			if r.RangeID == 0 || (len(r.StartKey) == 0 && len(r.EndKey) == 0) {
-				t.Errorf("unexpected empty/unpopulated range descriptor: %+v", r)
-			}
+	for _, r := range hotRangesResp.Ranges {
+		if r.RangeID == 0 || r.NodeID == 0 {
+			t.Errorf("unexpected empty/unpopulated range descriptor: %+v", r)
 		}
 	}
 }