Merge cockroachdb#129001

129001: ui: display very large ranges in problematic ranges r=iskettaneh a=iskettaneh This commit adds a column in the DB Console's page problematic ranges that displays the ranges that are too large. The threshold is set to be: 8 * the current range max size. The two screenshots below show how that looks like. In order to generate that, I had to set the threshold multiplier to be 0.8 (instead of 8). <img width="600" alt="Screenshot 2024-08-15 at 11 13 28 AM" src="https://github.com/user-attachments/assets/66aae63a-1140-4079-a12e-b925110d975e"> <img width="600" alt="Screenshot 2024-08-15 at 11 14 15 AM" src="https://github.com/user-attachments/assets/fb7e9f07-1fd6-473c-bdd6-b902af3c3306"> Fixes: cockroachdb#127843 Epic: None Release note (ui change): Adding too large ranges to the problematic ranges page in DB Console. If a range is larger than twice the max range size, it will show up in the problematic ranges page. Co-authored-by: Ibrahim Kettaneh <[email protected]>
xinhaoz · Aug 20, 2024 · 7379814 · 7379814
2 parents b42ad64 + b0e1aaa
commit 7379814
Show file tree

Hide file tree

Showing 9 changed files with 82 additions and 21 deletions.
diff --git a/docs/generated/http/full.md b/docs/generated/http/full.md
@@ -1326,6 +1326,7 @@ RangeProblems describes issues reported by a range. For internal use only.
 | raft_log_too_large | [bool](#cockroach.server.serverpb.RaftDebugResponse-bool) |  | When the raft log is too large, it can be a symptom of other issues. | [reserved](#support-status) |
 | circuit_breaker_error | [bool](#cockroach.server.serverpb.RaftDebugResponse-bool) |  |  | [reserved](#support-status) |
 | paused_followers | [bool](#cockroach.server.serverpb.RaftDebugResponse-bool) |  |  | [reserved](#support-status) |
+| range_too_large | [bool](#cockroach.server.serverpb.RaftDebugResponse-bool) |  |  | [reserved](#support-status) |
 
 
 
@@ -1575,6 +1576,7 @@ RangeProblems describes issues reported by a range. For internal use only.
 | raft_log_too_large | [bool](#cockroach.server.serverpb.RangesResponse-bool) |  | When the raft log is too large, it can be a symptom of other issues. | [reserved](#support-status) |
 | circuit_breaker_error | [bool](#cockroach.server.serverpb.RangesResponse-bool) |  |  | [reserved](#support-status) |
 | paused_followers | [bool](#cockroach.server.serverpb.RangesResponse-bool) |  |  | [reserved](#support-status) |
+| range_too_large | [bool](#cockroach.server.serverpb.RangesResponse-bool) |  |  | [reserved](#support-status) |
 
 
 
@@ -3408,6 +3410,7 @@ Support status: [reserved](#support-status)
 | raft_log_too_large_range_ids | [int64](#cockroach.server.serverpb.ProblemRangesResponse-int64) | repeated |  | [reserved](#support-status) |
 | circuit_breaker_error_range_ids | [int64](#cockroach.server.serverpb.ProblemRangesResponse-int64) | repeated |  | [reserved](#support-status) |
 | paused_replica_ids | [int64](#cockroach.server.serverpb.ProblemRangesResponse-int64) | repeated |  | [reserved](#support-status) |
+| too_large_range_ids | [int64](#cockroach.server.serverpb.ProblemRangesResponse-int64) | repeated |  | [reserved](#support-status) |
 
 
 
@@ -3956,6 +3959,7 @@ RangeProblems describes issues reported by a range. For internal use only.
 | raft_log_too_large | [bool](#cockroach.server.serverpb.RangeResponse-bool) |  | When the raft log is too large, it can be a symptom of other issues. | [reserved](#support-status) |
 | circuit_breaker_error | [bool](#cockroach.server.serverpb.RangeResponse-bool) |  |  | [reserved](#support-status) |
 | paused_followers | [bool](#cockroach.server.serverpb.RangeResponse-bool) |  |  | [reserved](#support-status) |
+| range_too_large | [bool](#cockroach.server.serverpb.RangeResponse-bool) |  |  | [reserved](#support-status) |
 
 
 

diff --git a/pkg/kv/kvserver/replica_metrics.go b/pkg/kv/kvserver/replica_metrics.go
@@ -52,6 +52,7 @@ type ReplicaMetrics struct {
 	Underreplicated          bool
 	Overreplicated           bool
 	RaftLogTooLarge          bool
+	RangeTooLarge            bool
 	BehindCount              int64
 	PausedFollowerCount      int64
 	PendingRaftProposalCount int64
@@ -109,6 +110,7 @@ func (r *Replica) Metrics(
 		lockTableMetrics:         lockTableMetrics,
 		raftLogSize:              r.mu.raftLogSize,
 		raftLogSizeTrusted:       r.mu.raftLogSizeTrusted,
+		rangeSize:                r.mu.state.Stats.Total(),
 		qpUsed:                   qpUsed,
 		qpCapacity:               qpCap,
 		paused:                   r.mu.pausedFollowers,
@@ -138,6 +140,7 @@ type calcReplicaMetricsInput struct {
 	lockTableMetrics         concurrency.LockTableMetrics
 	raftLogSize              int64
 	raftLogSizeTrusted       bool
+	rangeSize                int64
 	qpUsed, qpCapacity       int64 // quota pool used and capacity bytes
 	paused                   map[roachpb.ReplicaID]struct{}
 	pendingRaftProposalCount int64
@@ -164,8 +167,14 @@ func calcReplicaMetrics(d calcReplicaMetricsInput) ReplicaMetrics {
 		}
 	}
 
-	rangeCounter, unavailable, underreplicated, overreplicated := calcRangeCounter(
-		d.storeID, d.desc, d.leaseStatus, d.vitalityMap, d.conf.GetNumVoters(), d.conf.NumReplicas, d.clusterNodes)
+	const (
+		raftLogTooLargeMultiple = 4
+		rangeTooLargeMultiple   = 2
+	)
+	largeRangeThreshold := rangeTooLargeMultiple * d.conf.RangeMaxBytes
+	rangeCounter, unavailable, underreplicated, overreplicated, tooLarge := calcRangeCounter(
+		d.storeID, d.desc, d.leaseStatus, d.vitalityMap, d.conf.GetNumVoters(), d.conf.NumReplicas,
+		d.clusterNodes, largeRangeThreshold, d.rangeSize)
 
 	// The raft leader computes the number of raft entries that replicas are
 	// behind.
@@ -176,7 +185,6 @@ func calcReplicaMetrics(d calcReplicaMetricsInput) ReplicaMetrics {
 		leaderPausedFollowerCount = int64(len(d.paused))
 	}
 
-	const raftLogTooLargeMultiple = 4
 	return ReplicaMetrics{
 		Leader:                    leader,
 		LeaseValid:                validLease,
@@ -194,6 +202,7 @@ func calcReplicaMetrics(d calcReplicaMetricsInput) ReplicaMetrics {
 		Overreplicated:            overreplicated,
 		RaftLogTooLarge: d.raftLogSizeTrusted &&
 			d.raftLogSize > raftLogTooLargeMultiple*d.raftCfg.RaftLogTruncationThreshold,
+		RangeTooLarge:            tooLarge,
 		BehindCount:              leaderBehindCount,
 		PausedFollowerCount:      leaderPausedFollowerCount,
 		PendingRaftProposalCount: d.pendingRaftProposalCount,
@@ -217,9 +226,10 @@ func calcQuotaPoolPercentUsed(qpUsed, qpCapacity int64) int64 {
 
 // calcRangeCounter returns whether this replica is designated as the replica in
 // the range responsible for range-level metrics, whether the range doesn't have
-// a quorum of live voting replicas, and whether the range is currently
+// a quorum of live voting replicas, whether the range is currently
 // under-replicated (with regards to either the number of voting replicas or the
-// number of non-voting replicas).
+// number of non-voting replicas), and whether the range is considered too
+// large.
 //
 // Note: we compute an estimated range count across the cluster by counting the
 // leaseholder of each descriptor if it's live, otherwise the first live
@@ -232,7 +242,8 @@ func calcRangeCounter(
 	vitalityMap livenesspb.NodeVitalityMap,
 	numVoters, numReplicas int32,
 	clusterNodes int,
-) (rangeCounter, unavailable, underreplicated, overreplicated bool) {
+	rangeTooLargeThreshold, rangeSize int64,
+) (rangeCounter, unavailable, underreplicated, overreplicated, tooLarge bool) {
 	// If there is a live leaseholder (regardless of whether the lease is still
 	// valid) that leaseholder is responsible for range-level metrics.
 	if vitalityMap[leaseStatus.Lease.Replica.NodeID].IsLive(livenesspb.Metrics) {
@@ -267,6 +278,7 @@ func calcRangeCounter(
 		} else if neededVoters < liveVoters || neededNonVoters < liveNonVoters {
 			overreplicated = true
 		}
+		tooLarge = rangeSize > rangeTooLargeThreshold
 	}
 	return
 }

diff --git a/pkg/kv/kvserver/replica_metrics_test.go b/pkg/kv/kvserver/replica_metrics_test.go
@@ -55,9 +55,9 @@ func TestCalcRangeCounterIsLiveMap(t *testing.T) {
 		}))
 
 	{
-		ctr, down, under, over := calcRangeCounter(1100, threeVotersAndSingleNonVoter, leaseStatus, livenesspb.NodeVitalityMap{
+		ctr, down, under, over, _ := calcRangeCounter(1100, threeVotersAndSingleNonVoter, leaseStatus, livenesspb.NodeVitalityMap{
 			1000: livenesspb.FakeNodeVitality(true), // by NodeID
-		}, 3 /* numVoters */, 4 /* numReplicas */, 4 /* clusterNodes */)
+		}, 3 /* numVoters */, 4 /* numReplicas */, 4 /* clusterNodes */, 0 /* rangeTooLargeThreshold */, 0 /* rangeSize */)
 
 		require.True(t, ctr)
 		require.True(t, down)
@@ -66,9 +66,9 @@ func TestCalcRangeCounterIsLiveMap(t *testing.T) {
 	}
 
 	{
-		ctr, down, under, over := calcRangeCounter(1000, threeVotersAndSingleNonVoter, leaseStatus, livenesspb.NodeVitalityMap{
+		ctr, down, under, over, _ := calcRangeCounter(1000, threeVotersAndSingleNonVoter, leaseStatus, livenesspb.NodeVitalityMap{
 			1000: livenesspb.FakeNodeVitality(false),
-		}, 3 /* numVoters */, 4 /* numReplicas */, 4 /* clusterNodes */)
+		}, 3 /* numVoters */, 4 /* numReplicas */, 4 /* clusterNodes */, 0 /* rangeTooLargeThreshold */, 0 /* rangeSize */)
 
 		// Does not confuse a non-live entry for a live one. In other words,
 		// does not think that the liveness map has only entries for live nodes.
@@ -79,12 +79,12 @@ func TestCalcRangeCounterIsLiveMap(t *testing.T) {
 	}
 
 	{
-		ctr, down, under, over := calcRangeCounter(11, threeVotersAndSingleNonVoter, leaseStatus, livenesspb.NodeVitalityMap{
+		ctr, down, under, over, _ := calcRangeCounter(11, threeVotersAndSingleNonVoter, leaseStatus, livenesspb.NodeVitalityMap{
 			10:   livenesspb.FakeNodeVitality(true),
 			100:  livenesspb.FakeNodeVitality(true),
 			1000: livenesspb.FakeNodeVitality(true),
 			2000: livenesspb.FakeNodeVitality(true),
-		}, 3 /* numVoters */, 4 /* numReplicas */, 4 /* clusterNodes */)
+		}, 3 /* numVoters */, 4 /* numReplicas */, 4 /* clusterNodes */, 0 /* rangeTooLargeThreshold */, 0 /* rangeSize */)
 
 		require.True(t, ctr)
 		require.False(t, down)
@@ -94,12 +94,12 @@ func TestCalcRangeCounterIsLiveMap(t *testing.T) {
 
 	{
 		// Single non-voter dead
-		ctr, down, under, over := calcRangeCounter(11, oneVoterAndThreeNonVoters, leaseStatus, livenesspb.NodeVitalityMap{
+		ctr, down, under, over, _ := calcRangeCounter(11, oneVoterAndThreeNonVoters, leaseStatus, livenesspb.NodeVitalityMap{
 			10:   livenesspb.FakeNodeVitality(true),
 			100:  livenesspb.FakeNodeVitality(true),
 			1000: livenesspb.FakeNodeVitality(false),
 			2000: livenesspb.FakeNodeVitality(true),
-		}, 1 /* numVoters */, 4 /* numReplicas */, 4 /* clusterNodes */)
+		}, 1 /* numVoters */, 4 /* numReplicas */, 4 /* clusterNodes */, 0 /* rangeTooLargeThreshold */, 0 /* rangeSize */)
 
 		require.True(t, ctr)
 		require.False(t, down)
@@ -109,12 +109,12 @@ func TestCalcRangeCounterIsLiveMap(t *testing.T) {
 
 	{
 		// All non-voters are dead, but range is not unavailable
-		ctr, down, under, over := calcRangeCounter(11, oneVoterAndThreeNonVoters, leaseStatus, livenesspb.NodeVitalityMap{
+		ctr, down, under, over, _ := calcRangeCounter(11, oneVoterAndThreeNonVoters, leaseStatus, livenesspb.NodeVitalityMap{
 			10:   livenesspb.FakeNodeVitality(true),
 			100:  livenesspb.FakeNodeVitality(false),
 			1000: livenesspb.FakeNodeVitality(false),
 			2000: livenesspb.FakeNodeVitality(false),
-		}, 1 /* numVoters */, 4 /* numReplicas */, 4 /* clusterNodes */)
+		}, 1 /* numVoters */, 4 /* numReplicas */, 4 /* clusterNodes */, 0 /* rangeTooLargeThreshold */, 0 /* rangeSize */)
 
 		require.True(t, ctr)
 		require.False(t, down)
@@ -124,18 +124,38 @@ func TestCalcRangeCounterIsLiveMap(t *testing.T) {
 
 	{
 		// More non-voters than needed
-		ctr, down, under, over := calcRangeCounter(11, oneVoterAndThreeNonVoters, leaseStatus, livenesspb.NodeVitalityMap{
+		ctr, down, under, over, _ := calcRangeCounter(11, oneVoterAndThreeNonVoters, leaseStatus, livenesspb.NodeVitalityMap{
 			10:   livenesspb.FakeNodeVitality(true),
 			100:  livenesspb.FakeNodeVitality(true),
 			1000: livenesspb.FakeNodeVitality(true),
 			2000: livenesspb.FakeNodeVitality(true),
-		}, 1 /* numVoters */, 3 /* numReplicas */, 4 /* clusterNodes */)
+		}, 1 /* numVoters */, 3 /* numReplicas */, 4 /* clusterNodes */, 0 /* rangeTooLargeThreshold */, 0 /* rangeSize */)
 
 		require.True(t, ctr)
 		require.False(t, down)
 		require.False(t, under)
 		require.True(t, over)
 	}
+
+	{
+		// Range larger than the threshold.
+		ctr, _, _, _, large := calcRangeCounter(1100, threeVotersAndSingleNonVoter, leaseStatus, livenesspb.NodeVitalityMap{
+			1000: livenesspb.FakeNodeVitality(true), // by NodeID
+		}, 3 /* numVoters */, 4 /* numReplicas */, 4 /* clusterNodes */, 1000 /* rangeTooLargeThreshold */, 2000 /* rangeSize */)
+
+		require.True(t, ctr)
+		require.True(t, large)
+	}
+
+	{
+		ctr, _, _, _, large := calcRangeCounter(1000, threeVotersAndSingleNonVoter, leaseStatus, livenesspb.NodeVitalityMap{
+			1000: livenesspb.FakeNodeVitality(false),
+		}, 3 /* numVoters */, 4 /* numReplicas */, 4 /* clusterNodes */, 1000 /* rangeTooLargeThreshold */, 2000 /* rangeSize */)
+		require.False(t, ctr)
+		// Only the node responsible for the range can report if the range is too
+		// large.
+		require.False(t, large)
+	}
 }
 
 func TestCalcRangeCounterLeaseHolder(t *testing.T) {
@@ -242,8 +262,8 @@ func TestCalcRangeCounterLeaseHolder(t *testing.T) {
 			for _, nodeID := range tc.liveNodes {
 				livenessMap[nodeID] = livenesspb.FakeNodeVitality(true)
 			}
-			ctr, _, _, _ := calcRangeCounter(tc.storeID, rangeDesc, tc.leaseStatus, livenessMap,
-				3 /* numVoters */, 4 /* numReplicas */, 4 /* clusterNodes */)
+			ctr, _, _, _, _ := calcRangeCounter(tc.storeID, rangeDesc, tc.leaseStatus, livenessMap,
+				3 /* numVoters */, 4 /* numReplicas */, 4 /* clusterNodes */, 0 /* rangeTooLargeThreshold */, 0 /* rangeSize */)
 			require.Equal(t, tc.expectCounter, ctr)
 		})
 	}

diff --git a/pkg/server/problem_ranges.go b/pkg/server/problem_ranges.go
@@ -141,6 +141,10 @@ func (s *systemStatusServer) ProblemRanges(
 					problems.PausedReplicaIDs =
 						append(problems.PausedReplicaIDs, info.State.Desc.RangeID)
 				}
+				if info.Problems.RangeTooLarge {
+					problems.TooLargeRangeIds =
+						append(problems.TooLargeRangeIds, info.State.Desc.RangeID)
+				}
 			}
 			slices.Sort(problems.UnavailableRangeIDs)
 			slices.Sort(problems.RaftLeaderNotLeaseHolderRangeIDs)
@@ -152,6 +156,7 @@ func (s *systemStatusServer) ProblemRanges(
 			slices.Sort(problems.RaftLogTooLargeRangeIDs)
 			slices.Sort(problems.CircuitBreakerErrorRangeIDs)
 			slices.Sort(problems.PausedReplicaIDs)
+			slices.Sort(problems.TooLargeRangeIds)
 			response.ProblemsByNodeID[resp.nodeID] = problems
 		case <-ctx.Done():
 			return nil, status.Errorf(codes.DeadlineExceeded, ctx.Err().Error())

diff --git a/pkg/server/serverpb/status.proto b/pkg/server/serverpb/status.proto
@@ -409,6 +409,7 @@ message RangeProblems {
   bool raft_log_too_large = 7;
   bool circuit_breaker_error = 9;
   bool paused_followers = 10;
+  bool range_too_large = 11;
 }
 
 // RangeStatistics describes statistics reported by a range. For internal use
@@ -1324,6 +1325,11 @@ message ProblemRangesResponse {
       (gogoproto.casttype) =
           "github.com/cockroachdb/cockroach/pkg/roachpb.RangeID"
     ];
+    repeated int64 too_large_range_ids = 12 [
+      (gogoproto.customname) = "TooLargeRangeIds",
+      (gogoproto.casttype) =
+        "github.com/cockroachdb/cockroach/pkg/roachpb.RangeID"
+    ];
   }
   reserved 1 to 7;
   // NodeID is the node that submitted all the requests.

diff --git a/pkg/server/status.go b/pkg/server/status.go
@@ -2495,6 +2495,7 @@ func (s *systemStatusServer) rangesHelper(
 				NoLease:                metrics.Leader && !metrics.LeaseValid && !metrics.Quiescent,
 				QuiescentEqualsTicking: raftStatus != nil && metrics.Quiescent == metrics.Ticking,
 				RaftLogTooLarge:        metrics.RaftLogTooLarge,
+				RangeTooLarge:          metrics.RangeTooLarge,
 				CircuitBreakerError:    len(state.CircuitBreakerError) > 0,
 				PausedFollowers:        metrics.PausedFollowerCount > 0,
 			},

diff --git a/pkg/ui/workspaces/db-console/src/views/reports/containers/problemRanges/connectionsTable.tsx b/pkg/ui/workspaces/db-console/src/views/reports/containers/problemRanges/connectionsTable.tsx
@@ -82,6 +82,10 @@ const connectionTableColumns: ConnectionTableColumn[] = [
     title: "Paused Replicas",
     extract: problem => problem.paused_replica_ids.length,
   },
+  {
+    title: "Range Too Large",
+    extract: problem => problem.too_large_range_ids.length,
+  },
   {
     title: "Total",
     extract: problem => {
@@ -95,7 +99,8 @@ const connectionTableColumns: ConnectionTableColumn[] = [
         problem.quiescent_equals_ticking_range_ids.length +
         problem.raft_log_too_large_range_ids.length +
         problem.circuit_breaker_error_range_ids.length +
-        problem.paused_replica_ids.length
+        problem.paused_replica_ids.length +
+        problem.too_large_range_ids.length
       );
     },
   },

diff --git a/pkg/ui/workspaces/db-console/src/views/reports/containers/problemRanges/index.tsx b/pkg/ui/workspaces/db-console/src/views/reports/containers/problemRanges/index.tsx
@@ -230,6 +230,11 @@ export class ProblemRanges extends React.Component<ProblemRangesProps, {}> {
           problems={problems}
           extract={problem => problem.paused_replica_ids}
         />
+        <ProblemRangeList
+          name="Range Too Large"
+          problems={problems}
+          extract={problem => problem.too_large_range_ids}
+        />
       </div>
     );
   }

diff --git a/pkg/ui/workspaces/db-console/src/views/reports/containers/range/rangeTable.tsx b/pkg/ui/workspaces/db-console/src/views/reports/containers/range/rangeTable.tsx
@@ -492,6 +492,9 @@ export default class RangeTable extends React.Component<RangeTableProps, {}> {
     if (problems.raft_log_too_large) {
       results = concat(results, "Raft log too large");
     }
+    if (problems.range_too_large) {
+      results = concat(results, "Range too large");
+    }
     if (awaitingGC) {
       results = concat(results, "Awaiting GC");
     }