changefeedccl: only checkpoint leading spans above lead threshold

This patch updates the span-level checkpointing code to only checkpoint
spans that are above the lead threshold (formerly lag threshold) so that
we can at least maintain some minimal level of progress for spans that
we go through the effort of checkpointing.

It also renames `changefeed.frontier_highwater_lag_checkpoint_threshold`
to `changefeed.span_checkpoint.lead_threshold` for consistency.

Release note (ops change): The cluster setting
`changefeed.frontier_highwater_lag_checkpoint_threshold` has been
renamed to `changefeed.span_checkpoint.lead_threshold`. The old name
remains available for backwards-compatibility.

docs/generated/settings/settings-for-tenants.txt

@@ -17,7 +17,7 @@ changefeed.default_range_distribution_strategy	enumeration	default	configures ho
 changefeed.event_consumer_worker_queue_size	integer	16	if changefeed.event_consumer_workers is enabled, this setting sets the maxmimum number of events which a worker can buffer	application
 changefeed.event_consumer_workers	integer	0	the number of workers to use when processing events: <0 disables, 0 assigns a reasonable default, >0 assigns the setting value. for experimental/core changefeeds and changefeeds using parquet format, this is disabled	application
 changefeed.fast_gzip.enabled	boolean	true	use fast gzip implementation	application
-changefeed.frontier_highwater_lag_checkpoint_threshold	duration	10m0s	controls the maximum the high-water mark is allowed to lag behind the leading spans of the frontier before per-span checkpointing is enabled; if 0, checkpointing due to high-water lag is disabled	application
+changefeed.span_checkpoint.lag_threshold (alias: changefeed.frontier_highwater_lag_checkpoint_threshold)	duration	10m0s	the amount of time a changefeed's slowest spans must lag behind its fastest spans before a span-level checkpoint due to lag is written; if 0, span-level checkpoints due to lag is disabled	application
 changefeed.memory.per_changefeed_limit	byte size	512 MiB	controls amount of data that can be buffered per changefeed	application
 changefeed.resolved_timestamp.min_update_interval (alias: changefeed.min_highwater_advance)	duration	0s	minimum amount of time that must have elapsed since the last time a changefeed's resolved timestamp was updated before it is eligible to be updated again; default of 0 means no minimum interval is enforced but updating will still be limited by the average time it takes to checkpoint progress	application
 changefeed.node_throttle_config	string		specifies node level throttling configuration for all changefeeeds	application

docs/generated/settings/settings.html

@@ -22,7 +22,7 @@
 <tr><td><div id="setting-changefeed-event-consumer-worker-queue-size" class="anchored"><code>changefeed.event_consumer_worker_queue_size</code></div></td><td>integer</td><td><code>16</code></td><td>if changefeed.event_consumer_workers is enabled, this setting sets the maxmimum number of events which a worker can buffer</td><td>Serverless/Dedicated/Self-Hosted</td></tr>
 <tr><td><div id="setting-changefeed-event-consumer-workers" class="anchored"><code>changefeed.event_consumer_workers</code></div></td><td>integer</td><td><code>0</code></td><td>the number of workers to use when processing events: &lt;0 disables, 0 assigns a reasonable default, &gt;0 assigns the setting value. for experimental/core changefeeds and changefeeds using parquet format, this is disabled</td><td>Serverless/Dedicated/Self-Hosted</td></tr>
 <tr><td><div id="setting-changefeed-fast-gzip-enabled" class="anchored"><code>changefeed.fast_gzip.enabled</code></div></td><td>boolean</td><td><code>true</code></td><td>use fast gzip implementation</td><td>Serverless/Dedicated/Self-Hosted</td></tr>
-<tr><td><div id="setting-changefeed-frontier-highwater-lag-checkpoint-threshold" class="anchored"><code>changefeed.frontier_highwater_lag_checkpoint_threshold</code></div></td><td>duration</td><td><code>10m0s</code></td><td>controls the maximum the high-water mark is allowed to lag behind the leading spans of the frontier before per-span checkpointing is enabled; if 0, checkpointing due to high-water lag is disabled</td><td>Serverless/Dedicated/Self-Hosted</td></tr>
+<tr><td><div id="setting-changefeed-frontier-highwater-lag-checkpoint-threshold" class="anchored"><code>changefeed.span_checkpoint.lag_threshold<br />(alias: changefeed.frontier_highwater_lag_checkpoint_threshold)</code></div></td><td>duration</td><td><code>10m0s</code></td><td>the amount of time a changefeed&#39;s slowest spans must lag behind its fastest spans before a span-level checkpoint due to lag is written; if 0, span-level checkpoints due to lag is disabled</td><td>Serverless/Dedicated/Self-Hosted</td></tr>
 <tr><td><div id="setting-changefeed-memory-per-changefeed-limit" class="anchored"><code>changefeed.memory.per_changefeed_limit</code></div></td><td>byte size</td><td><code>512 MiB</code></td><td>controls amount of data that can be buffered per changefeed</td><td>Serverless/Dedicated/Self-Hosted</td></tr>
 <tr><td><div id="setting-changefeed-min-highwater-advance" class="anchored"><code>changefeed.resolved_timestamp.min_update_interval<br />(alias: changefeed.min_highwater_advance)</code></div></td><td>duration</td><td><code>0s</code></td><td>minimum amount of time that must have elapsed since the last time a changefeed&#39;s resolved timestamp was updated before it is eligible to be updated again; default of 0 means no minimum interval is enforced but updating will still be limited by the average time it takes to checkpoint progress</td><td>Serverless/Dedicated/Self-Hosted</td></tr>
 <tr><td><div id="setting-changefeed-node-throttle-config" class="anchored"><code>changefeed.node_throttle_config</code></div></td><td>string</td><td><code></code></td><td>specifies node level throttling configuration for all changefeeeds</td><td>Serverless/Dedicated/Self-Hosted</td></tr>

pkg/ccl/changefeedccl/changefeed_processors.go

@@ -884,9 +884,10 @@ func (ca *changeAggregator) noteResolvedSpan(resolved jobspb.ResolvedSpan) (retu
 
 	// At a lower frequency, we checkpoint specific spans in the job progress
 	// either in backfills or if the highwater mark is excessively lagging behind.
+	inBackfill := ca.frontier.InBackfill(resolved)
+	hasLeadingSpans, _ := ca.frontier.HasLeadingSpans(sv)
 	checkpointSpans := ca.spec.JobID != 0 && /* enterprise changefeed */
-		(ca.frontier.InBackfill(resolved) || ca.frontier.HasLaggingSpans(sv)) &&
-		canCheckpointSpans(sv, ca.lastSpanFlush)
+		(inBackfill || hasLeadingSpans) && canCheckpointSpans(sv, ca.lastSpanFlush)
 
 	if checkpointSpans {
 		defer func() {
@@ -1670,13 +1671,19 @@ func (cf *changeFrontier) maybeCheckpointJob(
 	// highwater mark remains fixed while other spans may significantly outpace
 	// it, therefore to avoid losing that progress on changefeed resumption we
 	// also store as many of those leading spans as we can in the job progress
-	updateCheckpoint := (inBackfill || cf.frontier.HasLaggingSpans(&cf.js.settings.SV)) && cf.js.canCheckpointSpans()
+	hasLeadingSpans, leadTS := cf.frontier.HasLeadingSpans(&cf.js.settings.SV)
+	updateCheckpoint := (inBackfill || hasLeadingSpans) && cf.js.canCheckpointSpans()
 
 	// If the highwater has moved an empty checkpoint will be saved
 	var checkpoint jobspb.ChangefeedProgress_Checkpoint
 	if updateCheckpoint {
 		maxBytes := changefeedbase.SpanCheckpointMaxBytes.Get(&cf.FlowCtx.Cfg.Settings.SV)
-		checkpoint = cf.frontier.MakeCheckpoint(maxBytes)
+		frontier := cf.frontier.Frontier()
+		if hasLeadingSpans {
+			checkpoint = cf.frontier.MakeCheckpoint(leadTS, maxBytes)
+		} else {
+			checkpoint = cf.frontier.MakeCheckpoint(frontier, maxBytes)
+		}
 	}
 
 	if updateCheckpoint || updateHighWater {

pkg/ccl/changefeedccl/changefeed_test.go

@@ -2340,7 +2340,7 @@ func TestChangefeedLaggingSpanCheckpointing(t *testing.T) {
 		context.Background(), &s.ClusterSettings().SV, 10*time.Millisecond)
 	changefeedbase.SpanCheckpointMaxBytes.Override(
 		context.Background(), &s.ClusterSettings().SV, 100<<20)
-	changefeedbase.FrontierHighwaterLagCheckpointThreshold.Override(
+	changefeedbase.SpanCheckpointLeadThreshold.Override(
 		context.Background(), &s.ClusterSettings().SV, 10*time.Millisecond)
 
 	// We'll start changefeed with the cursor.

pkg/ccl/changefeedccl/changefeedbase/settings.go

@@ -79,16 +79,20 @@ var SpanCheckpointInterval = settings.RegisterDurationSetting(
 	settings.WithName("changefeed.span_checkpoint.interval"),
 )
 
-// FrontierHighwaterLagCheckpointThreshold controls the amount the high-water
-// mark is allowed to lag behind the leading edge of the frontier before we
-// begin to attempt checkpointing spans above the high-water mark
-var FrontierHighwaterLagCheckpointThreshold = settings.RegisterDurationSetting(
+// SpanCheckpointLeadThreshold controls the amount of time a changefeed's
+// leading spans must be ahead of its slowest spans before a span-level
+// checkpoint is written.
+var SpanCheckpointLeadThreshold = settings.RegisterDurationSetting(
 	settings.ApplicationLevel,
 	"changefeed.frontier_highwater_lag_checkpoint_threshold",
-	"controls the maximum the high-water mark is allowed to lag behind the leading spans of the frontier before per-span checkpointing is enabled; if 0, checkpointing due to high-water lag is disabled",
+	"the amount of time a changefeed's leading spans must be ahead of its "+
+		"slowest spans before a span-level checkpoint to save leading span progress "+
+		"is written; if 0, span-level checkpoints due to leading spans is disabled",
 	10*time.Minute,
 	settings.NonNegativeDuration,
-	settings.WithPublic)
+	settings.WithPublic,
+	settings.WithName("changefeed.span_checkpoint.lead_threshold"),
+)
 
 // SpanCheckpointMaxBytes controls the maximum number of key bytes that will be added
 // to a span-level checkpoint record.

pkg/ccl/changefeedccl/checkpoint/checkpoint.go

@@ -18,25 +18,28 @@ import (
 type SpanIter func(forEachSpan span.Operation)
 
 // Make creates a checkpoint with as many spans that should be checkpointed (are
-// above the highwater mark) as can fit in maxBytes, along with the earliest
+// above the provided timestamp) as can fit in maxBytes, along with the earliest
 // timestamp of the checkpointed spans. A SpanGroup is used to merge adjacent
-// spans above the high-water mark.
+// spans above the provided timestamp.
 func Make(
-	frontier hlc.Timestamp, forEachSpan SpanIter, maxBytes int64,
+	baselineTS hlc.Timestamp, forEachSpan SpanIter, maxBytes int64,
 ) jobspb.ChangefeedProgress_Checkpoint {
 	// Collect leading spans into a SpanGroup to merge adjacent spans and store
 	// the lowest timestamp found.
 	var checkpointSpanGroup roachpb.SpanGroup
-	checkpointTS := hlc.Timestamp{}
+	checkpointTS := hlc.MaxTimestamp
 	forEachSpan(func(s roachpb.Span, ts hlc.Timestamp) span.OpResult {
-		if frontier.Less(ts) {
+		if baselineTS.Less(ts) {
 			checkpointSpanGroup.Add(s)
-			if checkpointTS.IsEmpty() || ts.Less(checkpointTS) {
+			if ts.Less(checkpointTS) {
 				checkpointTS = ts
 			}
 		}
 		return span.ContinueMatch
 	})
+	if checkpointSpanGroup.Len() == 0 {
+		return jobspb.ChangefeedProgress_Checkpoint{}
+	}
 
 	// Ensure we only return up to maxBytes spans.
 	var checkpointSpans []roachpb.Span

pkg/ccl/changefeedccl/checkpoint/checkpoint_test.go

@@ -52,10 +52,11 @@ func TestCheckpointMake(t *testing.T) {
 	}
 
 	for name, tc := range map[string]struct {
-		frontier hlc.Timestamp
-		spans    checkpointSpans
-		maxBytes int64
-		expected jobspb.ChangefeedProgress_Checkpoint
+		frontier      hlc.Timestamp
+		spans         checkpointSpans
+		maxBytes      int64
+		leadThreshold int64
+		expected      jobspb.ChangefeedProgress_Checkpoint
 	}{
 		"all spans ahead of frontier checkpointed": {
 			frontier: ts(1),
@@ -126,6 +127,51 @@ func TestCheckpointMake(t *testing.T) {
 				Spans:     []roachpb.Span{{Key: roachpb.Key("b"), EndKey: roachpb.Key("d")}},
 			},
 		},
+		"all spans ahead of frontier checkpointed due to lead threshold": {
+			frontier: ts(1),
+			spans: checkpointSpans{
+				{span: roachpb.Span{Key: roachpb.Key("a"), EndKey: roachpb.Key("b")}, ts: ts(1)},
+				{span: roachpb.Span{Key: roachpb.Key("b"), EndKey: roachpb.Key("c")}, ts: ts(3)},
+				{span: roachpb.Span{Key: roachpb.Key("c"), EndKey: roachpb.Key("d")}, ts: ts(1)},
+				{span: roachpb.Span{Key: roachpb.Key("d"), EndKey: roachpb.Key("e")}, ts: ts(4)},
+			},
+			maxBytes:      100,
+			leadThreshold: 1,
+			expected: jobspb.ChangefeedProgress_Checkpoint{
+				Timestamp: ts(3),
+				Spans: []roachpb.Span{
+					{Key: roachpb.Key("b"), EndKey: roachpb.Key("c")},
+					{Key: roachpb.Key("d"), EndKey: roachpb.Key("e")},
+				},
+			},
+		},
+		"only some spans ahead of frontier checkpointed due to lead threshold": {
+			frontier: ts(1),
+			spans: checkpointSpans{
+				{span: roachpb.Span{Key: roachpb.Key("a"), EndKey: roachpb.Key("b")}, ts: ts(1)},
+				{span: roachpb.Span{Key: roachpb.Key("b"), EndKey: roachpb.Key("c")}, ts: ts(3)},
+				{span: roachpb.Span{Key: roachpb.Key("c"), EndKey: roachpb.Key("d")}, ts: ts(1)},
+				{span: roachpb.Span{Key: roachpb.Key("d"), EndKey: roachpb.Key("e")}, ts: ts(4)},
+			},
+			maxBytes:      100,
+			leadThreshold: 2,
+			expected: jobspb.ChangefeedProgress_Checkpoint{
+				Timestamp: ts(4),
+				Spans:     []roachpb.Span{{Key: roachpb.Key("d"), EndKey: roachpb.Key("e")}},
+			},
+		},
+		"no spans ahead of frontier checkpointed due to lead threshold": {
+			frontier: ts(1),
+			spans: checkpointSpans{
+				{span: roachpb.Span{Key: roachpb.Key("a"), EndKey: roachpb.Key("b")}, ts: ts(1)},
+				{span: roachpb.Span{Key: roachpb.Key("b"), EndKey: roachpb.Key("c")}, ts: ts(3)},
+				{span: roachpb.Span{Key: roachpb.Key("c"), EndKey: roachpb.Key("d")}, ts: ts(1)},
+				{span: roachpb.Span{Key: roachpb.Key("d"), EndKey: roachpb.Key("e")}, ts: ts(4)},
+			},
+			maxBytes:      100,
+			leadThreshold: 3,
+			expected:      jobspb.ChangefeedProgress_Checkpoint{},
+		},
 	} {
 		t.Run(name, func(t *testing.T) {
 			forEachSpan := func(fn span.Operation) {
@@ -134,7 +180,8 @@ func TestCheckpointMake(t *testing.T) {
 				}
 			}
 
-			actual := checkpoint.Make(tc.frontier, forEachSpan, tc.maxBytes)
+			baselineTS := tc.frontier.Add(tc.leadThreshold, 0)
+			actual := checkpoint.Make(baselineTS, forEachSpan, tc.maxBytes)
 			require.Equal(t, tc.expected, actual)
 		})
 	}

pkg/ccl/changefeedccl/resolvedspan/frontier.go

@@ -170,7 +170,9 @@ func (f *CoordinatorFrontier) InBackfill(r jobspb.ResolvedSpan) bool {
 }
 
 // MakeCheckpoint creates a checkpoint based on the current state of the frontier.
-func (f *CoordinatorFrontier) MakeCheckpoint(maxBytes int64) jobspb.ChangefeedProgress_Checkpoint {
+func (f *CoordinatorFrontier) MakeCheckpoint(
+	baselineTS hlc.Timestamp, maxBytes int64,
+) jobspb.ChangefeedProgress_Checkpoint {
 	return checkpoint.Make(f.Frontier(), f.Entries, maxBytes)
 }
 
@@ -293,19 +295,21 @@ func (f *resolvedSpanFrontier) assertBoundaryNotEarlier(
 	return nil
 }
 
-// HasLaggingSpans returns whether the frontier has lagging spans as defined
+// HasLeadingSpans returns whether the frontier has lagging spans as defined
 // by whether the frontier trails the latest timestamp by at least
-// changefeedbase.FrontierHighwaterLagCheckpointThreshold.
-func (f *resolvedSpanFrontier) HasLaggingSpans(sv *settings.Values) bool {
-	lagThresholdNanos := int64(changefeedbase.FrontierHighwaterLagCheckpointThreshold.Get(sv))
+// changefeedbase.SpanCheckpointLeadThreshold. If there are, it returns the
+// lead timestamp (frontier + lead threshold).
+func (f *resolvedSpanFrontier) HasLeadingSpans(sv *settings.Values) (bool, hlc.Timestamp) {
+	lagThresholdNanos := int64(changefeedbase.SpanCheckpointLeadThreshold.Get(sv))
 	if lagThresholdNanos == 0 {
-		return false
+		return false, hlc.Timestamp{}
 	}
 	frontier := f.Frontier()
 	if frontier.IsEmpty() {
 		frontier = f.statementTime
 	}
-	return frontier.Add(lagThresholdNanos, 0).Less(f.latestTS)
+	leadTS := frontier.Add(lagThresholdNanos, 0)
+	return leadTS.Less(f.latestTS), leadTS
 }
 
 // resolvedSpanBoundary encapsulates a resolved span boundary, which is