Merge cockroachdb#50497

50497: sql: add telemetry for OOM during stats histogram generation r=mjibson a=mjibson With the addition of inverted stats, there's more memory usage during histogram collection. Record when an OOM occurs so we can decide whether to improve this situation. For example, we could split up histogram generation to be per column instead of all columns at the same time. Release note: None Co-authored-by: Matt Jibson <[email protected]>
lunevalex · Jun 26, 2020 · d3791a8 · d3791a8
2 parents c5133c6 + 7bedcaf
commit d3791a8
Show file tree

Hide file tree

Showing 3 changed files with 10 additions and 0 deletions.
diff --git a/pkg/sql/rowexec/sample_aggregator.go b/pkg/sql/rowexec/sample_aggregator.go
@@ -17,12 +17,14 @@ import (
 	"github.com/axiomhq/hyperloglog"
 	"github.com/cockroachdb/cockroach/pkg/jobs"
 	"github.com/cockroachdb/cockroach/pkg/kv"
+	"github.com/cockroachdb/cockroach/pkg/server/telemetry"
 	"github.com/cockroachdb/cockroach/pkg/sql/execinfra"
 	"github.com/cockroachdb/cockroach/pkg/sql/execinfrapb"
 	"github.com/cockroachdb/cockroach/pkg/sql/pgwire/pgcode"
 	"github.com/cockroachdb/cockroach/pkg/sql/pgwire/pgerror"
 	"github.com/cockroachdb/cockroach/pkg/sql/sem/tree"
 	"github.com/cockroachdb/cockroach/pkg/sql/sqlbase"
+	"github.com/cockroachdb/cockroach/pkg/sql/sqltelemetry"
 	"github.com/cockroachdb/cockroach/pkg/sql/stats"
 	"github.com/cockroachdb/cockroach/pkg/sql/types"
 	"github.com/cockroachdb/cockroach/pkg/util"
@@ -371,6 +373,7 @@ func (s *sampleAggregator) sampleRow(
 		// disable histogram sample collection.
 		sr.Disable()
 		log.Info(ctx, "disabling histogram collection due to excessive memory utilization")
+		telemetry.Inc(sqltelemetry.StatsHistogramOOMCounter)
 	}
 	return nil
 }

diff --git a/pkg/sql/rowexec/sampler.go b/pkg/sql/rowexec/sampler.go
@@ -18,12 +18,14 @@ import (
 
 	"github.com/axiomhq/hyperloglog"
 	"github.com/cockroachdb/cockroach/pkg/geo/geoindex"
+	"github.com/cockroachdb/cockroach/pkg/server/telemetry"
 	"github.com/cockroachdb/cockroach/pkg/sql/execinfra"
 	"github.com/cockroachdb/cockroach/pkg/sql/execinfrapb"
 	"github.com/cockroachdb/cockroach/pkg/sql/pgwire/pgcode"
 	"github.com/cockroachdb/cockroach/pkg/sql/pgwire/pgerror"
 	"github.com/cockroachdb/cockroach/pkg/sql/sem/tree"
 	"github.com/cockroachdb/cockroach/pkg/sql/sqlbase"
+	"github.com/cockroachdb/cockroach/pkg/sql/sqltelemetry"
 	"github.com/cockroachdb/cockroach/pkg/sql/stats"
 	"github.com/cockroachdb/cockroach/pkg/sql/types"
 	"github.com/cockroachdb/cockroach/pkg/util"
@@ -459,6 +461,7 @@ func (s *samplerProcessor) sampleRow(
 		// disable histogram sample collection.
 		sr.Disable()
 		log.Info(ctx, "disabling histogram collection due to excessive memory utilization")
+		telemetry.Inc(sqltelemetry.StatsHistogramOOMCounter)
 
 		// Send a metadata record so the sample aggregator will also disable
 		// histogram collection.

diff --git a/pkg/sql/sqltelemetry/planning.go b/pkg/sql/sqltelemetry/planning.go
@@ -113,6 +113,10 @@ var TurnAutoStatsOnUseCounter = telemetry.GetCounterOnce("sql.plan.automatic-sta
 // collection is explicitly disabled.
 var TurnAutoStatsOffUseCounter = telemetry.GetCounterOnce("sql.plan.automatic-stats.disabled")
 
+// StatsHistogramOOMCounter is to be incremented whenever statistics histogram
+// generation is disabled due to an out of memory error.
+var StatsHistogramOOMCounter = telemetry.GetCounterOnce("sql.plan.stats.histogram-oom")
+
 // JoinAlgoHashUseCounter is to be incremented whenever a hash join node is
 // planned.
 var JoinAlgoHashUseCounter = telemetry.GetCounterOnce("sql.plan.opt.node.join.algo.hash")