Merge #76719

76719: sql: extend resolverQueue to support waiter transactions r=Azhng a=Azhng

Reviewer note: only last two commits are relevant

---

# First commit

sql: add waiter txn id to extended contention event

Previously, contention event only include the txnID of the transaction
that held the lock (a.k.a. blocking transaction). The txnID of the
waiter transaction is missing from the contention event.

This commit includes the txnID of the waiter transaction into contention
events.

Release note: None

---

# Second commit

sql: extend resolverQueue to resolve waiter txn id

Previously, the resovlerQueue used in the contention event store only
resolved the txnID of the blocking transaction.
This commit, the resolverQueue would also resolve the txnID of the
waiting transaction.

Release note: None



Co-authored-by: Azhng <[email protected]>

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

@@ -79,6 +79,7 @@ server.web_session.purge.max_deletions_per_cycle	integer	10	the maximum number o
 server.web_session.purge.period	duration	1h0m0s	the time until old sessions are deleted
 server.web_session.purge.ttl	duration	1h0m0s	if nonzero, entries in system.web_sessions older than this duration are periodically purged
 server.web_session_timeout	duration	168h0m0s	the duration that a newly created web session will be valid
+sql.contention.event_store.capacity	byte size	64 MiB	the in-memory storage capacity per-node of contention event store
 sql.contention.txn_id_cache.max_size	byte size	0 B	the maximum byte size TxnID cache will use (set to 0 to disable)
 sql.cross_db_fks.enabled	boolean	false	if true, creating foreign key references across databases is allowed
 sql.cross_db_sequence_owners.enabled	boolean	false	if true, creating sequences owned by tables from other databases is allowed

docs/generated/settings/settings.html

@@ -91,6 +91,7 @@
 <tr><td><code>server.web_session.purge.period</code></td><td>duration</td><td><code>1h0m0s</code></td><td>the time until old sessions are deleted</td></tr>
 <tr><td><code>server.web_session.purge.ttl</code></td><td>duration</td><td><code>1h0m0s</code></td><td>if nonzero, entries in system.web_sessions older than this duration are periodically purged</td></tr>
 <tr><td><code>server.web_session_timeout</code></td><td>duration</td><td><code>168h0m0s</code></td><td>the duration that a newly created web session will be valid</td></tr>
+<tr><td><code>sql.contention.event_store.capacity</code></td><td>byte size</td><td><code>64 MiB</code></td><td>the in-memory storage capacity per-node of contention event store</td></tr>
 <tr><td><code>sql.contention.txn_id_cache.max_size</code></td><td>byte size</td><td><code>0 B</code></td><td>the maximum byte size TxnID cache will use (set to 0 to disable)</td></tr>
 <tr><td><code>sql.cross_db_fks.enabled</code></td><td>boolean</td><td><code>false</code></td><td>if true, creating foreign key references across databases is allowed</td></tr>
 <tr><td><code>sql.cross_db_sequence_owners.enabled</code></td><td>boolean</td><td><code>false</code></td><td>if true, creating sequences owned by tables from other databases is allowed</td></tr>

pkg/BUILD.bazel

@@ -254,6 +254,7 @@ ALL_TESTS = [
     "//pkg/sql/contention/contentionutils:contentionutils_test",
     "//pkg/sql/contention/txnidcache:txnidcache_test",
     "//pkg/sql/contention:contention_test",
+    "//pkg/sql/contentionpb:contentionpb_test",
     "//pkg/sql/covering:covering_test",
     "//pkg/sql/delegate:delegate_test",
     "//pkg/sql/distsql:distsql_test",

pkg/server/server_sql.go

@@ -659,6 +659,8 @@ func newSQLServer(ctx context.Context, cfg sqlServerArgs) (*SQLServer, error) {
 	if hasNodeLiveness {
 		traceCollector = collector.New(cfg.nodeDialer, nodeLiveness, cfg.Tracer)
 	}
+	contentionRegistry := contention.NewRegistry(cfg.Settings, cfg.sqlStatusServer.TxnIDResolution)
+	contentionRegistry.Start(ctx, cfg.stopper)
 
 	*execCfg = sql.ExecutorConfig{
 		Settings:                cfg.Settings,
@@ -681,7 +683,7 @@ func newSQLServer(ctx context.Context, cfg sqlServerArgs) (*SQLServer, error) {
 		SQLStatusServer:         cfg.sqlStatusServer,
 		RegionsServer:           cfg.regionsServer,
 		SessionRegistry:         cfg.sessionRegistry,
-		ContentionRegistry:      contention.NewRegistry(),
+		ContentionRegistry:      contentionRegistry,
 		SQLLiveness:             cfg.sqlLivenessProvider,
 		JobRegistry:             jobRegistry,
 		VirtualSchemas:          virtualSchemas,

pkg/sql/contention/BUILD.bazel

@@ -2,42 +2,62 @@ load("@io_bazel_rules_go//go:def.bzl", "go_library", "go_test")
 
 go_library(
     name = "contention",
-    srcs = ["registry.go"],
+    srcs = [
+        "cluster_settings.go",
+        "event_store.go",
+        "registry.go",
+        "resolver.go",
+        "test_utils.go",
+    ],
     importpath = "github.com/cockroachdb/cockroach/pkg/sql/contention",
     visibility = ["//visibility:public"],
     deps = [
         "//pkg/keys",
         "//pkg/roachpb",
+        "//pkg/server/serverpb",
+        "//pkg/settings",
+        "//pkg/settings/cluster",
         "//pkg/sql/catalog/descpb",
+        "//pkg/sql/contention/contentionutils",
         "//pkg/sql/contentionpb",
         "//pkg/util/cache",
+        "//pkg/util/log",
+        "//pkg/util/stop",
         "//pkg/util/syncutil",
+        "//pkg/util/timeutil",
         "//pkg/util/uuid",
         "@com_github_biogo_store//llrb",
+        "@com_github_cockroachdb_errors//:errors",
     ],
 )
 
 go_test(
     name = "contention_test",
     size = "small",
     srcs = [
+        "event_store_test.go",
         "registry_test.go",
+        "resolver_test.go",
         "utils_test.go",
     ],
     data = glob(["testdata/**"]),
     embed = [":contention"],
     deps = [
         "//pkg/keys",
         "//pkg/roachpb",
+        "//pkg/settings/cluster",
         "//pkg/sql/contentionpb",
         "//pkg/storage/enginepb",
         "//pkg/testutils",
         "//pkg/util/cache",
         "//pkg/util/encoding",
         "//pkg/util/leaktest",
         "//pkg/util/randutil",
+        "//pkg/util/stop",
+        "//pkg/util/timeutil",
         "//pkg/util/uuid",
         "@com_github_cockroachdb_datadriven//:datadriven",
+        "@com_github_cockroachdb_errors//:errors",
         "@com_github_stretchr_testify//require",
     ],
 )

pkg/sql/contention/cluster_settings.go

@@ -0,0 +1,36 @@
+// Copyright 2022 The Cockroach Authors.
+//
+// Use of this software is governed by the Business Source License
+// included in the file licenses/BSL.txt.
+//
+// As of the Change Date specified in that file, in accordance with
+// the Business Source License, use of this software will be governed
+// by the Apache License, Version 2.0, included in the file
+// licenses/APL.txt.
+
+package contention
+
+import (
+	"time"
+
+	"github.com/cockroachdb/cockroach/pkg/settings"
+)
+
+// TxnIDResolutionInterval is the cluster setting that controls how often the
+// Transaction ID Resolution is performed.
+var TxnIDResolutionInterval = settings.RegisterDurationSetting(
+	settings.TenantWritable,
+	"sql.contention.event_store.resolution_interval",
+	"the interval at which transaction fingerprint ID resolution is "+
+		"performed (set to 0 to disable)",
+	time.Second*30,
+)
+
+// StoreCapacity is the cluster setting that controls the
+// maximum size of the contention event store.
+var StoreCapacity = settings.RegisterByteSizeSetting(
+	settings.TenantWritable,
+	"sql.contention.event_store.capacity",
+	"the in-memory storage capacity per-node of contention event store",
+	64*1024*1024, // 64 MB per node.
+).WithPublic()