SERVER-99077 Add background read size and thread count configuration to live restore API (#31351)

GitOrigin-RevId: df37eb9

Author: Sean04

.github/CODEOWNERS

@@ -1573,6 +1573,9 @@ WORKSPACE.bazel @10gen/devprod-build @svc-auto-approve-bot
 # The following patterns are parsed from ./src/mongo/db/modules/enterprise/jstests/live_import/OWNERS.yml
 /src/mongo/db/modules/enterprise/jstests/live_import/**/* @10gen/server-backup-restore @svc-auto-approve-bot
 
+# The following patterns are parsed from ./src/mongo/db/modules/enterprise/jstests/live_restore/OWNERS.yml
+/src/mongo/db/modules/enterprise/jstests/live_restore/**/* @10gen/server-backup-restore @svc-auto-approve-bot
+
 # The following patterns are parsed from ./src/mongo/db/modules/enterprise/jstests/no_passthrough/OWNERS.yml
 /src/mongo/db/modules/enterprise/jstests/no_passthrough/**/* @10gen/query-integration-fle @10gen/server-security @svc-auto-approve-bot
 

buildscripts/resmokeconfig/suites/no_passthrough.yml

@@ -11,6 +11,7 @@ selector:
     - jstests/noPassthrough/**/*.js
     - src/mongo/db/modules/*/jstests/hot_backups/**/*.js
     - src/mongo/db/modules/*/jstests/live_import/*.js
+    - src/mongo/db/modules/*/jstests/live_restore/*.js
     - src/mongo/db/modules/*/jstests/no_passthrough/*.js
 
     # Self-tests for the Concurrency testing framework are run as part of this test suite.
@@ -26,6 +27,7 @@ executor:
       - jstests/noPassthrough/backup*.js
       - jstests/noPassthrough/oplog_writes_only_permitted_on_standalone.js
       - jstests/noPassthrough/wt_unclean_shutdown.js
+      - src/mongo/db/modules/*/jstests/live_restore/*.js
       - src/mongo/db/modules/enterprise/jstests/hot_backups/**/*.js
   config:
     shell_options:

src/mongo/db/storage/wiredtiger/wiredtiger_global_options.cpp

@@ -75,4 +75,21 @@ Status WiredTigerGlobalOptions::validateWiredTigerCompressor(const std::string&
     return Status::OK();
 }
 
+Status WiredTigerGlobalOptions::validateWiredTigerLiveRestoreReadSizeMB(const int value) {
+    if (value < 1) {
+        return {ErrorCodes::BadValue,
+                "Live restore read size must be greater than or equal to 1MB."};
+    }
+
+    if (value > 16) {
+        return {ErrorCodes::BadValue, "Live restore read size must be less than or equal to 16MB."};
+    }
+
+    if ((value & (value - 1)) != 0) {
+        return {ErrorCodes::BadValue, "Live restore read size must be a power of two."};
+    }
+
+    return Status::OK();
+}
+
 }  // namespace mongo

src/mongo/db/storage/wiredtiger/wiredtiger_global_options.h

@@ -48,6 +48,8 @@ class WiredTigerGlobalOptions {
           zstdCompressorLevel(0),
           directoryForIndexes(false),
           maxCacheOverflowFileSizeGBDeprecated(0),
+          liveRestoreThreads(0),
+          liveRestoreReadSizeMB(0),
           useCollectionPrefixCompression(false),
           useIndexPrefixCompression(false){};
 
@@ -61,6 +63,8 @@ class WiredTigerGlobalOptions {
     double maxCacheOverflowFileSizeGBDeprecated;
     std::string engineConfig;
     std::string liveRestoreSource;
+    int liveRestoreThreads;
+    double liveRestoreReadSizeMB;
 
     std::string collectionBlockCompressor;
     bool useCollectionPrefixCompression;
@@ -69,6 +73,8 @@ class WiredTigerGlobalOptions {
     std::string indexConfig;
 
     static Status validateWiredTigerCompressor(const std::string&);
+    static Status validateWiredTigerLiveRestoreReadSizeMB(int);
+
 
     /**
      * Returns current history file size limit in MB.

src/mongo/db/storage/wiredtiger/wiredtiger_global_options.idl

@@ -102,6 +102,25 @@ configs:
         arg_vartype: String
         cpp_varname: 'wiredTigerGlobalOptions.liveRestoreSource'
         short_name: wiredTigerLiveRestoreSource
+    "storage.wiredTiger.engineConfig.wiredTigerLiveRestoreThreads":
+        description: 'Number of live restore background threads.'
+        arg_vartype: Int
+        cpp_varname: 'wiredTigerGlobalOptions.liveRestoreThreads'
+        short_name: wiredTigerLiveRestoreThreads
+        validator:
+            gte: 0
+            lte: 12
+        default: 8
+    "storage.wiredTiger.engineConfig.wiredTigerLiveRestoreReadSizeMB":
+        description: >-
+            'The read size for data migration, in MB, must be a power of two.
+            This setting is a best effort. It does not force every read to be this size.'
+        arg_vartype: Int
+        cpp_varname: 'wiredTigerGlobalOptions.liveRestoreReadSizeMB'
+        short_name: wiredTigerLiveRestoreReadSizeMB
+        validator:
+            callback: WiredTigerGlobalOptions::validateWiredTigerLiveRestoreReadSizeMB
+        default: 1
 
     # WiredTiger collection options
     "storage.wiredTiger.collectionConfig.blockCompressor":

src/mongo/db/storage/wiredtiger/wiredtiger_kv_engine.cpp

@@ -483,7 +483,9 @@ WiredTigerKVEngine::WiredTigerKVEngine(const std::string& canonicalName,
 
     if (!wiredTigerGlobalOptions.liveRestoreSource.empty() && !_ephemeral) {
         ss << "live_restore=(enabled=true,path=\"" << wiredTigerGlobalOptions.liveRestoreSource
-           << "\"),";
+           << "\",threads_max=" << wiredTigerGlobalOptions.liveRestoreThreads
+           << ",read_size=" << wiredTigerGlobalOptions.liveRestoreReadSizeMB << "MB"
+           << "),";
     }
 
     ss << WiredTigerCustomizationHooks::get(getGlobalServiceContext())

src/mongo/db/storage/wiredtiger/wiredtiger_util.cpp

@@ -530,9 +530,14 @@ logv2::LogComponent getWTLogComponent(const BSONObj& obj) {
             return logv2::LogComponent::kWiredTigerCompact;
         case WT_VERB_EVICTION:
             return logv2::LogComponent::kWiredTigerEviction;
+        case WT_VERB_FILEOPS:
+            return logv2::LogComponent::kWiredTigerFileOps;
         case WT_VERB_HS:
         case WT_VERB_HS_ACTIVITY:
             return logv2::LogComponent::kWiredTigerHS;
+        case WT_VERB_LIVE_RESTORE:
+        case WT_VERB_LIVE_RESTORE_PROGRESS:
+            return logv2::LogComponent::kWiredTigerLiveRestore;
         case WT_VERB_RECOVERY:
         case WT_VERB_RECOVERY_PROGRESS:
             return logv2::LogComponent::kWiredTigerRecovery;
@@ -1165,7 +1170,9 @@ std::string WiredTigerUtil::generateWTVerboseConfiguration() {
         {logv2::LogComponent::kWiredTigerCheckpoint, "checkpoint"},
         {logv2::LogComponent::kWiredTigerCompact, "compact"},
         {logv2::LogComponent::kWiredTigerEviction, "eviction"},
+        {logv2::LogComponent::kWiredTigerFileOps, "fileops"},
         {logv2::LogComponent::kWiredTigerHS, "history_store"},
+        {logv2::LogComponent::kWiredTigerLiveRestore, "live_restore"},
         {logv2::LogComponent::kWiredTigerRecovery, "recovery"},
         {logv2::LogComponent::kWiredTigerRTS, "rts"},
         {logv2::LogComponent::kWiredTigerSalvage, "salvage"},
@@ -1182,7 +1189,7 @@ std::string WiredTigerUtil::generateWTVerboseConfiguration() {
     cfg << "verbose=[";
 
     // Enable WiredTiger progress messages.
-    cfg << "recovery_progress:1,checkpoint_progress:1,compact_progress:1";
+    cfg << "recovery_progress:1,checkpoint_progress:1,compact_progress:1,live_restore_progress:1";
 
     // Process each LOGV2 WiredTiger component and set the desired verbosity level.
     for (const auto& [component, componentStr] : *wtVerboseComponents) {

src/mongo/logv2/log_component.h

@@ -85,7 +85,9 @@ namespace mongo::logv2 {
     X(kWiredTigerCheckpoint,   , "wtCheckpoint"          , "WTCHKPT" , kWiredTiger) \
     X(kWiredTigerCompact,      , "wtCompact"             , "WTCMPCT" , kWiredTiger) \
     X(kWiredTigerEviction,     , "wtEviction"            , "WTEVICT" , kWiredTiger) \
+    X(kWiredTigerFileOps,      , "wtFileOps"             , "WTFILEOPS", kWiredTiger) \
     X(kWiredTigerHS,           , "wtHS"                  , "WTHS"    , kWiredTiger) \
+    X(kWiredTigerLiveRestore,  , "wtLiveRestore"         , "WTLR"    , kWiredTiger) \
     X(kWiredTigerRecovery,     , "wtRecovery"            , "WTRECOV" , kWiredTiger) \
     X(kWiredTigerRTS,          , "wtRTS"                 , "WTRTS"   , kWiredTiger) \
     X(kWiredTigerSalvage,      , "wtSalvage"             , "WTSLVG"  , kWiredTiger) \