Merge branch 'develop' into vault

cfg/rippled-example.cfg

@@ -420,6 +420,7 @@
 #     - r.ripple.com 51235
 #     - sahyadri.isrdc.in 51235
 #     - hubs.xrpkuwait.com 51235
+#     - hub.xrpl-commons.org 51235
 #
 #   Examples:
 #

src/xrpld/app/main/Application.cpp

@@ -1555,10 +1555,10 @@ ApplicationImp::run()
     if (!config_->standalone())
     {
         // VFALCO NOTE This seems unnecessary. If we properly refactor the load
-        //             manager then the deadlock detector can just always be
+        //             manager then the stall detector can just always be
         //             "armed"
         //
-        getLoadManager().activateDeadlockDetector();
+        getLoadManager().activateStallDetector();
     }
 
     {

src/xrpld/app/main/LoadManager.cpp

@@ -32,7 +32,7 @@
 namespace ripple {
 
 LoadManager::LoadManager(Application& app, beast::Journal journal)
-    : app_(app), journal_(journal), deadLock_(), armed_(false)
+    : app_(app), journal_(journal), lastHeartbeat_(), armed_(false)
 {
 }
 
@@ -53,19 +53,19 @@ LoadManager::~LoadManager()
 //------------------------------------------------------------------------------
 
 void
-LoadManager::activateDeadlockDetector()
+LoadManager::activateStallDetector()
 {
     std::lock_guard sl(mutex_);
     armed_ = true;
-    deadLock_ = std::chrono::steady_clock::now();
+    lastHeartbeat_ = std::chrono::steady_clock::now();
 }
 
 void
-LoadManager::resetDeadlockDetector()
+LoadManager::heartbeat()
 {
-    auto const detector_start = std::chrono::steady_clock::now();
+    auto const heartbeat = std::chrono::steady_clock::now();
     std::lock_guard sl(mutex_);
-    deadLock_ = detector_start;
+    lastHeartbeat_ = heartbeat;
 }
 
 //------------------------------------------------------------------------------
@@ -118,63 +118,62 @@ LoadManager::run()
             break;
 
         // Copy out shared data under a lock.  Use copies outside lock.
-        auto const deadLock = deadLock_;
+        auto const lastHeartbeat = lastHeartbeat_;
         auto const armed = armed_;
         sl.unlock();
 
-        // Measure the amount of time we have been deadlocked, in seconds.
+        // Measure the amount of time we have been stalled, in seconds.
         using namespace std::chrono;
-        auto const timeSpentDeadlocked =
+        auto const timeSpentStalled =
-            duration_cast<seconds>(steady_clock::now() - deadLock);
+            duration_cast<seconds>(steady_clock::now() - lastHeartbeat);
 
         constexpr auto reportingIntervalSeconds = 10s;
-        constexpr auto deadlockFatalLogMessageTimeLimit = 90s;
+        constexpr auto stallFatalLogMessageTimeLimit = 90s;
-        constexpr auto deadlockLogicErrorTimeLimit = 600s;
+        constexpr auto stallLogicErrorTimeLimit = 600s;
 
-        if (armed && (timeSpentDeadlocked >= reportingIntervalSeconds))
+        if (armed && (timeSpentStalled >= reportingIntervalSeconds))
         {
-            // Report the deadlocked condition every
+            // Report the stalled condition every reportingIntervalSeconds
-            // reportingIntervalSeconds
+            if ((timeSpentStalled % reportingIntervalSeconds) == 0s)
-            if ((timeSpentDeadlocked % reportingIntervalSeconds) == 0s)
             {
-                if (timeSpentDeadlocked < deadlockFatalLogMessageTimeLimit)
+                if (timeSpentStalled < stallFatalLogMessageTimeLimit)
                 {
                     JLOG(journal_.warn())
-                        << "Server stalled for " << timeSpentDeadlocked.count()
+                        << "Server stalled for " << timeSpentStalled.count()
                         << " seconds.";
+
                     if (app_.getJobQueue().isOverloaded())
                     {
-                        JLOG(journal_.warn()) << app_.getJobQueue().getJson(0);
+                        JLOG(journal_.warn())
+                            << "JobQueue: " << app_.getJobQueue().getJson(0);
                     }
                 }
                 else
                 {
                     JLOG(journal_.fatal())
-                        << "Deadlock detected. Deadlocked time: "
+                        << "Server stalled for " << timeSpentStalled.count()
-                        << timeSpentDeadlocked.count() << "s";
+                        << " seconds.";
                     JLOG(journal_.fatal())
                         << "JobQueue: " << app_.getJobQueue().getJson(0);
                 }
             }
 
-            // If we go over the deadlockTimeLimit spent deadlocked, it
+            // If we go over the stallLogicErrorTimeLimit spent stalled, it
-            // means that the deadlock resolution code has failed, which
+            // means that the stall resolution code has failed, which qualifies
-            // qualifies as undefined behavior.
+            // as a LogicError
-            //
+            if (timeSpentStalled >= stallLogicErrorTimeLimit)
-            if (timeSpentDeadlocked >= deadlockLogicErrorTimeLimit)
             {
                 JLOG(journal_.fatal())
-                    << "LogicError: Deadlock detected. Deadlocked time: "
+                    << "LogicError: Fatal server stall detected. Stalled time: "
-                    << timeSpentDeadlocked.count() << "s";
+                    << timeSpentStalled.count() << "s";
                 JLOG(journal_.fatal())
                     << "JobQueue: " << app_.getJobQueue().getJson(0);
-                LogicError("Deadlock detected");
+                LogicError("Fatal server stall detected");
             }
         }
     }
 
-    bool change;
+    bool change = false;
-
     if (app_.getJobQueue().isOverloaded())
     {
         JLOG(journal_.info()) << "Raising local fee (JQ overload): "

src/xrpld/app/main/LoadManager.h

@@ -58,28 +58,28 @@ public:
     */
     ~LoadManager();
 
-    /** Turn on deadlock detection.
+    /** Turn on stall detection.
 
-        The deadlock detector begins in a disabled state. After this function
+        The stall detector begins in a disabled state. After this function
-        is called, it will report deadlocks using a separate thread whenever
+        is called, it will report stalls using a separate thread whenever
         the reset function is not called at least once per 10 seconds.
 
-        @see resetDeadlockDetector
+        @see resetStallDetector
     */
-    // VFALCO NOTE it seems that the deadlock detector has an "armed" state
+    // VFALCO NOTE it seems that the stall detector has an "armed" state
     //             to prevent it from going off during program startup if
     //             there's a lengthy initialization operation taking place?
     //
     void
-    activateDeadlockDetector();
+    activateStallDetector();
 
-    /** Reset the deadlock detection timer.
+    /** Reset the stall detection timer.
 
-        A dedicated thread monitors the deadlock timer, and if too much
+        A dedicated thread monitors the stall timer, and if too much
         time passes it will produce log warnings.
     */
     void
-    resetDeadlockDetector();
+    heartbeat();
 
     //--------------------------------------------------------------------------
 
@@ -98,12 +98,12 @@ private:
     beast::Journal const journal_;
 
     std::thread thread_;
-    std::mutex mutex_;  // Guards deadLock_, armed_, cv_
+    std::mutex mutex_;  // Guards lastHeartbeat_, armed_, cv_
     std::condition_variable cv_;
     bool stop_ = false;
 
-    std::chrono::steady_clock::time_point
+    // Detect server stalls
-        deadLock_;  // Detect server deadlocks.
+    std::chrono::steady_clock::time_point lastHeartbeat_;
     bool armed_;
 
     friend std::unique_ptr<LoadManager>

src/xrpld/app/misc/NetworkOPs.cpp

@@ -1015,7 +1015,7 @@ NetworkOPsImp::processHeartbeatTimer()
 
         // VFALCO NOTE This is for diagnosing a crash on exit
         LoadManager& mgr(app_.getLoadManager());
-        mgr.resetDeadlockDetector();
+        mgr.heartbeat();
 
         std::size_t const numPeers = app_.overlay().size();
 

src/xrpld/overlay/detail/OverlayImpl.cpp

@@ -503,6 +503,9 @@ OverlayImpl::start()
 
         // Pool of servers operated by @Xrpkuwait - https://xrpkuwait.com
         bootstrapIps.push_back("hubs.xrpkuwait.com 51235");
+
+        // Pool of servers operated by XRPL Commons - https://xrpl-commons.org
+        bootstrapIps.push_back("hub.xrpl-commons.org 51235");
     }
 
     m_resolver.resolve(