Paper/patches/server/0239-Add-Early-Warning-Feature-to-WatchDog.patch

From 0000000000000000000000000000000000000000 Mon Sep 17 00:00:00 2001
From: miclebrick <miclebrick@outlook.com>
Date: Wed, 8 Aug 2018 15:30:52 -0400
Subject: [PATCH] Add Early Warning Feature to WatchDog

Detect when the server has been hung for a long duration, and start printing
thread dumps at an interval until the point of crash.

This will help diagnose what was going on in that time before the crash.

diff --git a/src/main/java/net/minecraft/server/MinecraftServer.java b/src/main/java/net/minecraft/server/MinecraftServer.java
index 274f3a213e60c9b67dbc43c8b53efb676b8bd41e..387c423dfe69a3310f8e7aa3dbb7faafc9501e50 100644
--- a/src/main/java/net/minecraft/server/MinecraftServer.java
+++ b/src/main/java/net/minecraft/server/MinecraftServer.java
@@ -1052,6 +1052,7 @@ public abstract class MinecraftServer extends ReentrantBlockableEventLoop<TickTa
             this.status = this.buildServerStatus();
 
             // Spigot start
+            org.spigotmc.WatchdogThread.hasStarted = true; // Paper
             Arrays.fill( recentTps, 20 );
             long start = System.nanoTime(), curTime, tickSection = start; // Paper - Further improve server tick loop
             lastTick = start - TICK_TIME; // Paper
diff --git a/src/main/java/net/minecraft/server/dedicated/DedicatedServer.java b/src/main/java/net/minecraft/server/dedicated/DedicatedServer.java
index e6826cd0a596f063e8737dcde3c8c6c5b3f71970..1a2607d1b257cea65c82c661a6b3d46c1526a775 100644
--- a/src/main/java/net/minecraft/server/dedicated/DedicatedServer.java
+++ b/src/main/java/net/minecraft/server/dedicated/DedicatedServer.java
@@ -197,6 +197,7 @@ public class DedicatedServer extends MinecraftServer implements ServerInterface
         // Paper start
         paperConfigurations.initializeGlobalConfiguration();
         paperConfigurations.initializeWorldDefaultsConfiguration();
+        org.spigotmc.WatchdogThread.doStart(org.spigotmc.SpigotConfig.timeoutTime, org.spigotmc.SpigotConfig.restartOnCrash);
         io.papermc.paper.command.PaperCommands.registerCommands(this);
         com.destroystokyo.paper.Metrics.PaperMetrics.startMetrics();
         com.destroystokyo.paper.VersionHistoryManager.INSTANCE.getClass(); // load version history now
diff --git a/src/main/java/org/bukkit/craftbukkit/CraftServer.java b/src/main/java/org/bukkit/craftbukkit/CraftServer.java
index f6e7cfc4d86dfdc0ff87642fbc7af330df8eb725..9894d5d10c76469ad6f26100ce3924b65597e738 100644
--- a/src/main/java/org/bukkit/craftbukkit/CraftServer.java
+++ b/src/main/java/org/bukkit/craftbukkit/CraftServer.java
@@ -902,6 +902,7 @@ public final class CraftServer implements Server {
 
     @Override
     public void reload() {
+        org.spigotmc.WatchdogThread.hasStarted = false; // Paper - Disable watchdog early timeout on reload
         this.reloadCount++;
         this.configuration = YamlConfiguration.loadConfiguration(this.getConfigFile());
         this.commandsConfiguration = YamlConfiguration.loadConfiguration(this.getCommandsConfigFile());
@@ -991,6 +992,7 @@ public final class CraftServer implements Server {
         this.enablePlugins(PluginLoadOrder.STARTUP);
         this.enablePlugins(PluginLoadOrder.POSTWORLD);
         this.getPluginManager().callEvent(new ServerLoadEvent(ServerLoadEvent.LoadType.RELOAD));
+        org.spigotmc.WatchdogThread.hasStarted = true; // Paper - Disable watchdog early timeout on reload
     }
 
     @Override
diff --git a/src/main/java/org/spigotmc/SpigotConfig.java b/src/main/java/org/spigotmc/SpigotConfig.java
index ddc5f2d9aa7dda6aff132392927e3d7e3674dbff..3ac48dafe2300ff4cf4591569fec9ce4916503cd 100644
--- a/src/main/java/org/spigotmc/SpigotConfig.java
+++ b/src/main/java/org/spigotmc/SpigotConfig.java
@@ -229,7 +229,7 @@ public class SpigotConfig
         SpigotConfig.restartScript = SpigotConfig.getString( "settings.restart-script", SpigotConfig.restartScript );
         SpigotConfig.restartMessage = SpigotConfig.transform( SpigotConfig.getString( "messages.restart", "Server is restarting" ) );
         SpigotConfig.commands.put( "restart", new RestartCommand( "restart" ) );
-        WatchdogThread.doStart( timeoutTime, restartOnCrash );
+        // WatchdogThread.doStart( timeoutTime, restartOnCrash ); // Paper - moved to after paper config initialization
     }
 
     public static boolean bungee;
diff --git a/src/main/java/org/spigotmc/WatchdogThread.java b/src/main/java/org/spigotmc/WatchdogThread.java
index a9897c494b3dc56d900356d74030359832febbaa..b47d043144c499b1499f6b4be5a16a3f75c9fcb8 100644
--- a/src/main/java/org/spigotmc/WatchdogThread.java
+++ b/src/main/java/org/spigotmc/WatchdogThread.java
@@ -14,6 +14,10 @@ public final class WatchdogThread extends io.papermc.paper.util.TickThread // Pa
     private static WatchdogThread instance;
     private long timeoutTime;
     private boolean restart;
+    private final long earlyWarningEvery; // Paper - Timeout time for just printing a dump but not restarting
+    private final long earlyWarningDelay; // Paper
+    public static volatile boolean hasStarted; // Paper
+    private long lastEarlyWarning; // Paper - Keep track of short dump times to avoid spamming console with short dumps
     private volatile long lastTick;
     private volatile boolean stopping;
 
@@ -22,6 +26,8 @@ public final class WatchdogThread extends io.papermc.paper.util.TickThread // Pa
         super( "Paper Watchdog Thread" );
         this.timeoutTime = timeoutTime;
         this.restart = restart;
+        earlyWarningEvery = Math.min(io.papermc.paper.configuration.GlobalConfiguration.get().watchdog.earlyWarningEvery, timeoutTime); // Paper
+        earlyWarningDelay = Math.min(io.papermc.paper.configuration.GlobalConfiguration.get().watchdog.earlyWarningDelay, timeoutTime); // Paper
     }
 
     private static long monotonicMillis()
@@ -61,9 +67,18 @@ public final class WatchdogThread extends io.papermc.paper.util.TickThread // Pa
         while ( !this.stopping )
         {
             //
-            if ( this.lastTick != 0 && this.timeoutTime > 0 && WatchdogThread.monotonicMillis() > this.lastTick + this.timeoutTime && !Boolean.getBoolean("disable.watchdog")) // Paper - Add property to disable
+            // Paper start
+            Logger log = Bukkit.getServer().getLogger();
+            long currentTime = WatchdogThread.monotonicMillis();
+            if ( this.lastTick != 0 && this.timeoutTime > 0 && currentTime > this.lastTick + this.earlyWarningEvery && !Boolean.getBoolean("disable.watchdog")) // Paper - Add property to disable
             {
-                Logger log = Bukkit.getServer().getLogger();
+                boolean isLongTimeout = currentTime > lastTick + timeoutTime;
+                // Don't spam early warning dumps
+                if ( !isLongTimeout && (earlyWarningEvery <= 0 || !hasStarted || currentTime < lastEarlyWarning + earlyWarningEvery || currentTime < lastTick + earlyWarningDelay)) continue;
+                if ( !isLongTimeout && MinecraftServer.getServer().hasStopped()) continue; // Don't spam early watchdog warnings during shutdown, we'll come back to this...
+                lastEarlyWarning = currentTime;
+                if (isLongTimeout) {
+                // Paper end
                 log.log( Level.SEVERE, "------------------------------" );
                 log.log( Level.SEVERE, "The server has stopped responding! This is (probably) not a Paper bug." ); // Paper
                 log.log( Level.SEVERE, "If you see a plugin in the Server thread dump below, then please report it to that author" );
@@ -93,30 +108,46 @@ public final class WatchdogThread extends io.papermc.paper.util.TickThread // Pa
                     }
                 }
                 // Paper end
+                } else
+                {
+                    log.log(Level.SEVERE, "--- DO NOT REPORT THIS TO PAPER - THIS IS NOT A BUG OR A CRASH  - " + Bukkit.getServer().getVersion() + " ---");
+                    log.log(Level.SEVERE, "The server has not responded for " + (currentTime - lastTick) / 1000 + " seconds! Creating thread dump");
+                }
+                // Paper end - Different message for short timeout
                 log.log( Level.SEVERE, "------------------------------" );
                 log.log( Level.SEVERE, "Server thread dump (Look for plugins here before reporting to Paper!):" ); // Paper
                 io.papermc.paper.chunk.system.scheduling.ChunkTaskScheduler.dumpAllChunkLoadInfo(isLongTimeout); // Paper // Paper - rewrite chunk system
                 WatchdogThread.dumpThread( ManagementFactory.getThreadMXBean().getThreadInfo( MinecraftServer.getServer().serverThread.getId(), Integer.MAX_VALUE ), log );
                 log.log( Level.SEVERE, "------------------------------" );
                 //
+                // Paper start - Only print full dump on long timeouts
+                if ( isLongTimeout )
+                {
                 log.log( Level.SEVERE, "Entire Thread Dump:" );
                 ThreadInfo[] threads = ManagementFactory.getThreadMXBean().dumpAllThreads( true, true );
                 for ( ThreadInfo thread : threads )
                 {
                     WatchdogThread.dumpThread( thread, log );
                 }
+                } else {
+                    log.log(Level.SEVERE, "--- DO NOT REPORT THIS TO PAPER - THIS IS NOT A BUG OR A CRASH ---");
+                }
+
                 log.log( Level.SEVERE, "------------------------------" );
 
+                if ( isLongTimeout )
+                {
                 if ( this.restart && !MinecraftServer.getServer().hasStopped() )
                 {
                     RestartCommand.restart();
                 }
                 break;
+                } // Paper end
             }
 
             try
             {
-                sleep( 10000 );
+                sleep( 1000 ); // Paper - Reduce check time to every second instead of every ten seconds, more consistent and allows for short timeout
             } catch ( InterruptedException ex )
             {
                 interrupt();