core
/
testserver
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity
testserver/Spigot-Server-Patches/0437-Improved-Watchdog-Supp...

577 lines
32 KiB
Diff
Raw Normal View History

Rebuild all patches using the new rebuild pattern
2020-05-06 09:48:49 +00:00
From 0000000000000000000000000000000000000000 Mon Sep 17 00:00:00 2001
Forced Watchdog Crash support and Improve Async Shutdown If the request to shut down the server is received while we are in a watchdog hang, immediately treat it as a crash and begin the shutdown process. Shutdown process is now improved to also shutdown cleanly when not using restart scripts either. If a server is deadlocked, a server owner can send SIGHUP (or any other signal the JVM understands to shut down as it currently does) and the watchdog will no longer need to wait until the full timeout, allowing you to trigger a close process and try to shut the server down gracefully, saving player and world data. Previously there was no way to trigger this outside of waiting for a full watchdog timeout, which may be set to a really long time... Additionally, fix everything to do with shutting the server down asynchronously. Previously, nearly everything about the process was fragile and unsafe. Main might not have actually been frozen, and might still be manipulating state. Or, some reuest might ask main to do something in the shutdown but main is dead. Or worse, other things might start closing down items such as the Console or Thread Pool before we are fully shutdown. This change tries to resolve all of these issues by moving everything into the stop method and guaranteeing only one thread is stopping the server. We then issue Thread Death to the main thread of another thread initiates the stop process. We have to ensure Thread Death propagates correctly though to stop main completely. This is to ensure that if main isn't truely stuck, it's not manipulating state we are trying to save.
2020-04-16 08:53:50 +00:00
From: Aikar <aikar@aikar.co>
Date: Sun, 12 Apr 2020 15:50:48 -0400
Force Plugins that use delayed tasks for init back in their place Will now run those tasks just before we print "Done" so that startup time is appropriately accounted for a plugin, as well as will no longer trip the watchdog on startup. Any plugin that tries to bypass this is just going to then trip watchdog on Spigot too, so don't you dare. Stop trying to cheat the delay your plugin added to startup time. This isn't a behavior change because the first thing the tick does.... was run these tasks.... So it's just moving it slightly a few lines to be before a watchdog tick and to account for it in "Done" time. Fixes #3294
2020-05-07 00:18:31 +00:00
Subject: [PATCH] Improved Watchdog Support
Forced Watchdog Crash support and Improve Async Shutdown
Forced Watchdog Crash support and Improve Async Shutdown If the request to shut down the server is received while we are in a watchdog hang, immediately treat it as a crash and begin the shutdown process. Shutdown process is now improved to also shutdown cleanly when not using restart scripts either. If a server is deadlocked, a server owner can send SIGHUP (or any other signal the JVM understands to shut down as it currently does) and the watchdog will no longer need to wait until the full timeout, allowing you to trigger a close process and try to shut the server down gracefully, saving player and world data. Previously there was no way to trigger this outside of waiting for a full watchdog timeout, which may be set to a really long time... Additionally, fix everything to do with shutting the server down asynchronously. Previously, nearly everything about the process was fragile and unsafe. Main might not have actually been frozen, and might still be manipulating state. Or, some reuest might ask main to do something in the shutdown but main is dead. Or worse, other things might start closing down items such as the Console or Thread Pool before we are fully shutdown. This change tries to resolve all of these issues by moving everything into the stop method and guaranteeing only one thread is stopping the server. We then issue Thread Death to the main thread of another thread initiates the stop process. We have to ensure Thread Death propagates correctly though to stop main completely. This is to ensure that if main isn't truely stuck, it's not manipulating state we are trying to save.
2020-04-16 08:53:50 +00:00
If the request to shut down the server is received while we are in
a watchdog hang, immediately treat it as a crash and begin the shutdown
process. Shutdown process is now improved to also shutdown cleanly when
not using restart scripts either.
If a server is deadlocked, a server owner can send SIGUP (or any other signal
the JVM understands to shut down as it currently does) and the watchdog
will no longer need to wait until the full timeout, allowing you to trigger
a close process and try to shut the server down gracefully, saving player and
world data.
Previously there was no way to trigger this outside of waiting for a full watchdog
timeout, which may be set to a really long time...
Additionally, fix everything to do with shutting the server down asynchronously.
Previously, nearly everything about the process was fragile and unsafe. Main might
not have actually been frozen, and might still be manipulating state.
Or, some reuest might ask main to do something in the shutdown but main is dead.
Or worse, other things might start closing down items such as the Console or Thread Pool
before we are fully shutdown.
This change tries to resolve all of these issues by moving everything into the stop
method and guaranteeing only one thread is stopping the server.
We then issue Thread Death to the main thread of another thread initiates the stop process.
We have to ensure Thread Death propagates correctly though to stop main completely.
This is to ensure that if main isn't truely stuck, it's not manipulating state we are trying to save.
Force Plugins that use delayed tasks for init back in their place Will now run those tasks just before we print "Done" so that startup time is appropriately accounted for a plugin, as well as will no longer trip the watchdog on startup. Any plugin that tries to bypass this is just going to then trip watchdog on Spigot too, so don't you dare. Stop trying to cheat the delay your plugin added to startup time. This isn't a behavior change because the first thing the tick does.... was run these tasks.... So it's just moving it slightly a few lines to be before a watchdog tick and to account for it in "Done" time. Fixes #3294
2020-05-07 00:18:31 +00:00
This also moves all plugins who register "delayed init" tasks to occur just before "Done" so they
are properly accounted for and wont trip watchdog on init.
Preload important classes such as Logger and JLine This is for 2 reasons: 1) Ensuring our log4j is mostly loaded at OUR version. I've seen stack traces with line numbers that do not match our version. This means that some plugin has shaded in log4j and their loaded version is mixing with ours.... So by at least trying to load a bunch of log4j classes before we load plugins, we can be more sure mixed versions are not loading. 2) If the jar file is replaced while the server is runnimg class not found errors galore This will preloaod a bunch of classes commonly seen to error during shutdown due to this. The goal here is to help let the server shutdown gracefully as possible. Some plugins will still blow up here if they access a class that hadn't been loaded yet, but goal is to at least stop freezing the shutdown process as it does with JLine and Log4j errors requiring an external kill. Ideally you should not replace jars while the server is running, but it is something that happens in development for testing. Updated test server to do a copy though to avoid this happening in Paper development.
2020-05-24 17:09:02 +00:00
diff --git a/src/main/java/com/destroystokyo/paper/Metrics.java b/src/main/java/com/destroystokyo/paper/Metrics.java
Add legacy plugin count to metrics
2020-05-11 20:18:58 +00:00
index e4923ca371012188793b284cb0ed4280c31f50b6..03c95b30d8a0557482d39468c058e7966114f824 100644
Preload important classes such as Logger and JLine This is for 2 reasons: 1) Ensuring our log4j is mostly loaded at OUR version. I've seen stack traces with line numbers that do not match our version. This means that some plugin has shaded in log4j and their loaded version is mixing with ours.... So by at least trying to load a bunch of log4j classes before we load plugins, we can be more sure mixed versions are not loading. 2) If the jar file is replaced while the server is runnimg class not found errors galore This will preloaod a bunch of classes commonly seen to error during shutdown due to this. The goal here is to help let the server shutdown gracefully as possible. Some plugins will still blow up here if they access a class that hadn't been loaded yet, but goal is to at least stop freezing the shutdown process as it does with JLine and Log4j errors requiring an external kill. Ideally you should not replace jars while the server is running, but it is something that happens in development for testing. Updated test server to do a copy though to avoid this happening in Paper development.
2020-05-24 17:09:02 +00:00
--- a/src/main/java/com/destroystokyo/paper/Metrics.java
+++ b/src/main/java/com/destroystokyo/paper/Metrics.java
Add legacy plugin count to metrics
2020-05-11 20:18:58 +00:00
@@ -90,6 +90,9 @@ public class Metrics {
Preload important classes such as Logger and JLine This is for 2 reasons: 1) Ensuring our log4j is mostly loaded at OUR version. I've seen stack traces with line numbers that do not match our version. This means that some plugin has shaded in log4j and their loaded version is mixing with ours.... So by at least trying to load a bunch of log4j classes before we load plugins, we can be more sure mixed versions are not loading. 2) If the jar file is replaced while the server is runnimg class not found errors galore This will preloaod a bunch of classes commonly seen to error during shutdown due to this. The goal here is to help let the server shutdown gracefully as possible. Some plugins will still blow up here if they access a class that hadn't been loaded yet, but goal is to at least stop freezing the shutdown process as it does with JLine and Log4j errors requiring an external kill. Ideally you should not replace jars while the server is running, but it is something that happens in development for testing. Updated test server to do a copy though to avoid this happening in Paper development.
2020-05-24 17:09:02 +00:00
timer.scheduleAtFixedRate(new TimerTask() {
@Override
public void run() {
+ if (MinecraftServer.getServer().hasStopped()) {
+ return;
+ }
submitData();
}
}, 1000 * 60 * 5, 1000 * 60 * 30);
Forced Watchdog Crash support and Improve Async Shutdown If the request to shut down the server is received while we are in a watchdog hang, immediately treat it as a crash and begin the shutdown process. Shutdown process is now improved to also shutdown cleanly when not using restart scripts either. If a server is deadlocked, a server owner can send SIGHUP (or any other signal the JVM understands to shut down as it currently does) and the watchdog will no longer need to wait until the full timeout, allowing you to trigger a close process and try to shut the server down gracefully, saving player and world data. Previously there was no way to trigger this outside of waiting for a full watchdog timeout, which may be set to a really long time... Additionally, fix everything to do with shutting the server down asynchronously. Previously, nearly everything about the process was fragile and unsafe. Main might not have actually been frozen, and might still be manipulating state. Or, some reuest might ask main to do something in the shutdown but main is dead. Or worse, other things might start closing down items such as the Console or Thread Pool before we are fully shutdown. This change tries to resolve all of these issues by moving everything into the stop method and guaranteeing only one thread is stopping the server. We then issue Thread Death to the main thread of another thread initiates the stop process. We have to ensure Thread Death propagates correctly though to stop main completely. This is to ensure that if main isn't truely stuck, it's not manipulating state we are trying to save.
2020-04-16 08:53:50 +00:00
diff --git a/src/main/java/net/minecraft/server/CrashReport.java b/src/main/java/net/minecraft/server/CrashReport.java
even even even even even even more work
2020-06-26 01:53:21 +00:00
index 95e6a6de7ccfc4445d0ac19c5f874c0d533b1712..cc6e6f245ee5e73bd570cf42381bf55ee0b364d3 100644
Forced Watchdog Crash support and Improve Async Shutdown If the request to shut down the server is received while we are in a watchdog hang, immediately treat it as a crash and begin the shutdown process. Shutdown process is now improved to also shutdown cleanly when not using restart scripts either. If a server is deadlocked, a server owner can send SIGHUP (or any other signal the JVM understands to shut down as it currently does) and the watchdog will no longer need to wait until the full timeout, allowing you to trigger a close process and try to shut the server down gracefully, saving player and world data. Previously there was no way to trigger this outside of waiting for a full watchdog timeout, which may be set to a really long time... Additionally, fix everything to do with shutting the server down asynchronously. Previously, nearly everything about the process was fragile and unsafe. Main might not have actually been frozen, and might still be manipulating state. Or, some reuest might ask main to do something in the shutdown but main is dead. Or worse, other things might start closing down items such as the Console or Thread Pool before we are fully shutdown. This change tries to resolve all of these issues by moving everything into the stop method and guaranteeing only one thread is stopping the server. We then issue Thread Death to the main thread of another thread initiates the stop process. We have to ensure Thread Death propagates correctly though to stop main completely. This is to ensure that if main isn't truely stuck, it's not manipulating state we are trying to save.
2020-04-16 08:53:50 +00:00
--- a/src/main/java/net/minecraft/server/CrashReport.java
+++ b/src/main/java/net/minecraft/server/CrashReport.java
@@ -257,6 +257,7 @@ public class CrashReport {
}
public static CrashReport a(Throwable throwable, String s) {
+ if (throwable instanceof ThreadDeath) com.destroystokyo.paper.util.SneakyThrow.sneaky(throwable); // Paper
while (throwable instanceof CompletionException && throwable.getCause() != null) {
throwable = throwable.getCause();
}
diff --git a/src/main/java/net/minecraft/server/DedicatedServer.java b/src/main/java/net/minecraft/server/DedicatedServer.java
Updated Upstream (Bukkit/CraftBukkit/Spigot) (#4728) Upstream has released updates that appears to apply and compile correctly. This update has not been tested by PaperMC and as with ANY update, please do your own testing Bukkit Changes: 30885166 Update to Minecraft 1.16.4 CraftBukkit Changes: 3af81c71 Update to Minecraft 1.16.4 Spigot Changes: f011ca24 Update to Minecraft 1.16.4 Co-authored-by: Mariell Hoversholm <proximyst@proximyst.com>
2020-11-03 02:22:15 +00:00
index a4db2527ff2defcfd25f08dd24722a37794a4a7e..f5792b999ce42acb13ae9a62ceb2ddec39abe209 100644
Forced Watchdog Crash support and Improve Async Shutdown If the request to shut down the server is received while we are in a watchdog hang, immediately treat it as a crash and begin the shutdown process. Shutdown process is now improved to also shutdown cleanly when not using restart scripts either. If a server is deadlocked, a server owner can send SIGHUP (or any other signal the JVM understands to shut down as it currently does) and the watchdog will no longer need to wait until the full timeout, allowing you to trigger a close process and try to shut the server down gracefully, saving player and world data. Previously there was no way to trigger this outside of waiting for a full watchdog timeout, which may be set to a really long time... Additionally, fix everything to do with shutting the server down asynchronously. Previously, nearly everything about the process was fragile and unsafe. Main might not have actually been frozen, and might still be manipulating state. Or, some reuest might ask main to do something in the shutdown but main is dead. Or worse, other things might start closing down items such as the Console or Thread Pool before we are fully shutdown. This change tries to resolve all of these issues by moving everything into the stop method and guaranteeing only one thread is stopping the server. We then issue Thread Death to the main thread of another thread initiates the stop process. We have to ensure Thread Death propagates correctly though to stop main completely. This is to ensure that if main isn't truely stuck, it's not manipulating state we are trying to save.
2020-04-16 08:53:50 +00:00
--- a/src/main/java/net/minecraft/server/DedicatedServer.java
+++ b/src/main/java/net/minecraft/server/DedicatedServer.java
Updated Upstream (Bukkit/CraftBukkit/Spigot) (#4728) Upstream has released updates that appears to apply and compile correctly. This update has not been tested by PaperMC and as with ANY update, please do your own testing Bukkit Changes: 30885166 Update to Minecraft 1.16.4 CraftBukkit Changes: 3af81c71 Update to Minecraft 1.16.4 Spigot Changes: f011ca24 Update to Minecraft 1.16.4 Co-authored-by: Mariell Hoversholm <proximyst@proximyst.com>
2020-11-03 02:22:15 +00:00
@@ -240,7 +240,7 @@ public class DedicatedServer extends MinecraftServer implements IMinecraftServer
even even even even even even more work
2020-06-26 01:53:21 +00:00
long j = SystemUtils.getMonotonicNanos() - i;
String s = String.format(Locale.ROOT, "%.3fs", (double) j / 1.0E9D);
Move delayed init down later in tick, improve accuracy of startup time Now tracks the full startup time for "Done" message at end, as apparently Vanillas was done in a place that skipped tracking a lot of code too. This fixes an issue with ViaVersion
2020-05-07 01:24:43 +00:00
even even even even even even more work
2020-06-26 01:53:21 +00:00
- DedicatedServer.LOGGER.info("Done ({})! For help, type \"help\"", s);
+ //DedicatedServer.LOGGER.info("Done ({})! For help, type \"help\"", s); // Paper moved to after init
Move delayed init down later in tick, improve accuracy of startup time Now tracks the full startup time for "Done" message at end, as apparently Vanillas was done in a place that skipped tracking a lot of code too. This fixes an issue with ViaVersion
2020-05-07 01:24:43 +00:00
if (dedicatedserverproperties.announcePlayerAchievements != null) {
((GameRules.GameRuleBoolean) this.getGameRules().get(GameRules.ANNOUNCE_ADVANCEMENTS)).a(dedicatedserverproperties.announcePlayerAchievements, (MinecraftServer) this);
}
Updated Upstream (Bukkit/CraftBukkit/Spigot) (#4728) Upstream has released updates that appears to apply and compile correctly. This update has not been tested by PaperMC and as with ANY update, please do your own testing Bukkit Changes: 30885166 Update to Minecraft 1.16.4 CraftBukkit Changes: 3af81c71 Update to Minecraft 1.16.4 Spigot Changes: f011ca24 Update to Minecraft 1.16.4 Co-authored-by: Mariell Hoversholm <proximyst@proximyst.com>
2020-11-03 02:22:15 +00:00
@@ -368,6 +368,7 @@ public class DedicatedServer extends MinecraftServer implements IMinecraftServer
Fix issues with SIGTERM not shutting down gracefully - Fixes #3062 We had 2 issues. 1) Log4J2 Shutdown hook seemed to be causing issues as it shutdown logger while we still needed it 2) ServerShutdownThread needs to stay alive until server is shutdown to keep jvm open. It appears SIGINT is handled differently than SIGTERM, as SIGINT worked correctly. But this will make both methods work.
2020-06-11 03:40:28 +00:00
//this.remoteStatusListener.b(); // Paper - don't wait for remote connections
}
+ hasFullyShutdown = true; // Paper
System.exit(0); // CraftBukkit
}
Updated Upstream (Bukkit/CraftBukkit/Spigot) (#4728) Upstream has released updates that appears to apply and compile correctly. This update has not been tested by PaperMC and as with ANY update, please do your own testing Bukkit Changes: 30885166 Update to Minecraft 1.16.4 CraftBukkit Changes: 3af81c71 Update to Minecraft 1.16.4 Spigot Changes: f011ca24 Update to Minecraft 1.16.4 Co-authored-by: Mariell Hoversholm <proximyst@proximyst.com>
2020-11-03 02:22:15 +00:00
@@ -701,7 +702,7 @@ public class DedicatedServer extends MinecraftServer implements IMinecraftServer
Forced Watchdog Crash support and Improve Async Shutdown If the request to shut down the server is received while we are in a watchdog hang, immediately treat it as a crash and begin the shutdown process. Shutdown process is now improved to also shutdown cleanly when not using restart scripts either. If a server is deadlocked, a server owner can send SIGHUP (or any other signal the JVM understands to shut down as it currently does) and the watchdog will no longer need to wait until the full timeout, allowing you to trigger a close process and try to shut the server down gracefully, saving player and world data. Previously there was no way to trigger this outside of waiting for a full watchdog timeout, which may be set to a really long time... Additionally, fix everything to do with shutting the server down asynchronously. Previously, nearly everything about the process was fragile and unsafe. Main might not have actually been frozen, and might still be manipulating state. Or, some reuest might ask main to do something in the shutdown but main is dead. Or worse, other things might start closing down items such as the Console or Thread Pool before we are fully shutdown. This change tries to resolve all of these issues by moving everything into the stop method and guaranteeing only one thread is stopping the server. We then issue Thread Death to the main thread of another thread initiates the stop process. We have to ensure Thread Death propagates correctly though to stop main completely. This is to ensure that if main isn't truely stuck, it's not manipulating state we are trying to save.
2020-04-16 08:53:50 +00:00
@Override
public void stop() {
super.stop();
even even even even even even more work
2020-06-26 01:53:21 +00:00
- SystemUtils.h();
+ //SystemUtils.h(); // Paper - moved into super
Forced Watchdog Crash support and Improve Async Shutdown If the request to shut down the server is received while we are in a watchdog hang, immediately treat it as a crash and begin the shutdown process. Shutdown process is now improved to also shutdown cleanly when not using restart scripts either. If a server is deadlocked, a server owner can send SIGHUP (or any other signal the JVM understands to shut down as it currently does) and the watchdog will no longer need to wait until the full timeout, allowing you to trigger a close process and try to shut the server down gracefully, saving player and world data. Previously there was no way to trigger this outside of waiting for a full watchdog timeout, which may be set to a really long time... Additionally, fix everything to do with shutting the server down asynchronously. Previously, nearly everything about the process was fragile and unsafe. Main might not have actually been frozen, and might still be manipulating state. Or, some reuest might ask main to do something in the shutdown but main is dead. Or worse, other things might start closing down items such as the Console or Thread Pool before we are fully shutdown. This change tries to resolve all of these issues by moving everything into the stop method and guaranteeing only one thread is stopping the server. We then issue Thread Death to the main thread of another thread initiates the stop process. We have to ensure Thread Death propagates correctly though to stop main completely. This is to ensure that if main isn't truely stuck, it's not manipulating state we are trying to save.
2020-04-16 08:53:50 +00:00
}
@Override
Improvements to watchdog changes catch a thread death, fix invalid watchdog timeouts <= 0 seconds will now change to 5 minutes.
2020-04-24 04:27:28 +00:00
diff --git a/src/main/java/net/minecraft/server/IAsyncTaskHandler.java b/src/main/java/net/minecraft/server/IAsyncTaskHandler.java
Updated Upstream (Bukkit/CraftBukkit/Spigot) (#4728) Upstream has released updates that appears to apply and compile correctly. This update has not been tested by PaperMC and as with ANY update, please do your own testing Bukkit Changes: 30885166 Update to Minecraft 1.16.4 CraftBukkit Changes: 3af81c71 Update to Minecraft 1.16.4 Spigot Changes: f011ca24 Update to Minecraft 1.16.4 Co-authored-by: Mariell Hoversholm <proximyst@proximyst.com>
2020-11-03 02:22:15 +00:00
index bc15da4640a4a6107c9c186a01ce76df87511b41..27db247aa40e0516302c74b9bf00c631a8607af5 100644
Improvements to watchdog changes catch a thread death, fix invalid watchdog timeouts <= 0 seconds will now change to 5 minutes.
2020-04-24 04:27:28 +00:00
--- a/src/main/java/net/minecraft/server/IAsyncTaskHandler.java
+++ b/src/main/java/net/minecraft/server/IAsyncTaskHandler.java
@@ -135,6 +135,7 @@ public abstract class IAsyncTaskHandler<R extends Runnable> implements Mailbox<R
try {
r0.run();
} catch (Exception exception) {
+ if (exception.getCause() instanceof ThreadDeath) throw exception; // Paper
Updated Upstream (Bukkit/CraftBukkit/Spigot) (#4728) Upstream has released updates that appears to apply and compile correctly. This update has not been tested by PaperMC and as with ANY update, please do your own testing Bukkit Changes: 30885166 Update to Minecraft 1.16.4 CraftBukkit Changes: 3af81c71 Update to Minecraft 1.16.4 Spigot Changes: f011ca24 Update to Minecraft 1.16.4 Co-authored-by: Mariell Hoversholm <proximyst@proximyst.com>
2020-11-03 02:22:15 +00:00
IAsyncTaskHandler.LOGGER.fatal("Error executing task on {}", this.bj(), exception);
Improvements to watchdog changes catch a thread death, fix invalid watchdog timeouts <= 0 seconds will now change to 5 minutes.
2020-04-24 04:27:28 +00:00
}
Forced Watchdog Crash support and Improve Async Shutdown If the request to shut down the server is received while we are in a watchdog hang, immediately treat it as a crash and begin the shutdown process. Shutdown process is now improved to also shutdown cleanly when not using restart scripts either. If a server is deadlocked, a server owner can send SIGHUP (or any other signal the JVM understands to shut down as it currently does) and the watchdog will no longer need to wait until the full timeout, allowing you to trigger a close process and try to shut the server down gracefully, saving player and world data. Previously there was no way to trigger this outside of waiting for a full watchdog timeout, which may be set to a really long time... Additionally, fix everything to do with shutting the server down asynchronously. Previously, nearly everything about the process was fragile and unsafe. Main might not have actually been frozen, and might still be manipulating state. Or, some reuest might ask main to do something in the shutdown but main is dead. Or worse, other things might start closing down items such as the Console or Thread Pool before we are fully shutdown. This change tries to resolve all of these issues by moving everything into the stop method and guaranteeing only one thread is stopping the server. We then issue Thread Death to the main thread of another thread initiates the stop process. We have to ensure Thread Death propagates correctly though to stop main completely. This is to ensure that if main isn't truely stuck, it's not manipulating state we are trying to save.
2020-04-16 08:53:50 +00:00
diff --git a/src/main/java/net/minecraft/server/MinecraftServer.java b/src/main/java/net/minecraft/server/MinecraftServer.java
Fix keep-spawn-loaded stopping force-loaded chunks from loading Also, cleanup the patch a little bit
2020-12-01 13:23:44 +00:00
index 0c469ad1fd477fdb6181c657b7b9e193c790246d..fe88179037f1a56cec3543910cfab67b0e2ebbe4 100644
Forced Watchdog Crash support and Improve Async Shutdown If the request to shut down the server is received while we are in a watchdog hang, immediately treat it as a crash and begin the shutdown process. Shutdown process is now improved to also shutdown cleanly when not using restart scripts either. If a server is deadlocked, a server owner can send SIGHUP (or any other signal the JVM understands to shut down as it currently does) and the watchdog will no longer need to wait until the full timeout, allowing you to trigger a close process and try to shut the server down gracefully, saving player and world data. Previously there was no way to trigger this outside of waiting for a full watchdog timeout, which may be set to a really long time... Additionally, fix everything to do with shutting the server down asynchronously. Previously, nearly everything about the process was fragile and unsafe. Main might not have actually been frozen, and might still be manipulating state. Or, some reuest might ask main to do something in the shutdown but main is dead. Or worse, other things might start closing down items such as the Console or Thread Pool before we are fully shutdown. This change tries to resolve all of these issues by moving everything into the stop method and guaranteeing only one thread is stopping the server. We then issue Thread Death to the main thread of another thread initiates the stop process. We have to ensure Thread Death propagates correctly though to stop main completely. This is to ensure that if main isn't truely stuck, it's not manipulating state we are trying to save.
2020-04-16 08:53:50 +00:00
--- a/src/main/java/net/minecraft/server/MinecraftServer.java
+++ b/src/main/java/net/minecraft/server/MinecraftServer.java
[WIP|1.16] Misc fixes (#3627)
2020-06-26 16:20:03 +00:00
@@ -159,7 +159,7 @@ public abstract class MinecraftServer extends IAsyncTaskHandlerReentrant<TickTas
public int autosavePeriod;
Forced Watchdog Crash support and Improve Async Shutdown If the request to shut down the server is received while we are in a watchdog hang, immediately treat it as a crash and begin the shutdown process. Shutdown process is now improved to also shutdown cleanly when not using restart scripts either. If a server is deadlocked, a server owner can send SIGHUP (or any other signal the JVM understands to shut down as it currently does) and the watchdog will no longer need to wait until the full timeout, allowing you to trigger a close process and try to shut the server down gracefully, saving player and world data. Previously there was no way to trigger this outside of waiting for a full watchdog timeout, which may be set to a really long time... Additionally, fix everything to do with shutting the server down asynchronously. Previously, nearly everything about the process was fragile and unsafe. Main might not have actually been frozen, and might still be manipulating state. Or, some reuest might ask main to do something in the shutdown but main is dead. Or worse, other things might start closing down items such as the Console or Thread Pool before we are fully shutdown. This change tries to resolve all of these issues by moving everything into the stop method and guaranteeing only one thread is stopping the server. We then issue Thread Death to the main thread of another thread initiates the stop process. We have to ensure Thread Death propagates correctly though to stop main completely. This is to ensure that if main isn't truely stuck, it's not manipulating state we are trying to save.
2020-04-16 08:53:50 +00:00
public boolean serverAutoSave = false; // Paper
public CommandDispatcher vanillaCommandDispatcher;
- private boolean forceTicks;
+ public boolean forceTicks; // Paper
// CraftBukkit end
// Spigot start
public static final int TPS = 20;
[WIP|1.16] Misc fixes (#3627)
2020-06-26 16:20:03 +00:00
@@ -169,6 +169,8 @@ public abstract class MinecraftServer extends IAsyncTaskHandlerReentrant<TickTas
even even even even even even more work
2020-06-26 01:53:21 +00:00
public final SlackActivityAccountant slackActivityAccountant = new SlackActivityAccountant();
// Spigot end
+ public volatile Thread shutdownThread; // Paper
+
public static <S extends MinecraftServer> S a(Function<Thread, S> function) {
AtomicReference<S> atomicreference = new AtomicReference();
Thread thread = new Thread(() -> {
Fix keep-spawn-loaded stopping force-loaded chunks from loading Also, cleanup the patch a little bit
2020-12-01 13:23:44 +00:00
@@ -731,6 +733,7 @@ public abstract class MinecraftServer extends IAsyncTaskHandlerReentrant<TickTas
Fix issues with SIGTERM not shutting down gracefully - Fixes #3062 We had 2 issues. 1) Log4J2 Shutdown hook seemed to be causing issues as it shutdown logger while we still needed it 2) ServerShutdownThread needs to stay alive until server is shutdown to keep jvm open. It appears SIGINT is handled differently than SIGTERM, as SIGINT worked correctly. But this will make both methods work.
2020-06-11 03:40:28 +00:00
// CraftBukkit start
private boolean hasStopped = false;
+ public volatile boolean hasFullyShutdown = false; // Paper
private final Object stopLock = new Object();
public final boolean hasStopped() {
synchronized (stopLock) {
Fix keep-spawn-loaded stopping force-loaded chunks from loading Also, cleanup the patch a little bit
2020-12-01 13:23:44 +00:00
@@ -745,6 +748,19 @@ public abstract class MinecraftServer extends IAsyncTaskHandlerReentrant<TickTas
Forced Watchdog Crash support and Improve Async Shutdown If the request to shut down the server is received while we are in a watchdog hang, immediately treat it as a crash and begin the shutdown process. Shutdown process is now improved to also shutdown cleanly when not using restart scripts either. If a server is deadlocked, a server owner can send SIGHUP (or any other signal the JVM understands to shut down as it currently does) and the watchdog will no longer need to wait until the full timeout, allowing you to trigger a close process and try to shut the server down gracefully, saving player and world data. Previously there was no way to trigger this outside of waiting for a full watchdog timeout, which may be set to a really long time... Additionally, fix everything to do with shutting the server down asynchronously. Previously, nearly everything about the process was fragile and unsafe. Main might not have actually been frozen, and might still be manipulating state. Or, some reuest might ask main to do something in the shutdown but main is dead. Or worse, other things might start closing down items such as the Console or Thread Pool before we are fully shutdown. This change tries to resolve all of these issues by moving everything into the stop method and guaranteeing only one thread is stopping the server. We then issue Thread Death to the main thread of another thread initiates the stop process. We have to ensure Thread Death propagates correctly though to stop main completely. This is to ensure that if main isn't truely stuck, it's not manipulating state we are trying to save.
2020-04-16 08:53:50 +00:00
if (hasStopped) return;
hasStopped = true;
}
Fix deadlock issue with watchdog stopping Fixes #4008
2020-07-29 02:19:01 +00:00
+ // Paper start - kill main thread, and kill it hard
+ shutdownThread = Thread.currentThread();
+ org.spigotmc.WatchdogThread.doStop(); // Paper
+ if (!isMainThread()) {
+ MinecraftServer.LOGGER.info("Stopping main thread (Ignore any thread death message you see! - DO NOT REPORT THREAD DEATH TO PAPER)");
+ while (this.getThread().isAlive()) {
+ this.getThread().stop();
+ try {
+ Thread.sleep(1);
+ } catch (InterruptedException e) {}
+ }
+ }
+ // Paper end
Forced Watchdog Crash support and Improve Async Shutdown If the request to shut down the server is received while we are in a watchdog hang, immediately treat it as a crash and begin the shutdown process. Shutdown process is now improved to also shutdown cleanly when not using restart scripts either. If a server is deadlocked, a server owner can send SIGHUP (or any other signal the JVM understands to shut down as it currently does) and the watchdog will no longer need to wait until the full timeout, allowing you to trigger a close process and try to shut the server down gracefully, saving player and world data. Previously there was no way to trigger this outside of waiting for a full watchdog timeout, which may be set to a really long time... Additionally, fix everything to do with shutting the server down asynchronously. Previously, nearly everything about the process was fragile and unsafe. Main might not have actually been frozen, and might still be manipulating state. Or, some reuest might ask main to do something in the shutdown but main is dead. Or worse, other things might start closing down items such as the Console or Thread Pool before we are fully shutdown. This change tries to resolve all of these issues by moving everything into the stop method and guaranteeing only one thread is stopping the server. We then issue Thread Death to the main thread of another thread initiates the stop process. We have to ensure Thread Death propagates correctly though to stop main completely. This is to ensure that if main isn't truely stuck, it's not manipulating state we are trying to save.
2020-04-16 08:53:50 +00:00
// CraftBukkit end
Fix deadlock issue with watchdog stopping Fixes #4008
2020-07-29 02:19:01 +00:00
MinecraftServer.LOGGER.info("Stopping server");
Ensure ThreadDeath propagates fully - Fixes #3521
2020-06-08 15:54:12 +00:00
MinecraftTimings.stopServer(); // Paper
Fix keep-spawn-loaded stopping force-loaded chunks from loading Also, cleanup the patch a little bit
2020-12-01 13:23:44 +00:00
@@ -810,7 +826,18 @@ public abstract class MinecraftServer extends IAsyncTaskHandlerReentrant<TickTas
1.16.2 Release (#4123) PaperMC believes that 1.16.2 is now ready for general release as we fixed the main issue plagueing the 1.16.x release, the MapLike data conversion issues. Until now, it was not safe for a server to convert a world to 1.16.2 without data conversion issues around villages and potentially other things. If you did, those MapLike errors meant something went wrong. This is now resolved. Big thanks to all those that helped, notably @BillyGalbreath and @Proximyst who did large parts of the update process with me. Please as always, backup your worlds and test before updating to 1.16.2! If you update to 1.16.2, there is no going back to an older build than this. --------------------------------- Co-authored-by: William Blake Galbreath <Blake.Galbreath@GMail.com> Co-authored-by: Mariell Hoversholm <proximyst@proximyst.com> Co-authored-by: krolik-exe <69214078+krolik-exe@users.noreply.github.com> Co-authored-by: BillyGalbreath <BillyGalbreath@users.noreply.github.com> Co-authored-by: stonar96 <minecraft.stonar96@gmail.com> Co-authored-by: Shane Freeder <theboyetronic@gmail.com> Co-authored-by: Jason <jasonpenilla2@me.com> Co-authored-by: kashike <kashike@vq.lc> Co-authored-by: Aurora <21148213+aurorasmiles@users.noreply.github.com> Co-authored-by: KennyTV <kennytv@t-online.de> Co-authored-by: commandblockguy <commandblockguy1@gmail.com> Co-authored-by: DigitalRegent <misterwener@gmail.com> Co-authored-by: ishland <ishlandmc@yeah.net>
2020-08-25 02:22:08 +00:00
this.getUserCache().b(false); // Paper
Forced Watchdog Crash support and Improve Async Shutdown If the request to shut down the server is received while we are in a watchdog hang, immediately treat it as a crash and begin the shutdown process. Shutdown process is now improved to also shutdown cleanly when not using restart scripts either. If a server is deadlocked, a server owner can send SIGHUP (or any other signal the JVM understands to shut down as it currently does) and the watchdog will no longer need to wait until the full timeout, allowing you to trigger a close process and try to shut the server down gracefully, saving player and world data. Previously there was no way to trigger this outside of waiting for a full watchdog timeout, which may be set to a really long time... Additionally, fix everything to do with shutting the server down asynchronously. Previously, nearly everything about the process was fragile and unsafe. Main might not have actually been frozen, and might still be manipulating state. Or, some reuest might ask main to do something in the shutdown but main is dead. Or worse, other things might start closing down items such as the Console or Thread Pool before we are fully shutdown. This change tries to resolve all of these issues by moving everything into the stop method and guaranteeing only one thread is stopping the server. We then issue Thread Death to the main thread of another thread initiates the stop process. We have to ensure Thread Death propagates correctly though to stop main completely. This is to ensure that if main isn't truely stuck, it's not manipulating state we are trying to save.
2020-04-16 08:53:50 +00:00
}
// Spigot end
+ // Paper start - move final shutdown items here
+ LOGGER.info("Flushing Chunk IO");
com.destroystokyo.paper.io.PaperFileIOThread.Holder.INSTANCE.close(true, true); // Paper
+ LOGGER.info("Closing Thread Pool");
+ SystemUtils.shutdownServerThreadPool(); // Paper
+ LOGGER.info("Closing Server");
+ try {
+ net.minecrell.terminalconsole.TerminalConsoleAppender.close(); // Paper - Use TerminalConsoleAppender
+ } catch (Exception e) {
+ }
+ this.exit();
+ // Paper end
}
public String getServerIp() {
Fix keep-spawn-loaded stopping force-loaded chunks from loading Also, cleanup the patch a little bit
2020-12-01 13:23:44 +00:00
@@ -903,6 +930,7 @@ public abstract class MinecraftServer extends IAsyncTaskHandlerReentrant<TickTas
Move delayed init down later in tick, improve accuracy of startup time Now tracks the full startup time for "Done" message at end, as apparently Vanillas was done in a place that skipped tracking a lot of code too. This fixes an issue with ViaVersion
2020-05-07 01:24:43 +00:00
1.16.2 Release (#4123) PaperMC believes that 1.16.2 is now ready for general release as we fixed the main issue plagueing the 1.16.x release, the MapLike data conversion issues. Until now, it was not safe for a server to convert a world to 1.16.2 without data conversion issues around villages and potentially other things. If you did, those MapLike errors meant something went wrong. This is now resolved. Big thanks to all those that helped, notably @BillyGalbreath and @Proximyst who did large parts of the update process with me. Please as always, backup your worlds and test before updating to 1.16.2! If you update to 1.16.2, there is no going back to an older build than this. --------------------------------- Co-authored-by: William Blake Galbreath <Blake.Galbreath@GMail.com> Co-authored-by: Mariell Hoversholm <proximyst@proximyst.com> Co-authored-by: krolik-exe <69214078+krolik-exe@users.noreply.github.com> Co-authored-by: BillyGalbreath <BillyGalbreath@users.noreply.github.com> Co-authored-by: stonar96 <minecraft.stonar96@gmail.com> Co-authored-by: Shane Freeder <theboyetronic@gmail.com> Co-authored-by: Jason <jasonpenilla2@me.com> Co-authored-by: kashike <kashike@vq.lc> Co-authored-by: Aurora <21148213+aurorasmiles@users.noreply.github.com> Co-authored-by: KennyTV <kennytv@t-online.de> Co-authored-by: commandblockguy <commandblockguy1@gmail.com> Co-authored-by: DigitalRegent <misterwener@gmail.com> Co-authored-by: ishland <ishlandmc@yeah.net>
2020-08-25 02:22:08 +00:00
protected void w() {
Move delayed init down later in tick, improve accuracy of startup time Now tracks the full startup time for "Done" message at end, as apparently Vanillas was done in a place that skipped tracking a lot of code too. This fixes an issue with ViaVersion
2020-05-07 01:24:43 +00:00
try {
+ long serverStartTime = SystemUtils.getMonotonicNanos(); // Paper
if (this.init()) {
this.nextTick = SystemUtils.getMonotonicMillis();
this.serverPing.setMOTD(new ChatComponentText(this.motd));
Fix keep-spawn-loaded stopping force-loaded chunks from loading Also, cleanup the patch a little bit
2020-12-01 13:23:44 +00:00
@@ -910,6 +938,18 @@ public abstract class MinecraftServer extends IAsyncTaskHandlerReentrant<TickTas
Don't process watchdog until server has fully started and ticked. I think adjusting watchdog settings may of caused instant crashes. Fixes #3209
2020-04-23 08:46:53 +00:00
this.a(this.serverPing);
Forced Watchdog Crash support and Improve Async Shutdown If the request to shut down the server is received while we are in a watchdog hang, immediately treat it as a crash and begin the shutdown process. Shutdown process is now improved to also shutdown cleanly when not using restart scripts either. If a server is deadlocked, a server owner can send SIGHUP (or any other signal the JVM understands to shut down as it currently does) and the watchdog will no longer need to wait until the full timeout, allowing you to trigger a close process and try to shut the server down gracefully, saving player and world data. Previously there was no way to trigger this outside of waiting for a full watchdog timeout, which may be set to a really long time... Additionally, fix everything to do with shutting the server down asynchronously. Previously, nearly everything about the process was fragile and unsafe. Main might not have actually been frozen, and might still be manipulating state. Or, some reuest might ask main to do something in the shutdown but main is dead. Or worse, other things might start closing down items such as the Console or Thread Pool before we are fully shutdown. This change tries to resolve all of these issues by moving everything into the stop method and guaranteeing only one thread is stopping the server. We then issue Thread Death to the main thread of another thread initiates the stop process. We have to ensure Thread Death propagates correctly though to stop main completely. This is to ensure that if main isn't truely stuck, it's not manipulating state we are trying to save.
2020-04-16 08:53:50 +00:00
Don't process watchdog until server has fully started and ticked. I think adjusting watchdog settings may of caused instant crashes. Fixes #3209
2020-04-23 08:46:53 +00:00
// Spigot start
Move delayed init down later in tick, improve accuracy of startup time Now tracks the full startup time for "Done" message at end, as apparently Vanillas was done in a place that skipped tracking a lot of code too. This fixes an issue with ViaVersion
2020-05-07 01:24:43 +00:00
+ // Paper start - move done tracking
+ LOGGER.info("Running delayed init tasks");
+ this.server.getScheduler().mainThreadHeartbeat(this.ticks); // run all 1 tick delay tasks during init,
+ // this is going to be the first thing the tick process does anyways, so move done and run it after
+ // everything is init before watchdog tick.
+ // anything at 3+ won't be caught here but also will trip watchdog....
+ // tasks are default scheduled at -1 + delay, and first tick will tick at 1
+ String doneTime = String.format(java.util.Locale.ROOT, "%.3fs", (double) (SystemUtils.getMonotonicNanos() - serverStartTime) / 1.0E9D);
+ LOGGER.info("Done ({})! For help, type \"help\"", doneTime);
+ // Paper end
+
Don't process watchdog until server has fully started and ticked. I think adjusting watchdog settings may of caused instant crashes. Fixes #3209
2020-04-23 08:46:53 +00:00
+ org.spigotmc.WatchdogThread.tick(); // Paper
org.spigotmc.WatchdogThread.hasStarted = true; // Paper
Arrays.fill( recentTps, 20 );
long start = System.nanoTime(), curTime, tickSection = start; // Paper - Further improve server tick loop
Fix keep-spawn-loaded stopping force-loaded chunks from loading Also, cleanup the patch a little bit
2020-12-01 13:23:44 +00:00
@@ -965,6 +1005,12 @@ public abstract class MinecraftServer extends IAsyncTaskHandlerReentrant<TickTas
Forced Watchdog Crash support and Improve Async Shutdown If the request to shut down the server is received while we are in a watchdog hang, immediately treat it as a crash and begin the shutdown process. Shutdown process is now improved to also shutdown cleanly when not using restart scripts either. If a server is deadlocked, a server owner can send SIGHUP (or any other signal the JVM understands to shut down as it currently does) and the watchdog will no longer need to wait until the full timeout, allowing you to trigger a close process and try to shut the server down gracefully, saving player and world data. Previously there was no way to trigger this outside of waiting for a full watchdog timeout, which may be set to a really long time... Additionally, fix everything to do with shutting the server down asynchronously. Previously, nearly everything about the process was fragile and unsafe. Main might not have actually been frozen, and might still be manipulating state. Or, some reuest might ask main to do something in the shutdown but main is dead. Or worse, other things might start closing down items such as the Console or Thread Pool before we are fully shutdown. This change tries to resolve all of these issues by moving everything into the stop method and guaranteeing only one thread is stopping the server. We then issue Thread Death to the main thread of another thread initiates the stop process. We have to ensure Thread Death propagates correctly though to stop main completely. This is to ensure that if main isn't truely stuck, it's not manipulating state we are trying to save.
2020-04-16 08:53:50 +00:00
this.a((CrashReport) null);
}
} catch (Throwable throwable) {
+ // Paper start
+ if (throwable instanceof ThreadDeath) {
+ MinecraftServer.LOGGER.error("Main thread terminated by WatchDog due to hard crash", throwable);
+ return;
+ }
+ // Paper end
MinecraftServer.LOGGER.error("Encountered an unexpected exception", throwable);
// Spigot Start
if ( throwable.getCause() != null )
Fix keep-spawn-loaded stopping force-loaded chunks from loading Also, cleanup the patch a little bit
2020-12-01 13:23:44 +00:00
@@ -996,14 +1042,14 @@ public abstract class MinecraftServer extends IAsyncTaskHandlerReentrant<TickTas
Forced Watchdog Crash support and Improve Async Shutdown If the request to shut down the server is received while we are in a watchdog hang, immediately treat it as a crash and begin the shutdown process. Shutdown process is now improved to also shutdown cleanly when not using restart scripts either. If a server is deadlocked, a server owner can send SIGHUP (or any other signal the JVM understands to shut down as it currently does) and the watchdog will no longer need to wait until the full timeout, allowing you to trigger a close process and try to shut the server down gracefully, saving player and world data. Previously there was no way to trigger this outside of waiting for a full watchdog timeout, which may be set to a really long time... Additionally, fix everything to do with shutting the server down asynchronously. Previously, nearly everything about the process was fragile and unsafe. Main might not have actually been frozen, and might still be manipulating state. Or, some reuest might ask main to do something in the shutdown but main is dead. Or worse, other things might start closing down items such as the Console or Thread Pool before we are fully shutdown. This change tries to resolve all of these issues by moving everything into the stop method and guaranteeing only one thread is stopping the server. We then issue Thread Death to the main thread of another thread initiates the stop process. We have to ensure Thread Death propagates correctly though to stop main completely. This is to ensure that if main isn't truely stuck, it's not manipulating state we are trying to save.
2020-04-16 08:53:50 +00:00
} catch (Throwable throwable1) {
MinecraftServer.LOGGER.error("Exception stopping the server", throwable1);
} finally {
- org.spigotmc.WatchdogThread.doStop(); // Spigot
+ //org.spigotmc.WatchdogThread.doStop(); // Spigot // Paper - move into stop
// CraftBukkit start - Restore terminal to original settings
try {
- net.minecrell.terminalconsole.TerminalConsoleAppender.close(); // Paper - Use TerminalConsoleAppender
+ //net.minecrell.terminalconsole.TerminalConsoleAppender.close(); // Paper - Move into stop
} catch (Exception ignored) {
}
// CraftBukkit end
- this.exit();
+ //this.exit(); // Paper - moved into stop
}
}
Fix keep-spawn-loaded stopping force-loaded chunks from loading Also, cleanup the patch a little bit
2020-12-01 13:23:44 +00:00
@@ -1059,6 +1105,12 @@ public abstract class MinecraftServer extends IAsyncTaskHandlerReentrant<TickTas
Forced Watchdog Crash support and Improve Async Shutdown If the request to shut down the server is received while we are in a watchdog hang, immediately treat it as a crash and begin the shutdown process. Shutdown process is now improved to also shutdown cleanly when not using restart scripts either. If a server is deadlocked, a server owner can send SIGHUP (or any other signal the JVM understands to shut down as it currently does) and the watchdog will no longer need to wait until the full timeout, allowing you to trigger a close process and try to shut the server down gracefully, saving player and world data. Previously there was no way to trigger this outside of waiting for a full watchdog timeout, which may be set to a really long time... Additionally, fix everything to do with shutting the server down asynchronously. Previously, nearly everything about the process was fragile and unsafe. Main might not have actually been frozen, and might still be manipulating state. Or, some reuest might ask main to do something in the shutdown but main is dead. Or worse, other things might start closing down items such as the Console or Thread Pool before we are fully shutdown. This change tries to resolve all of these issues by moving everything into the stop method and guaranteeing only one thread is stopping the server. We then issue Thread Death to the main thread of another thread initiates the stop process. We have to ensure Thread Death propagates correctly though to stop main completely. This is to ensure that if main isn't truely stuck, it's not manipulating state we are trying to save.
2020-04-16 08:53:50 +00:00
@Override
protected TickTask postToMainThread(Runnable runnable) {
+ // Paper start - anything that does try to post to main during watchdog crash, run on watchdog
+ if (this.hasStopped && Thread.currentThread().equals(shutdownThread)) {
+ runnable.run();
+ runnable = () -> {};
+ }
+ // Paper end
return new TickTask(this.ticks, runnable);
}
Fix keep-spawn-loaded stopping force-loaded chunks from loading Also, cleanup the patch a little bit
2020-12-01 13:23:44 +00:00
@@ -1301,6 +1353,7 @@ public abstract class MinecraftServer extends IAsyncTaskHandlerReentrant<TickTas
even even even even even even more work
2020-06-26 01:53:21 +00:00
try {
crashreport = CrashReport.a(throwable, "Exception ticking world");
} catch (Throwable t) {
+ if (throwable instanceof ThreadDeath) { throw (ThreadDeath)throwable; } // Paper
throw new RuntimeException("Error generating crash report", t);
}
// Spigot End
Fix keep-spawn-loaded stopping force-loaded chunks from loading Also, cleanup the patch a little bit
2020-12-01 13:23:44 +00:00
@@ -1758,7 +1811,8 @@ public abstract class MinecraftServer extends IAsyncTaskHandlerReentrant<TickTas
Fix Reloadable save players error on stop Not sure what plugins or panels are doing to have triggered this. Reloading resource packs appear to have been happening during shutdown So this will abort the end of the reload process if we are not on the correct thread and stop saving players as part of this task too.
2020-06-29 00:06:26 +00:00
this.resourcePackRepository.a(collection);
this.saveData.a(a(this.resourcePackRepository));
datapackresources.i();
- this.getPlayerList().savePlayers();
+ if (Thread.currentThread() != this.serverThread) return; // Paper
+ //this.getPlayerList().savePlayers(); // Paper - we don't need to do this
this.getPlayerList().reload();
this.customFunctionData.a(this.dataPackResources.a());
this.ak.a(this.dataPackResources.h());
Preload important classes such as Logger and JLine This is for 2 reasons: 1) Ensuring our log4j is mostly loaded at OUR version. I've seen stack traces with line numbers that do not match our version. This means that some plugin has shaded in log4j and their loaded version is mixing with ours.... So by at least trying to load a bunch of log4j classes before we load plugins, we can be more sure mixed versions are not loading. 2) If the jar file is replaced while the server is runnimg class not found errors galore This will preloaod a bunch of classes commonly seen to error during shutdown due to this. The goal here is to help let the server shutdown gracefully as possible. Some plugins will still blow up here if they access a class that hadn't been loaded yet, but goal is to at least stop freezing the shutdown process as it does with JLine and Log4j errors requiring an external kill. Ideally you should not replace jars while the server is running, but it is something that happens in development for testing. Updated test server to do a copy though to avoid this happening in Paper development.
2020-05-24 17:09:02 +00:00
diff --git a/src/main/java/net/minecraft/server/PlayerChunkMap.java b/src/main/java/net/minecraft/server/PlayerChunkMap.java
1.16.2 Release (#4123) PaperMC believes that 1.16.2 is now ready for general release as we fixed the main issue plagueing the 1.16.x release, the MapLike data conversion issues. Until now, it was not safe for a server to convert a world to 1.16.2 without data conversion issues around villages and potentially other things. If you did, those MapLike errors meant something went wrong. This is now resolved. Big thanks to all those that helped, notably @BillyGalbreath and @Proximyst who did large parts of the update process with me. Please as always, backup your worlds and test before updating to 1.16.2! If you update to 1.16.2, there is no going back to an older build than this. --------------------------------- Co-authored-by: William Blake Galbreath <Blake.Galbreath@GMail.com> Co-authored-by: Mariell Hoversholm <proximyst@proximyst.com> Co-authored-by: krolik-exe <69214078+krolik-exe@users.noreply.github.com> Co-authored-by: BillyGalbreath <BillyGalbreath@users.noreply.github.com> Co-authored-by: stonar96 <minecraft.stonar96@gmail.com> Co-authored-by: Shane Freeder <theboyetronic@gmail.com> Co-authored-by: Jason <jasonpenilla2@me.com> Co-authored-by: kashike <kashike@vq.lc> Co-authored-by: Aurora <21148213+aurorasmiles@users.noreply.github.com> Co-authored-by: KennyTV <kennytv@t-online.de> Co-authored-by: commandblockguy <commandblockguy1@gmail.com> Co-authored-by: DigitalRegent <misterwener@gmail.com> Co-authored-by: ishland <ishlandmc@yeah.net>
2020-08-25 02:22:08 +00:00
index d3a597d878de4cb625484cbf3a9838961d38ada0..16e4acdb0f834883a480829a864ef7570035bc26 100644
Preload important classes such as Logger and JLine This is for 2 reasons: 1) Ensuring our log4j is mostly loaded at OUR version. I've seen stack traces with line numbers that do not match our version. This means that some plugin has shaded in log4j and their loaded version is mixing with ours.... So by at least trying to load a bunch of log4j classes before we load plugins, we can be more sure mixed versions are not loading. 2) If the jar file is replaced while the server is runnimg class not found errors galore This will preloaod a bunch of classes commonly seen to error during shutdown due to this. The goal here is to help let the server shutdown gracefully as possible. Some plugins will still blow up here if they access a class that hadn't been loaded yet, but goal is to at least stop freezing the shutdown process as it does with JLine and Log4j errors requiring an external kill. Ideally you should not replace jars while the server is running, but it is something that happens in development for testing. Updated test server to do a copy though to avoid this happening in Paper development.
2020-05-24 17:09:02 +00:00
--- a/src/main/java/net/minecraft/server/PlayerChunkMap.java
+++ b/src/main/java/net/minecraft/server/PlayerChunkMap.java
1.16.2 Release (#4123) PaperMC believes that 1.16.2 is now ready for general release as we fixed the main issue plagueing the 1.16.x release, the MapLike data conversion issues. Until now, it was not safe for a server to convert a world to 1.16.2 without data conversion issues around villages and potentially other things. If you did, those MapLike errors meant something went wrong. This is now resolved. Big thanks to all those that helped, notably @BillyGalbreath and @Proximyst who did large parts of the update process with me. Please as always, backup your worlds and test before updating to 1.16.2! If you update to 1.16.2, there is no going back to an older build than this. --------------------------------- Co-authored-by: William Blake Galbreath <Blake.Galbreath@GMail.com> Co-authored-by: Mariell Hoversholm <proximyst@proximyst.com> Co-authored-by: krolik-exe <69214078+krolik-exe@users.noreply.github.com> Co-authored-by: BillyGalbreath <BillyGalbreath@users.noreply.github.com> Co-authored-by: stonar96 <minecraft.stonar96@gmail.com> Co-authored-by: Shane Freeder <theboyetronic@gmail.com> Co-authored-by: Jason <jasonpenilla2@me.com> Co-authored-by: kashike <kashike@vq.lc> Co-authored-by: Aurora <21148213+aurorasmiles@users.noreply.github.com> Co-authored-by: KennyTV <kennytv@t-online.de> Co-authored-by: commandblockguy <commandblockguy1@gmail.com> Co-authored-by: DigitalRegent <misterwener@gmail.com> Co-authored-by: ishland <ishlandmc@yeah.net>
2020-08-25 02:22:08 +00:00
@@ -487,6 +487,7 @@ public class PlayerChunkMap extends IChunkLoader implements PlayerChunk.d {
Preload important classes such as Logger and JLine This is for 2 reasons: 1) Ensuring our log4j is mostly loaded at OUR version. I've seen stack traces with line numbers that do not match our version. This means that some plugin has shaded in log4j and their loaded version is mixing with ours.... So by at least trying to load a bunch of log4j classes before we load plugins, we can be more sure mixed versions are not loading. 2) If the jar file is replaced while the server is runnimg class not found errors galore This will preloaod a bunch of classes commonly seen to error during shutdown due to this. The goal here is to help let the server shutdown gracefully as possible. Some plugins will still blow up here if they access a class that hadn't been loaded yet, but goal is to at least stop freezing the shutdown process as it does with JLine and Log4j errors requiring an external kill. Ideally you should not replace jars while the server is running, but it is something that happens in development for testing. Updated test server to do a copy though to avoid this happening in Paper development.
2020-05-24 17:09:02 +00:00
MutableBoolean mutableboolean = new MutableBoolean();
do {
+ boolean isShuttingDown = world.getMinecraftServer().hasStopped(); // Paper
mutableboolean.setFalse();
list.stream().map((playerchunk) -> {
CompletableFuture completablefuture;
Forced Watchdog Crash support and Improve Async Shutdown If the request to shut down the server is received while we are in a watchdog hang, immediately treat it as a crash and begin the shutdown process. Shutdown process is now improved to also shutdown cleanly when not using restart scripts either. If a server is deadlocked, a server owner can send SIGHUP (or any other signal the JVM understands to shut down as it currently does) and the watchdog will no longer need to wait until the full timeout, allowing you to trigger a close process and try to shut the server down gracefully, saving player and world data. Previously there was no way to trigger this outside of waiting for a full watchdog timeout, which may be set to a really long time... Additionally, fix everything to do with shutting the server down asynchronously. Previously, nearly everything about the process was fragile and unsafe. Main might not have actually been frozen, and might still be manipulating state. Or, some reuest might ask main to do something in the shutdown but main is dead. Or worse, other things might start closing down items such as the Console or Thread Pool before we are fully shutdown. This change tries to resolve all of these issues by moving everything into the stop method and guaranteeing only one thread is stopping the server. We then issue Thread Death to the main thread of another thread initiates the stop process. We have to ensure Thread Death propagates correctly though to stop main completely. This is to ensure that if main isn't truely stuck, it's not manipulating state we are trying to save.
2020-04-16 08:53:50 +00:00
diff --git a/src/main/java/net/minecraft/server/PlayerList.java b/src/main/java/net/minecraft/server/PlayerList.java
[CI-SKIP] Remove Waving banner fix (#4786) Mojang fixed it in MC-63720
2020-11-19 15:15:16 +00:00
index a1987b17adb5fa8fdeb601da948839fb1fd35aaf..fbd0ecb7bee67bf5affb6be6a57a8a0a8e2686b3 100644
Forced Watchdog Crash support and Improve Async Shutdown If the request to shut down the server is received while we are in a watchdog hang, immediately treat it as a crash and begin the shutdown process. Shutdown process is now improved to also shutdown cleanly when not using restart scripts either. If a server is deadlocked, a server owner can send SIGHUP (or any other signal the JVM understands to shut down as it currently does) and the watchdog will no longer need to wait until the full timeout, allowing you to trigger a close process and try to shut the server down gracefully, saving player and world data. Previously there was no way to trigger this outside of waiting for a full watchdog timeout, which may be set to a really long time... Additionally, fix everything to do with shutting the server down asynchronously. Previously, nearly everything about the process was fragile and unsafe. Main might not have actually been frozen, and might still be manipulating state. Or, some reuest might ask main to do something in the shutdown but main is dead. Or worse, other things might start closing down items such as the Console or Thread Pool before we are fully shutdown. This change tries to resolve all of these issues by moving everything into the stop method and guaranteeing only one thread is stopping the server. We then issue Thread Death to the main thread of another thread initiates the stop process. We have to ensure Thread Death propagates correctly though to stop main completely. This is to ensure that if main isn't truely stuck, it's not manipulating state we are trying to save.
2020-04-16 08:53:50 +00:00
--- a/src/main/java/net/minecraft/server/PlayerList.java
+++ b/src/main/java/net/minecraft/server/PlayerList.java
Updated Upstream (CraftBukkit) Upstream has released updates that appears to apply and compile correctly. This update has not been tested by PaperMC and as with ANY update, please do your own testing CraftBukkit Changes: 5c77bd28 SPIGOT-6147: InventoryCloseEvent does not fire after closing player inventory
2020-10-01 13:16:32 +00:00
@@ -432,7 +432,7 @@ public abstract class PlayerList {
Forced Watchdog Crash support and Improve Async Shutdown If the request to shut down the server is received while we are in a watchdog hang, immediately treat it as a crash and begin the shutdown process. Shutdown process is now improved to also shutdown cleanly when not using restart scripts either. If a server is deadlocked, a server owner can send SIGHUP (or any other signal the JVM understands to shut down as it currently does) and the watchdog will no longer need to wait until the full timeout, allowing you to trigger a close process and try to shut the server down gracefully, saving player and world data. Previously there was no way to trigger this outside of waiting for a full watchdog timeout, which may be set to a really long time... Additionally, fix everything to do with shutting the server down asynchronously. Previously, nearly everything about the process was fragile and unsafe. Main might not have actually been frozen, and might still be manipulating state. Or, some reuest might ask main to do something in the shutdown but main is dead. Or worse, other things might start closing down items such as the Console or Thread Pool before we are fully shutdown. This change tries to resolve all of these issues by moving everything into the stop method and guaranteeing only one thread is stopping the server. We then issue Thread Death to the main thread of another thread initiates the stop process. We have to ensure Thread Death propagates correctly though to stop main completely. This is to ensure that if main isn't truely stuck, it's not manipulating state we are trying to save.
2020-04-16 08:53:50 +00:00
cserver.getPluginManager().callEvent(playerQuitEvent);
entityplayer.getBukkitEntity().disconnect(playerQuitEvent.getQuitMessage());
1.15.2 - Updated Upstream (Bukkit/CraftBukkit) Upstream has released updates that appears to apply and compile correctly. This update has not been tested by PaperMC and as with ANY update, please do your own testing Bukkit Changes: 149527f7 SPIGOT-5782: Set Arrow Launched From Crossbow CraftBukkit Changes: be6aaf04 SPIGOT-5782: Set Arrow Launched From Crossbow 833da9c4 SPIGOT-5799: InventoryCloseEvent fires after PlayerQuitEvent 26c0084f SPIGOT-5675, SPIGOT-5798, MC-149563: Fix tracking of entities across dimensions 7f3e7c3f SPIGOT-5797: Zombie(Villagers) Instant Convert based on their lifetime
2020-06-24 00:25:28 +00:00
- entityplayer.playerTick(); // SPIGOT-924
+ if (server.isMainThread()) entityplayer.playerTick(); // SPIGOT-924 // Paper - don't tick during emergency shutdowns (Watchdog)
Forced Watchdog Crash support and Improve Async Shutdown If the request to shut down the server is received while we are in a watchdog hang, immediately treat it as a crash and begin the shutdown process. Shutdown process is now improved to also shutdown cleanly when not using restart scripts either. If a server is deadlocked, a server owner can send SIGHUP (or any other signal the JVM understands to shut down as it currently does) and the watchdog will no longer need to wait until the full timeout, allowing you to trigger a close process and try to shut the server down gracefully, saving player and world data. Previously there was no way to trigger this outside of waiting for a full watchdog timeout, which may be set to a really long time... Additionally, fix everything to do with shutting the server down asynchronously. Previously, nearly everything about the process was fragile and unsafe. Main might not have actually been frozen, and might still be manipulating state. Or, some reuest might ask main to do something in the shutdown but main is dead. Or worse, other things might start closing down items such as the Console or Thread Pool before we are fully shutdown. This change tries to resolve all of these issues by moving everything into the stop method and guaranteeing only one thread is stopping the server. We then issue Thread Death to the main thread of another thread initiates the stop process. We have to ensure Thread Death propagates correctly though to stop main completely. This is to ensure that if main isn't truely stuck, it's not manipulating state we are trying to save.
2020-04-16 08:53:50 +00:00
// CraftBukkit end
// Paper start - Remove from collideRule team if needed
diff --git a/src/main/java/net/minecraft/server/SystemUtils.java b/src/main/java/net/minecraft/server/SystemUtils.java
Updated Upstream (CraftBukkit/Spigot) (#4318) Upstream has released updates that appears to apply and compile correctly. This update has not been tested by PaperMC and as with ANY update, please do your own testing CraftBukkit Changes: d5a72960 SPIGOT-6063: ConsoleSender sending extra lines in Java 13+ Spigot Changes: 2740d5ae Rebuild patches
2020-09-12 19:57:21 +00:00
index 1fe1df445ba56b2f176ee25502a774aa0a7bd00b..916a6a1fc1ccc73e4fb974ad77310d16fd8bd7e6 100644
Forced Watchdog Crash support and Improve Async Shutdown If the request to shut down the server is received while we are in a watchdog hang, immediately treat it as a crash and begin the shutdown process. Shutdown process is now improved to also shutdown cleanly when not using restart scripts either. If a server is deadlocked, a server owner can send SIGHUP (or any other signal the JVM understands to shut down as it currently does) and the watchdog will no longer need to wait until the full timeout, allowing you to trigger a close process and try to shut the server down gracefully, saving player and world data. Previously there was no way to trigger this outside of waiting for a full watchdog timeout, which may be set to a really long time... Additionally, fix everything to do with shutting the server down asynchronously. Previously, nearly everything about the process was fragile and unsafe. Main might not have actually been frozen, and might still be manipulating state. Or, some reuest might ask main to do something in the shutdown but main is dead. Or worse, other things might start closing down items such as the Console or Thread Pool before we are fully shutdown. This change tries to resolve all of these issues by moving everything into the stop method and guaranteeing only one thread is stopping the server. We then issue Thread Death to the main thread of another thread initiates the stop process. We have to ensure Thread Death propagates correctly though to stop main completely. This is to ensure that if main isn't truely stuck, it's not manipulating state we are trying to save.
2020-04-16 08:53:50 +00:00
--- a/src/main/java/net/minecraft/server/SystemUtils.java
+++ b/src/main/java/net/minecraft/server/SystemUtils.java
1.16.2 Release (#4123) PaperMC believes that 1.16.2 is now ready for general release as we fixed the main issue plagueing the 1.16.x release, the MapLike data conversion issues. Until now, it was not safe for a server to convert a world to 1.16.2 without data conversion issues around villages and potentially other things. If you did, those MapLike errors meant something went wrong. This is now resolved. Big thanks to all those that helped, notably @BillyGalbreath and @Proximyst who did large parts of the update process with me. Please as always, backup your worlds and test before updating to 1.16.2! If you update to 1.16.2, there is no going back to an older build than this. --------------------------------- Co-authored-by: William Blake Galbreath <Blake.Galbreath@GMail.com> Co-authored-by: Mariell Hoversholm <proximyst@proximyst.com> Co-authored-by: krolik-exe <69214078+krolik-exe@users.noreply.github.com> Co-authored-by: BillyGalbreath <BillyGalbreath@users.noreply.github.com> Co-authored-by: stonar96 <minecraft.stonar96@gmail.com> Co-authored-by: Shane Freeder <theboyetronic@gmail.com> Co-authored-by: Jason <jasonpenilla2@me.com> Co-authored-by: kashike <kashike@vq.lc> Co-authored-by: Aurora <21148213+aurorasmiles@users.noreply.github.com> Co-authored-by: KennyTV <kennytv@t-online.de> Co-authored-by: commandblockguy <commandblockguy1@gmail.com> Co-authored-by: DigitalRegent <misterwener@gmail.com> Co-authored-by: ishland <ishlandmc@yeah.net>
2020-08-25 02:22:08 +00:00
@@ -123,6 +123,7 @@ public class SystemUtils {
even even even even even even more work
2020-06-26 01:53:21 +00:00
return SystemUtils.f;
Forced Watchdog Crash support and Improve Async Shutdown If the request to shut down the server is received while we are in a watchdog hang, immediately treat it as a crash and begin the shutdown process. Shutdown process is now improved to also shutdown cleanly when not using restart scripts either. If a server is deadlocked, a server owner can send SIGHUP (or any other signal the JVM understands to shut down as it currently does) and the watchdog will no longer need to wait until the full timeout, allowing you to trigger a close process and try to shut the server down gracefully, saving player and world data. Previously there was no way to trigger this outside of waiting for a full watchdog timeout, which may be set to a really long time... Additionally, fix everything to do with shutting the server down asynchronously. Previously, nearly everything about the process was fragile and unsafe. Main might not have actually been frozen, and might still be manipulating state. Or, some reuest might ask main to do something in the shutdown but main is dead. Or worse, other things might start closing down items such as the Console or Thread Pool before we are fully shutdown. This change tries to resolve all of these issues by moving everything into the stop method and guaranteeing only one thread is stopping the server. We then issue Thread Death to the main thread of another thread initiates the stop process. We have to ensure Thread Death propagates correctly though to stop main completely. This is to ensure that if main isn't truely stuck, it's not manipulating state we are trying to save.
2020-04-16 08:53:50 +00:00
}
even even even even even even more work
2020-06-26 01:53:21 +00:00
+ public static void shutdownServerThreadPool() { h(); } // Paper - OBFHELPER
public static void h() {
a(SystemUtils.e);
a(SystemUtils.f);
Forced Watchdog Crash support and Improve Async Shutdown If the request to shut down the server is received while we are in a watchdog hang, immediately treat it as a crash and begin the shutdown process. Shutdown process is now improved to also shutdown cleanly when not using restart scripts either. If a server is deadlocked, a server owner can send SIGHUP (or any other signal the JVM understands to shut down as it currently does) and the watchdog will no longer need to wait until the full timeout, allowing you to trigger a close process and try to shut the server down gracefully, saving player and world data. Previously there was no way to trigger this outside of waiting for a full watchdog timeout, which may be set to a really long time... Additionally, fix everything to do with shutting the server down asynchronously. Previously, nearly everything about the process was fragile and unsafe. Main might not have actually been frozen, and might still be manipulating state. Or, some reuest might ask main to do something in the shutdown but main is dead. Or worse, other things might start closing down items such as the Console or Thread Pool before we are fully shutdown. This change tries to resolve all of these issues by moving everything into the stop method and guaranteeing only one thread is stopping the server. We then issue Thread Death to the main thread of another thread initiates the stop process. We have to ensure Thread Death propagates correctly though to stop main completely. This is to ensure that if main isn't truely stuck, it's not manipulating state we are trying to save.
2020-04-16 08:53:50 +00:00
diff --git a/src/main/java/net/minecraft/server/World.java b/src/main/java/net/minecraft/server/World.java
Updated Upstream (Bukkit/CraftBukkit) Upstream has released updates that appear to apply and compile correctly. This update has not been tested by PaperMC and as with ANY update, please do your own testing Bukkit Changes: 19b7b7bd #561: Add clear weather World API 5929c808 #552: Add the ability to retrieve hit, step, fall, and other sounds from blocks. CraftBukkit Changes: e1ebdd92 #771: Add clear weather World API 424598d2 #752: Add the ability to retrieve hit, step, fall, and other sounds from blocks.
2020-11-25 23:45:25 +00:00
index 8c58581417ca3e0336359e7db65568a08e025d6d..0dac8af6680a7dc4ecbedf70707297690fb1e694 100644
Forced Watchdog Crash support and Improve Async Shutdown If the request to shut down the server is received while we are in a watchdog hang, immediately treat it as a crash and begin the shutdown process. Shutdown process is now improved to also shutdown cleanly when not using restart scripts either. If a server is deadlocked, a server owner can send SIGHUP (or any other signal the JVM understands to shut down as it currently does) and the watchdog will no longer need to wait until the full timeout, allowing you to trigger a close process and try to shut the server down gracefully, saving player and world data. Previously there was no way to trigger this outside of waiting for a full watchdog timeout, which may be set to a really long time... Additionally, fix everything to do with shutting the server down asynchronously. Previously, nearly everything about the process was fragile and unsafe. Main might not have actually been frozen, and might still be manipulating state. Or, some reuest might ask main to do something in the shutdown but main is dead. Or worse, other things might start closing down items such as the Console or Thread Pool before we are fully shutdown. This change tries to resolve all of these issues by moving everything into the stop method and guaranteeing only one thread is stopping the server. We then issue Thread Death to the main thread of another thread initiates the stop process. We have to ensure Thread Death propagates correctly though to stop main completely. This is to ensure that if main isn't truely stuck, it's not manipulating state we are trying to save.
2020-04-16 08:53:50 +00:00
--- a/src/main/java/net/minecraft/server/World.java
+++ b/src/main/java/net/minecraft/server/World.java
Optimize some methods for inlining Adds final to some methods to improve inlining ability
2020-09-19 15:51:01 +00:00
@@ -797,6 +797,7 @@ public abstract class World implements GeneratorAccess, AutoCloseable {
Forced Watchdog Crash support and Improve Async Shutdown If the request to shut down the server is received while we are in a watchdog hang, immediately treat it as a crash and begin the shutdown process. Shutdown process is now improved to also shutdown cleanly when not using restart scripts either. If a server is deadlocked, a server owner can send SIGHUP (or any other signal the JVM understands to shut down as it currently does) and the watchdog will no longer need to wait until the full timeout, allowing you to trigger a close process and try to shut the server down gracefully, saving player and world data. Previously there was no way to trigger this outside of waiting for a full watchdog timeout, which may be set to a really long time... Additionally, fix everything to do with shutting the server down asynchronously. Previously, nearly everything about the process was fragile and unsafe. Main might not have actually been frozen, and might still be manipulating state. Or, some reuest might ask main to do something in the shutdown but main is dead. Or worse, other things might start closing down items such as the Console or Thread Pool before we are fully shutdown. This change tries to resolve all of these issues by moving everything into the stop method and guaranteeing only one thread is stopping the server. We then issue Thread Death to the main thread of another thread initiates the stop process. We have to ensure Thread Death propagates correctly though to stop main completely. This is to ensure that if main isn't truely stuck, it's not manipulating state we are trying to save.
2020-04-16 08:53:50 +00:00
gameprofilerfiller.exit();
} catch (Throwable throwable) {
+ if (throwable instanceof ThreadDeath) throw throwable; // Paper
// Paper start - Prevent tile entity and entity crashes
String msg = "TileEntity threw exception at " + tileentity.world.getWorld().getName() + ":" + tileentity.position.getX() + "," + tileentity.position.getY() + "," + tileentity.position.getZ();
System.err.println(msg);
Optimize some methods for inlining Adds final to some methods to improve inlining ability
2020-09-19 15:51:01 +00:00
@@ -872,6 +873,7 @@ public abstract class World implements GeneratorAccess, AutoCloseable {
Forced Watchdog Crash support and Improve Async Shutdown If the request to shut down the server is received while we are in a watchdog hang, immediately treat it as a crash and begin the shutdown process. Shutdown process is now improved to also shutdown cleanly when not using restart scripts either. If a server is deadlocked, a server owner can send SIGHUP (or any other signal the JVM understands to shut down as it currently does) and the watchdog will no longer need to wait until the full timeout, allowing you to trigger a close process and try to shut the server down gracefully, saving player and world data. Previously there was no way to trigger this outside of waiting for a full watchdog timeout, which may be set to a really long time... Additionally, fix everything to do with shutting the server down asynchronously. Previously, nearly everything about the process was fragile and unsafe. Main might not have actually been frozen, and might still be manipulating state. Or, some reuest might ask main to do something in the shutdown but main is dead. Or worse, other things might start closing down items such as the Console or Thread Pool before we are fully shutdown. This change tries to resolve all of these issues by moving everything into the stop method and guaranteeing only one thread is stopping the server. We then issue Thread Death to the main thread of another thread initiates the stop process. We have to ensure Thread Death propagates correctly though to stop main completely. This is to ensure that if main isn't truely stuck, it's not manipulating state we are trying to save.
2020-04-16 08:53:50 +00:00
try {
consumer.accept(entity);
} catch (Throwable throwable) {
+ if (throwable instanceof ThreadDeath) throw throwable; // Paper
// Paper start - Prevent tile entity and entity crashes
String msg = "Entity threw exception at " + entity.world.getWorld().getName() + ":" + entity.locX() + "," + entity.locY() + "," + entity.locZ();
System.err.println(msg);
diff --git a/src/main/java/org/bukkit/craftbukkit/CraftServer.java b/src/main/java/org/bukkit/craftbukkit/CraftServer.java
[Auto] Updated Upstream (Bukkit/CraftBukkit/Spigot) Upstream has released updates that appear to apply and compile correctly. This update has not been tested by PaperMC and as with ANY update, please do your own testing Bukkit Changes: 8f495b8d #564: Add method to get max world size CraftBukkit Changes: 768d7fc2d #773: Add method to get max world size Spigot Changes: 628435a8 #103: Add async catchers to Chunk#getEntities
2020-12-13 00:42:05 +00:00
index f546f7ac0484b6224b878ad39285ffccd7410b96..38cb08adf3f71ac40c70b3443b238ac1c67d2b4a 100644
Forced Watchdog Crash support and Improve Async Shutdown If the request to shut down the server is received while we are in a watchdog hang, immediately treat it as a crash and begin the shutdown process. Shutdown process is now improved to also shutdown cleanly when not using restart scripts either. If a server is deadlocked, a server owner can send SIGHUP (or any other signal the JVM understands to shut down as it currently does) and the watchdog will no longer need to wait until the full timeout, allowing you to trigger a close process and try to shut the server down gracefully, saving player and world data. Previously there was no way to trigger this outside of waiting for a full watchdog timeout, which may be set to a really long time... Additionally, fix everything to do with shutting the server down asynchronously. Previously, nearly everything about the process was fragile and unsafe. Main might not have actually been frozen, and might still be manipulating state. Or, some reuest might ask main to do something in the shutdown but main is dead. Or worse, other things might start closing down items such as the Console or Thread Pool before we are fully shutdown. This change tries to resolve all of these issues by moving everything into the stop method and guaranteeing only one thread is stopping the server. We then issue Thread Death to the main thread of another thread initiates the stop process. We have to ensure Thread Death propagates correctly though to stop main completely. This is to ensure that if main isn't truely stuck, it's not manipulating state we are trying to save.
2020-04-16 08:53:50 +00:00
--- a/src/main/java/org/bukkit/craftbukkit/CraftServer.java
+++ b/src/main/java/org/bukkit/craftbukkit/CraftServer.java
[Auto] Updated Upstream (Bukkit/CraftBukkit/Spigot) Upstream has released updates that appear to apply and compile correctly. This update has not been tested by PaperMC and as with ANY update, please do your own testing Bukkit Changes: 8f495b8d #564: Add method to get max world size CraftBukkit Changes: 768d7fc2d #773: Add method to get max world size Spigot Changes: 628435a8 #103: Add async catchers to Chunk#getEntities
2020-12-13 00:42:05 +00:00
@@ -1793,7 +1793,7 @@ public final class CraftServer implements Server {
Forced Watchdog Crash support and Improve Async Shutdown If the request to shut down the server is received while we are in a watchdog hang, immediately treat it as a crash and begin the shutdown process. Shutdown process is now improved to also shutdown cleanly when not using restart scripts either. If a server is deadlocked, a server owner can send SIGHUP (or any other signal the JVM understands to shut down as it currently does) and the watchdog will no longer need to wait until the full timeout, allowing you to trigger a close process and try to shut the server down gracefully, saving player and world data. Previously there was no way to trigger this outside of waiting for a full watchdog timeout, which may be set to a really long time... Additionally, fix everything to do with shutting the server down asynchronously. Previously, nearly everything about the process was fragile and unsafe. Main might not have actually been frozen, and might still be manipulating state. Or, some reuest might ask main to do something in the shutdown but main is dead. Or worse, other things might start closing down items such as the Console or Thread Pool before we are fully shutdown. This change tries to resolve all of these issues by moving everything into the stop method and guaranteeing only one thread is stopping the server. We then issue Thread Death to the main thread of another thread initiates the stop process. We have to ensure Thread Death propagates correctly though to stop main completely. This is to ensure that if main isn't truely stuck, it's not manipulating state we are trying to save.
2020-04-16 08:53:50 +00:00
@Override
public boolean isPrimaryThread() {
- return Thread.currentThread().equals(console.serverThread); // Paper - Fix issues with detecting main thread properly
+ return Thread.currentThread().equals(console.serverThread) || Thread.currentThread().equals(net.minecraft.server.MinecraftServer.getServer().shutdownThread); // Paper - Fix issues with detecting main thread properly, the only time Watchdog will be used is during a crash shutdown which is a "try our best" scenario
}
@Override
Preload important classes such as Logger and JLine This is for 2 reasons: 1) Ensuring our log4j is mostly loaded at OUR version. I've seen stack traces with line numbers that do not match our version. This means that some plugin has shaded in log4j and their loaded version is mixing with ours.... So by at least trying to load a bunch of log4j classes before we load plugins, we can be more sure mixed versions are not loading. 2) If the jar file is replaced while the server is runnimg class not found errors galore This will preloaod a bunch of classes commonly seen to error during shutdown due to this. The goal here is to help let the server shutdown gracefully as possible. Some plugins will still blow up here if they access a class that hadn't been loaded yet, but goal is to at least stop freezing the shutdown process as it does with JLine and Log4j errors requiring an external kill. Ideally you should not replace jars while the server is running, but it is something that happens in development for testing. Updated test server to do a copy though to avoid this happening in Paper development.
2020-05-24 17:09:02 +00:00
diff --git a/src/main/java/org/bukkit/craftbukkit/Main.java b/src/main/java/org/bukkit/craftbukkit/Main.java
[Auto] Updated Upstream (Bukkit/CraftBukkit/Spigot) Upstream has released updates that appear to apply and compile correctly. This update has not been tested by PaperMC and as with ANY update, please do your own testing Bukkit Changes: 8f495b8d #564: Add method to get max world size CraftBukkit Changes: 768d7fc2d #773: Add method to get max world size Spigot Changes: 628435a8 #103: Add async catchers to Chunk#getEntities
2020-12-13 00:42:05 +00:00
index 67331fa5463dd31e4aea3aebf6204ec5cb43d99e..45da87c75a788037b638088d9249b20b7f9b2686 100644
Preload important classes such as Logger and JLine This is for 2 reasons: 1) Ensuring our log4j is mostly loaded at OUR version. I've seen stack traces with line numbers that do not match our version. This means that some plugin has shaded in log4j and their loaded version is mixing with ours.... So by at least trying to load a bunch of log4j classes before we load plugins, we can be more sure mixed versions are not loading. 2) If the jar file is replaced while the server is runnimg class not found errors galore This will preloaod a bunch of classes commonly seen to error during shutdown due to this. The goal here is to help let the server shutdown gracefully as possible. Some plugins will still blow up here if they access a class that hadn't been loaded yet, but goal is to at least stop freezing the shutdown process as it does with JLine and Log4j errors requiring an external kill. Ideally you should not replace jars while the server is running, but it is something that happens in development for testing. Updated test server to do a copy though to avoid this happening in Paper development.
2020-05-24 17:09:02 +00:00
--- a/src/main/java/org/bukkit/craftbukkit/Main.java
+++ b/src/main/java/org/bukkit/craftbukkit/Main.java
@@ -150,6 +150,37 @@ public class Main {
OptionSet options = null;
+ // Paper start - preload logger classes to avoid plugins mixing versions
+ tryPreloadClass("com.destroystokyo.paper.log.LogFullPolicy");
+ tryPreloadClass("org.apache.logging.log4j.core.Core");
+ tryPreloadClass("org.apache.logging.log4j.core.Appender");
+ tryPreloadClass("org.apache.logging.log4j.core.ContextDataInjector");
+ tryPreloadClass("org.apache.logging.log4j.core.Filter");
+ tryPreloadClass("org.apache.logging.log4j.core.ErrorHandler");
+ tryPreloadClass("org.apache.logging.log4j.core.LogEvent");
+ tryPreloadClass("org.apache.logging.log4j.core.Logger");
+ tryPreloadClass("org.apache.logging.log4j.core.LoggerContext");
+ tryPreloadClass("org.apache.logging.log4j.core.LogEventListener");
+ tryPreloadClass("org.apache.logging.log4j.core.AbstractLogEvent");
+ tryPreloadClass("org.apache.logging.log4j.message.AsynchronouslyFormattable");
+ tryPreloadClass("org.apache.logging.log4j.message.FormattedMessage");
+ tryPreloadClass("org.apache.logging.log4j.message.ParameterizedMessage");
+ tryPreloadClass("org.apache.logging.log4j.message.Message");
+ tryPreloadClass("org.apache.logging.log4j.message.MessageFactory");
+ tryPreloadClass("org.apache.logging.log4j.message.TimestampMessage");
+ tryPreloadClass("org.apache.logging.log4j.message.SimpleMessage");
+ tryPreloadClass("org.apache.logging.log4j.core.async.AsyncLogger");
+ tryPreloadClass("org.apache.logging.log4j.core.async.AsyncLoggerContext");
+ tryPreloadClass("org.apache.logging.log4j.core.async.AsyncQueueFullPolicy");
+ tryPreloadClass("org.apache.logging.log4j.core.async.AsyncLoggerDisruptor");
+ tryPreloadClass("org.apache.logging.log4j.core.async.RingBufferLogEvent");
+ tryPreloadClass("org.apache.logging.log4j.core.async.DisruptorUtil");
+ tryPreloadClass("org.apache.logging.log4j.core.async.RingBufferLogEventHandler");
+ tryPreloadClass("org.apache.logging.log4j.core.impl.ThrowableProxy");
+ tryPreloadClass("org.apache.logging.log4j.core.impl.ThrowableProxy$CacheEntry");
+ tryPreloadClass("org.apache.logging.log4j.core.impl.ExtendedClassInfo");
+ tryPreloadClass("org.apache.logging.log4j.core.impl.ExtendedStackTraceElement");
+ // Paper end
try {
options = parser.parse(args);
} catch (joptsimple.OptionException ex) {
(1.16.3) Missed some fixes to include in commit
2020-09-10 23:56:58 +00:00
@@ -245,8 +276,65 @@ public class Main {
Preload important classes such as Logger and JLine This is for 2 reasons: 1) Ensuring our log4j is mostly loaded at OUR version. I've seen stack traces with line numbers that do not match our version. This means that some plugin has shaded in log4j and their loaded version is mixing with ours.... So by at least trying to load a bunch of log4j classes before we load plugins, we can be more sure mixed versions are not loading. 2) If the jar file is replaced while the server is runnimg class not found errors galore This will preloaod a bunch of classes commonly seen to error during shutdown due to this. The goal here is to help let the server shutdown gracefully as possible. Some plugins will still blow up here if they access a class that hadn't been loaded yet, but goal is to at least stop freezing the shutdown process as it does with JLine and Log4j errors requiring an external kill. Ideally you should not replace jars while the server is running, but it is something that happens in development for testing. Updated test server to do a copy though to avoid this happening in Paper development.
2020-05-24 17:09:02 +00:00
} catch (Throwable t) {
t.printStackTrace();
}
+ // Paper start
+ // load some required classes to avoid errors during shutdown if jar is replaced
+ // also to guarantee our version loads over plugins
+ tryPreloadClass("com.destroystokyo.paper.util.SneakyThrow");
+ tryPreloadClass("com.google.common.collect.Iterators$PeekingImpl");
+ tryPreloadClass("com.google.common.collect.MapMakerInternalMap$Values");
+ tryPreloadClass("com.google.common.collect.MapMakerInternalMap$ValueIterator");
Don't unregister the primary dimensions on world unload
2020-05-27 04:15:57 +00:00
+ tryPreloadClass("com.google.common.collect.MapMakerInternalMap$WriteThroughEntry");
Preload important classes such as Logger and JLine This is for 2 reasons: 1) Ensuring our log4j is mostly loaded at OUR version. I've seen stack traces with line numbers that do not match our version. This means that some plugin has shaded in log4j and their loaded version is mixing with ours.... So by at least trying to load a bunch of log4j classes before we load plugins, we can be more sure mixed versions are not loading. 2) If the jar file is replaced while the server is runnimg class not found errors galore This will preloaod a bunch of classes commonly seen to error during shutdown due to this. The goal here is to help let the server shutdown gracefully as possible. Some plugins will still blow up here if they access a class that hadn't been loaded yet, but goal is to at least stop freezing the shutdown process as it does with JLine and Log4j errors requiring an external kill. Ideally you should not replace jars while the server is running, but it is something that happens in development for testing. Updated test server to do a copy though to avoid this happening in Paper development.
2020-05-24 17:09:02 +00:00
+ tryPreloadClass("com.google.common.collect.Iterables");
+ for (int i = 1; i <= 15; i++) {
+ tryPreloadClass("com.google.common.collect.Iterables$" + i, false);
+ }
+ tryPreloadClass("org.apache.commons.lang3.mutable.MutableBoolean");
+ tryPreloadClass("org.apache.commons.lang3.mutable.MutableInt");
+ tryPreloadClass("org.jline.terminal.impl.MouseSupport");
+ tryPreloadClass("org.jline.terminal.impl.MouseSupport$1");
+ tryPreloadClass("org.jline.terminal.Terminal$MouseTracking");
+ tryPreloadClass("co.aikar.timings.TimingHistory");
+ tryPreloadClass("co.aikar.timings.TimingHistory$MinuteReport");
+ tryPreloadClass("io.netty.channel.AbstractChannelHandlerContext");
+ tryPreloadClass("io.netty.channel.AbstractChannelHandlerContext$11");
(1.16.3) Missed some fixes to include in commit
2020-09-10 23:56:58 +00:00
+ tryPreloadClass("io.netty.channel.AbstractChannelHandlerContext$12");
+ tryPreloadClass("io.netty.channel.AbstractChannelHandlerContext$13");
Preload important classes such as Logger and JLine This is for 2 reasons: 1) Ensuring our log4j is mostly loaded at OUR version. I've seen stack traces with line numbers that do not match our version. This means that some plugin has shaded in log4j and their loaded version is mixing with ours.... So by at least trying to load a bunch of log4j classes before we load plugins, we can be more sure mixed versions are not loading. 2) If the jar file is replaced while the server is runnimg class not found errors galore This will preloaod a bunch of classes commonly seen to error during shutdown due to this. The goal here is to help let the server shutdown gracefully as possible. Some plugins will still blow up here if they access a class that hadn't been loaded yet, but goal is to at least stop freezing the shutdown process as it does with JLine and Log4j errors requiring an external kill. Ideally you should not replace jars while the server is running, but it is something that happens in development for testing. Updated test server to do a copy though to avoid this happening in Paper development.
2020-05-24 17:09:02 +00:00
+ tryPreloadClass("io.netty.channel.AbstractChannel$AbstractUnsafe$8");
+ tryPreloadClass("io.netty.util.concurrent.DefaultPromise");
+ tryPreloadClass("io.netty.util.concurrent.DefaultPromise$1");
+ tryPreloadClass("io.netty.util.internal.PromiseNotificationUtil");
+ tryPreloadClass("io.netty.util.internal.SystemPropertyUtil");
+ tryPreloadClass("org.bukkit.craftbukkit.scheduler.CraftScheduler");
+ tryPreloadClass("org.bukkit.craftbukkit.scheduler.CraftScheduler$1");
+ tryPreloadClass("org.bukkit.craftbukkit.scheduler.CraftScheduler$2");
+ tryPreloadClass("org.bukkit.craftbukkit.scheduler.CraftScheduler$3");
+ tryPreloadClass("org.bukkit.craftbukkit.scheduler.CraftScheduler$4");
+ tryPreloadClass("org.slf4j.helpers.MessageFormatter");
+ tryPreloadClass("org.slf4j.helpers.FormattingTuple");
+ tryPreloadClass("org.slf4j.helpers.BasicMarker");
+ tryPreloadClass("org.slf4j.helpers.Util");
Don't unregister the primary dimensions on world unload
2020-05-27 04:15:57 +00:00
+ tryPreloadClass("com.destroystokyo.paper.event.player.PlayerConnectionCloseEvent");
Fix Per World Difficulty / Remembering Difficulty Fixes per world difficulty with /difficulty command and also makes it so that the server keeps the last difficulty used instead of restoring the server.properties every single load.
2020-06-28 08:01:01 +00:00
+ tryPreloadClass("com.destroystokyo.paper.event.entity.EntityRemoveFromWorldEvent");
Preload important classes such as Logger and JLine This is for 2 reasons: 1) Ensuring our log4j is mostly loaded at OUR version. I've seen stack traces with line numbers that do not match our version. This means that some plugin has shaded in log4j and their loaded version is mixing with ours.... So by at least trying to load a bunch of log4j classes before we load plugins, we can be more sure mixed versions are not loading. 2) If the jar file is replaced while the server is runnimg class not found errors galore This will preloaod a bunch of classes commonly seen to error during shutdown due to this. The goal here is to help let the server shutdown gracefully as possible. Some plugins will still blow up here if they access a class that hadn't been loaded yet, but goal is to at least stop freezing the shutdown process as it does with JLine and Log4j errors requiring an external kill. Ideally you should not replace jars while the server is running, but it is something that happens in development for testing. Updated test server to do a copy though to avoid this happening in Paper development.
2020-05-24 17:09:02 +00:00
+ // Minecraft, seen during saving
+ tryPreloadClass("net.minecraft.server.LightEngineLayerEventListener$Void");
+ tryPreloadClass("net.minecraft.server.LightEngineLayerEventListener");
Fix missing chunks on client (View distance issue)
2020-06-28 04:58:25 +00:00
+ tryPreloadClass("net.minecraft.server.ExceptionSuppressor");
Preload important classes such as Logger and JLine This is for 2 reasons: 1) Ensuring our log4j is mostly loaded at OUR version. I've seen stack traces with line numbers that do not match our version. This means that some plugin has shaded in log4j and their loaded version is mixing with ours.... So by at least trying to load a bunch of log4j classes before we load plugins, we can be more sure mixed versions are not loading. 2) If the jar file is replaced while the server is runnimg class not found errors galore This will preloaod a bunch of classes commonly seen to error during shutdown due to this. The goal here is to help let the server shutdown gracefully as possible. Some plugins will still blow up here if they access a class that hadn't been loaded yet, but goal is to at least stop freezing the shutdown process as it does with JLine and Log4j errors requiring an external kill. Ideally you should not replace jars while the server is running, but it is something that happens in development for testing. Updated test server to do a copy though to avoid this happening in Paper development.
2020-05-24 17:09:02 +00:00
+ // Paper end
+ }
+ }
+
+ // Paper start
+ private static void tryPreloadClass(String className) {
+ tryPreloadClass(className, true);
+ }
+ private static void tryPreloadClass(String className, boolean printError) {
+ try {
+ Class.forName(className);
+ } catch (ClassNotFoundException e) {
+ if (printError) System.err.println("An expected class " + className + " was not found for preloading: " + e.getMessage());
}
}
+ // Paper end
private static List<String> asList(String... params) {
return Arrays.asList(params);
Forced Watchdog Crash support and Improve Async Shutdown If the request to shut down the server is received while we are in a watchdog hang, immediately treat it as a crash and begin the shutdown process. Shutdown process is now improved to also shutdown cleanly when not using restart scripts either. If a server is deadlocked, a server owner can send SIGHUP (or any other signal the JVM understands to shut down as it currently does) and the watchdog will no longer need to wait until the full timeout, allowing you to trigger a close process and try to shut the server down gracefully, saving player and world data. Previously there was no way to trigger this outside of waiting for a full watchdog timeout, which may be set to a really long time... Additionally, fix everything to do with shutting the server down asynchronously. Previously, nearly everything about the process was fragile and unsafe. Main might not have actually been frozen, and might still be manipulating state. Or, some reuest might ask main to do something in the shutdown but main is dead. Or worse, other things might start closing down items such as the Console or Thread Pool before we are fully shutdown. This change tries to resolve all of these issues by moving everything into the stop method and guaranteeing only one thread is stopping the server. We then issue Thread Death to the main thread of another thread initiates the stop process. We have to ensure Thread Death propagates correctly though to stop main completely. This is to ensure that if main isn't truely stuck, it's not manipulating state we are trying to save.
2020-04-16 08:53:50 +00:00
diff --git a/src/main/java/org/bukkit/craftbukkit/util/ServerShutdownThread.java b/src/main/java/org/bukkit/craftbukkit/util/ServerShutdownThread.java
Fix issues with SIGTERM not shutting down gracefully - Fixes #3062 We had 2 issues. 1) Log4J2 Shutdown hook seemed to be causing issues as it shutdown logger while we still needed it 2) ServerShutdownThread needs to stay alive until server is shutdown to keep jvm open. It appears SIGINT is handled differently than SIGTERM, as SIGINT worked correctly. But this will make both methods work.
2020-06-11 03:40:28 +00:00
index 449e99d1b673870ed6892f6ab2c715a2db35c35d..c7ed6e0f8a989cec97700df2b15198c9c481c549 100644
Forced Watchdog Crash support and Improve Async Shutdown If the request to shut down the server is received while we are in a watchdog hang, immediately treat it as a crash and begin the shutdown process. Shutdown process is now improved to also shutdown cleanly when not using restart scripts either. If a server is deadlocked, a server owner can send SIGHUP (or any other signal the JVM understands to shut down as it currently does) and the watchdog will no longer need to wait until the full timeout, allowing you to trigger a close process and try to shut the server down gracefully, saving player and world data. Previously there was no way to trigger this outside of waiting for a full watchdog timeout, which may be set to a really long time... Additionally, fix everything to do with shutting the server down asynchronously. Previously, nearly everything about the process was fragile and unsafe. Main might not have actually been frozen, and might still be manipulating state. Or, some reuest might ask main to do something in the shutdown but main is dead. Or worse, other things might start closing down items such as the Console or Thread Pool before we are fully shutdown. This change tries to resolve all of these issues by moving everything into the stop method and guaranteeing only one thread is stopping the server. We then issue Thread Death to the main thread of another thread initiates the stop process. We have to ensure Thread Death propagates correctly though to stop main completely. This is to ensure that if main isn't truely stuck, it's not manipulating state we are trying to save.
2020-04-16 08:53:50 +00:00
--- a/src/main/java/org/bukkit/craftbukkit/util/ServerShutdownThread.java
+++ b/src/main/java/org/bukkit/craftbukkit/util/ServerShutdownThread.java
Fix issues with SIGTERM not shutting down gracefully - Fixes #3062 We had 2 issues. 1) Log4J2 Shutdown hook seemed to be causing issues as it shutdown logger while we still needed it 2) ServerShutdownThread needs to stay alive until server is shutdown to keep jvm open. It appears SIGINT is handled differently than SIGTERM, as SIGINT worked correctly. But this will make both methods work.
2020-06-11 03:40:28 +00:00
@@ -12,12 +12,27 @@ public class ServerShutdownThread extends Thread {
Forced Watchdog Crash support and Improve Async Shutdown If the request to shut down the server is received while we are in a watchdog hang, immediately treat it as a crash and begin the shutdown process. Shutdown process is now improved to also shutdown cleanly when not using restart scripts either. If a server is deadlocked, a server owner can send SIGHUP (or any other signal the JVM understands to shut down as it currently does) and the watchdog will no longer need to wait until the full timeout, allowing you to trigger a close process and try to shut the server down gracefully, saving player and world data. Previously there was no way to trigger this outside of waiting for a full watchdog timeout, which may be set to a really long time... Additionally, fix everything to do with shutting the server down asynchronously. Previously, nearly everything about the process was fragile and unsafe. Main might not have actually been frozen, and might still be manipulating state. Or, some reuest might ask main to do something in the shutdown but main is dead. Or worse, other things might start closing down items such as the Console or Thread Pool before we are fully shutdown. This change tries to resolve all of these issues by moving everything into the stop method and guaranteeing only one thread is stopping the server. We then issue Thread Death to the main thread of another thread initiates the stop process. We have to ensure Thread Death propagates correctly though to stop main completely. This is to ensure that if main isn't truely stuck, it's not manipulating state we are trying to save.
2020-04-16 08:53:50 +00:00
@Override
public void run() {
try {
+ // Paper start - try to shutdown on main
+ server.safeShutdown(false, false);
+ for (int i = 1000; i > 0 && !server.hasStopped(); i -= 100) {
+ Thread.sleep(100);
+ }
+ if (server.hasStopped()) {
Fix issues with SIGTERM not shutting down gracefully - Fixes #3062 We had 2 issues. 1) Log4J2 Shutdown hook seemed to be causing issues as it shutdown logger while we still needed it 2) ServerShutdownThread needs to stay alive until server is shutdown to keep jvm open. It appears SIGINT is handled differently than SIGTERM, as SIGINT worked correctly. But this will make both methods work.
2020-06-11 03:40:28 +00:00
+ while (!server.hasFullyShutdown) Thread.sleep(1000);
Forced Watchdog Crash support and Improve Async Shutdown If the request to shut down the server is received while we are in a watchdog hang, immediately treat it as a crash and begin the shutdown process. Shutdown process is now improved to also shutdown cleanly when not using restart scripts either. If a server is deadlocked, a server owner can send SIGHUP (or any other signal the JVM understands to shut down as it currently does) and the watchdog will no longer need to wait until the full timeout, allowing you to trigger a close process and try to shut the server down gracefully, saving player and world data. Previously there was no way to trigger this outside of waiting for a full watchdog timeout, which may be set to a really long time... Additionally, fix everything to do with shutting the server down asynchronously. Previously, nearly everything about the process was fragile and unsafe. Main might not have actually been frozen, and might still be manipulating state. Or, some reuest might ask main to do something in the shutdown but main is dead. Or worse, other things might start closing down items such as the Console or Thread Pool before we are fully shutdown. This change tries to resolve all of these issues by moving everything into the stop method and guaranteeing only one thread is stopping the server. We then issue Thread Death to the main thread of another thread initiates the stop process. We have to ensure Thread Death propagates correctly though to stop main completely. This is to ensure that if main isn't truely stuck, it's not manipulating state we are trying to save.
2020-04-16 08:53:50 +00:00
+ return;
+ }
+ // Looks stalled, close async
org.spigotmc.AsyncCatcher.enabled = false; // Spigot
org.spigotmc.AsyncCatcher.shuttingDown = true; // Paper
+ server.forceTicks = true;
server.close();
Fix issues with SIGTERM not shutting down gracefully - Fixes #3062 We had 2 issues. 1) Log4J2 Shutdown hook seemed to be causing issues as it shutdown logger while we still needed it 2) ServerShutdownThread needs to stay alive until server is shutdown to keep jvm open. It appears SIGINT is handled differently than SIGTERM, as SIGINT worked correctly. But this will make both methods work.
2020-06-11 03:40:28 +00:00
+ while (!server.hasFullyShutdown) Thread.sleep(1000);
Forced Watchdog Crash support and Improve Async Shutdown If the request to shut down the server is received while we are in a watchdog hang, immediately treat it as a crash and begin the shutdown process. Shutdown process is now improved to also shutdown cleanly when not using restart scripts either. If a server is deadlocked, a server owner can send SIGHUP (or any other signal the JVM understands to shut down as it currently does) and the watchdog will no longer need to wait until the full timeout, allowing you to trigger a close process and try to shut the server down gracefully, saving player and world data. Previously there was no way to trigger this outside of waiting for a full watchdog timeout, which may be set to a really long time... Additionally, fix everything to do with shutting the server down asynchronously. Previously, nearly everything about the process was fragile and unsafe. Main might not have actually been frozen, and might still be manipulating state. Or, some reuest might ask main to do something in the shutdown but main is dead. Or worse, other things might start closing down items such as the Console or Thread Pool before we are fully shutdown. This change tries to resolve all of these issues by moving everything into the stop method and guaranteeing only one thread is stopping the server. We then issue Thread Death to the main thread of another thread initiates the stop process. We have to ensure Thread Death propagates correctly though to stop main completely. This is to ensure that if main isn't truely stuck, it's not manipulating state we are trying to save.
2020-04-16 08:53:50 +00:00
+ } catch (InterruptedException e) {
+ e.printStackTrace();
+ // Paper end
} finally {
try {
- net.minecrell.terminalconsole.TerminalConsoleAppender.close(); // Paper - Use TerminalConsoleAppender
+ //net.minecrell.terminalconsole.TerminalConsoleAppender.close(); // Paper - Move into stop
} catch (Exception e) {
}
}
diff --git a/src/main/java/org/spigotmc/RestartCommand.java b/src/main/java/org/spigotmc/RestartCommand.java
Rebuild all patches using the new rebuild pattern
2020-05-06 09:48:49 +00:00
index aefea3a9a8b9b75c62bd20018be7cd166a213001..123de5ac9026508e21cdc225f0962f5c3c46fed5 100644
Forced Watchdog Crash support and Improve Async Shutdown If the request to shut down the server is received while we are in a watchdog hang, immediately treat it as a crash and begin the shutdown process. Shutdown process is now improved to also shutdown cleanly when not using restart scripts either. If a server is deadlocked, a server owner can send SIGHUP (or any other signal the JVM understands to shut down as it currently does) and the watchdog will no longer need to wait until the full timeout, allowing you to trigger a close process and try to shut the server down gracefully, saving player and world data. Previously there was no way to trigger this outside of waiting for a full watchdog timeout, which may be set to a really long time... Additionally, fix everything to do with shutting the server down asynchronously. Previously, nearly everything about the process was fragile and unsafe. Main might not have actually been frozen, and might still be manipulating state. Or, some reuest might ask main to do something in the shutdown but main is dead. Or worse, other things might start closing down items such as the Console or Thread Pool before we are fully shutdown. This change tries to resolve all of these issues by moving everything into the stop method and guaranteeing only one thread is stopping the server. We then issue Thread Death to the main thread of another thread initiates the stop process. We have to ensure Thread Death propagates correctly though to stop main completely. This is to ensure that if main isn't truely stuck, it's not manipulating state we are trying to save.
2020-04-16 08:53:50 +00:00
--- a/src/main/java/org/spigotmc/RestartCommand.java
+++ b/src/main/java/org/spigotmc/RestartCommand.java
@@ -139,7 +139,7 @@ public class RestartCommand extends Command
// Paper end
// Paper start - copied from above and modified to return if the hook registered
- private static boolean addShutdownHook(String restartScript)
+ public static boolean addShutdownHook(String restartScript)
{
String[] split = restartScript.split( " " );
if ( split.length > 0 && new File( split[0] ).isFile() )
diff --git a/src/main/java/org/spigotmc/WatchdogThread.java b/src/main/java/org/spigotmc/WatchdogThread.java
Many improvements to chunk prioritization and bug fixes Fixed a few bugs, and made numerous improvements. Fixed issue where a sync chunk load could have its ticket removed and the priority ticket could expire... Still not perfect there but better than before. Also fixed few other misc issues such as watchdog cpu usage, chunk queue update had risk of double enqueue due to it no longer being a set. Added much more information about chunk state to watchdog prints. I see some more room for improvement even, but this is much better than before. Fixes #3407 Fixes #3411 Fixes #3395 Fixes #3389
2020-05-20 09:11:57 +00:00
index 5bdcdcf9e85b73086722783bff26321d03382bb9..513c1041c34ebb3ac1775674a3f4526693759c08 100644
Forced Watchdog Crash support and Improve Async Shutdown If the request to shut down the server is received while we are in a watchdog hang, immediately treat it as a crash and begin the shutdown process. Shutdown process is now improved to also shutdown cleanly when not using restart scripts either. If a server is deadlocked, a server owner can send SIGHUP (or any other signal the JVM understands to shut down as it currently does) and the watchdog will no longer need to wait until the full timeout, allowing you to trigger a close process and try to shut the server down gracefully, saving player and world data. Previously there was no way to trigger this outside of waiting for a full watchdog timeout, which may be set to a really long time... Additionally, fix everything to do with shutting the server down asynchronously. Previously, nearly everything about the process was fragile and unsafe. Main might not have actually been frozen, and might still be manipulating state. Or, some reuest might ask main to do something in the shutdown but main is dead. Or worse, other things might start closing down items such as the Console or Thread Pool before we are fully shutdown. This change tries to resolve all of these issues by moving everything into the stop method and guaranteeing only one thread is stopping the server. We then issue Thread Death to the main thread of another thread initiates the stop process. We have to ensure Thread Death propagates correctly though to stop main completely. This is to ensure that if main isn't truely stuck, it's not manipulating state we are trying to save.
2020-04-16 08:53:50 +00:00
--- a/src/main/java/org/spigotmc/WatchdogThread.java
+++ b/src/main/java/org/spigotmc/WatchdogThread.java
Many improvements to chunk prioritization and bug fixes Fixed a few bugs, and made numerous improvements. Fixed issue where a sync chunk load could have its ticket removed and the priority ticket could expire... Still not perfect there but better than before. Also fixed few other misc issues such as watchdog cpu usage, chunk queue update had risk of double enqueue due to it no longer being a set. Added much more information about chunk state to watchdog prints. I see some more room for improvement even, but this is much better than before. Fixes #3407 Fixes #3411 Fixes #3395 Fixes #3389
2020-05-20 09:11:57 +00:00
@@ -13,6 +13,7 @@ import org.bukkit.Bukkit;
public class WatchdogThread extends Thread
{
+ public static final boolean DISABLE_WATCHDOG = Boolean.getBoolean("disable.watchdog"); // Paper
private static WatchdogThread instance;
private final long timeoutTime;
private final long earlyWarningEvery; // Paper - Timeout time for just printing a dump but not restarting
@@ -41,6 +42,7 @@ public class WatchdogThread extends Thread
Improvements to watchdog changes catch a thread death, fix invalid watchdog timeouts <= 0 seconds will now change to 5 minutes.
2020-04-24 04:27:28 +00:00
{
if ( instance == null )
{
+ if (timeoutTime <= 0) timeoutTime = 300; // Paper
instance = new WatchdogThread( timeoutTime * 1000L, restart );
instance.start();
}
Many improvements to chunk prioritization and bug fixes Fixed a few bugs, and made numerous improvements. Fixed issue where a sync chunk load could have its ticket removed and the priority ticket could expire... Still not perfect there but better than before. Also fixed few other misc issues such as watchdog cpu usage, chunk queue update had risk of double enqueue due to it no longer being a set. Added much more information about chunk state to watchdog prints. I see some more room for improvement even, but this is much better than before. Fixes #3407 Fixes #3411 Fixes #3395 Fixes #3389
2020-05-20 09:11:57 +00:00
@@ -67,12 +69,13 @@ public class WatchdogThread extends Thread
Forced Watchdog Crash support and Improve Async Shutdown If the request to shut down the server is received while we are in a watchdog hang, immediately treat it as a crash and begin the shutdown process. Shutdown process is now improved to also shutdown cleanly when not using restart scripts either. If a server is deadlocked, a server owner can send SIGHUP (or any other signal the JVM understands to shut down as it currently does) and the watchdog will no longer need to wait until the full timeout, allowing you to trigger a close process and try to shut the server down gracefully, saving player and world data. Previously there was no way to trigger this outside of waiting for a full watchdog timeout, which may be set to a really long time... Additionally, fix everything to do with shutting the server down asynchronously. Previously, nearly everything about the process was fragile and unsafe. Main might not have actually been frozen, and might still be manipulating state. Or, some reuest might ask main to do something in the shutdown but main is dead. Or worse, other things might start closing down items such as the Console or Thread Pool before we are fully shutdown. This change tries to resolve all of these issues by moving everything into the stop method and guaranteeing only one thread is stopping the server. We then issue Thread Death to the main thread of another thread initiates the stop process. We have to ensure Thread Death propagates correctly though to stop main completely. This is to ensure that if main isn't truely stuck, it's not manipulating state we are trying to save.
2020-04-16 08:53:50 +00:00
// Paper start
Logger log = Bukkit.getServer().getLogger();
long currentTime = monotonicMillis();
- if ( lastTick != 0 && currentTime > lastTick + earlyWarningEvery && !Boolean.getBoolean("disable.watchdog") )
+ MinecraftServer server = MinecraftServer.getServer();
Many improvements to chunk prioritization and bug fixes Fixed a few bugs, and made numerous improvements. Fixed issue where a sync chunk load could have its ticket removed and the priority ticket could expire... Still not perfect there but better than before. Also fixed few other misc issues such as watchdog cpu usage, chunk queue update had risk of double enqueue due to it no longer being a set. Added much more information about chunk state to watchdog prints. I see some more room for improvement even, but this is much better than before. Fixes #3407 Fixes #3411 Fixes #3395 Fixes #3389
2020-05-20 09:11:57 +00:00
+ if (lastTick != 0 && hasStarted && (!server.isRunning() || (currentTime > lastTick + earlyWarningEvery && !DISABLE_WATCHDOG) ))
Forced Watchdog Crash support and Improve Async Shutdown If the request to shut down the server is received while we are in a watchdog hang, immediately treat it as a crash and begin the shutdown process. Shutdown process is now improved to also shutdown cleanly when not using restart scripts either. If a server is deadlocked, a server owner can send SIGHUP (or any other signal the JVM understands to shut down as it currently does) and the watchdog will no longer need to wait until the full timeout, allowing you to trigger a close process and try to shut the server down gracefully, saving player and world data. Previously there was no way to trigger this outside of waiting for a full watchdog timeout, which may be set to a really long time... Additionally, fix everything to do with shutting the server down asynchronously. Previously, nearly everything about the process was fragile and unsafe. Main might not have actually been frozen, and might still be manipulating state. Or, some reuest might ask main to do something in the shutdown but main is dead. Or worse, other things might start closing down items such as the Console or Thread Pool before we are fully shutdown. This change tries to resolve all of these issues by moving everything into the stop method and guaranteeing only one thread is stopping the server. We then issue Thread Death to the main thread of another thread initiates the stop process. We have to ensure Thread Death propagates correctly though to stop main completely. This is to ensure that if main isn't truely stuck, it's not manipulating state we are trying to save.
2020-04-16 08:53:50 +00:00
{
- boolean isLongTimeout = currentTime > lastTick + timeoutTime;
+ boolean isLongTimeout = currentTime > lastTick + timeoutTime || (!server.isRunning() && !server.hasStopped() && currentTime > lastTick + 1000);
// 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...
+ if ( !isLongTimeout && server.hasStopped()) continue; // Don't spam early watchdog warnings during shutdown, we'll come back to this...
lastEarlyWarning = currentTime;
if (isLongTimeout) {
// Paper end
Many improvements to chunk prioritization and bug fixes Fixed a few bugs, and made numerous improvements. Fixed issue where a sync chunk load could have its ticket removed and the priority ticket could expire... Still not perfect there but better than before. Also fixed few other misc issues such as watchdog cpu usage, chunk queue update had risk of double enqueue due to it no longer being a set. Added much more information about chunk state to watchdog prints. I see some more room for improvement even, but this is much better than before. Fixes #3407 Fixes #3411 Fixes #3395 Fixes #3389
2020-05-20 09:11:57 +00:00
@@ -114,7 +117,7 @@ public class WatchdogThread extends Thread
Forced Watchdog Crash support and Improve Async Shutdown If the request to shut down the server is received while we are in a watchdog hang, immediately treat it as a crash and begin the shutdown process. Shutdown process is now improved to also shutdown cleanly when not using restart scripts either. If a server is deadlocked, a server owner can send SIGHUP (or any other signal the JVM understands to shut down as it currently does) and the watchdog will no longer need to wait until the full timeout, allowing you to trigger a close process and try to shut the server down gracefully, saving player and world data. Previously there was no way to trigger this outside of waiting for a full watchdog timeout, which may be set to a really long time... Additionally, fix everything to do with shutting the server down asynchronously. Previously, nearly everything about the process was fragile and unsafe. Main might not have actually been frozen, and might still be manipulating state. Or, some reuest might ask main to do something in the shutdown but main is dead. Or worse, other things might start closing down items such as the Console or Thread Pool before we are fully shutdown. This change tries to resolve all of these issues by moving everything into the stop method and guaranteeing only one thread is stopping the server. We then issue Thread Death to the main thread of another thread initiates the stop process. We have to ensure Thread Death propagates correctly though to stop main completely. This is to ensure that if main isn't truely stuck, it's not manipulating state we are trying to save.
2020-04-16 08:53:50 +00:00
log.log( Level.SEVERE, "------------------------------" );
log.log( Level.SEVERE, "Server thread dump (Look for plugins here before reporting to Paper!):" ); // Paper
ChunkTaskManager.dumpAllChunkLoadInfo(); // Paper
- dumpThread( ManagementFactory.getThreadMXBean().getThreadInfo( MinecraftServer.getServer().serverThread.getId(), Integer.MAX_VALUE ), log );
+ dumpThread( ManagementFactory.getThreadMXBean().getThreadInfo( server.serverThread.getId(), Integer.MAX_VALUE ), log );
log.log( Level.SEVERE, "------------------------------" );
//
// Paper start - Only print full dump on long timeouts
Many improvements to chunk prioritization and bug fixes Fixed a few bugs, and made numerous improvements. Fixed issue where a sync chunk load could have its ticket removed and the priority ticket could expire... Still not perfect there but better than before. Also fixed few other misc issues such as watchdog cpu usage, chunk queue update had risk of double enqueue due to it no longer being a set. Added much more information about chunk state to watchdog prints. I see some more room for improvement even, but this is much better than before. Fixes #3407 Fixes #3411 Fixes #3395 Fixes #3389
2020-05-20 09:11:57 +00:00
@@ -135,9 +138,24 @@ public class WatchdogThread extends Thread
Forced Watchdog Crash support and Improve Async Shutdown If the request to shut down the server is received while we are in a watchdog hang, immediately treat it as a crash and begin the shutdown process. Shutdown process is now improved to also shutdown cleanly when not using restart scripts either. If a server is deadlocked, a server owner can send SIGHUP (or any other signal the JVM understands to shut down as it currently does) and the watchdog will no longer need to wait until the full timeout, allowing you to trigger a close process and try to shut the server down gracefully, saving player and world data. Previously there was no way to trigger this outside of waiting for a full watchdog timeout, which may be set to a really long time... Additionally, fix everything to do with shutting the server down asynchronously. Previously, nearly everything about the process was fragile and unsafe. Main might not have actually been frozen, and might still be manipulating state. Or, some reuest might ask main to do something in the shutdown but main is dead. Or worse, other things might start closing down items such as the Console or Thread Pool before we are fully shutdown. This change tries to resolve all of these issues by moving everything into the stop method and guaranteeing only one thread is stopping the server. We then issue Thread Death to the main thread of another thread initiates the stop process. We have to ensure Thread Death propagates correctly though to stop main completely. This is to ensure that if main isn't truely stuck, it's not manipulating state we are trying to save.
2020-04-16 08:53:50 +00:00
if ( isLongTimeout )
{
- if ( restart && !MinecraftServer.getServer().hasStopped() )
+ if ( !server.hasStopped() )
{
- RestartCommand.restart();
+ AsyncCatcher.enabled = false; // Disable async catcher incase it interferes with us
+ AsyncCatcher.shuttingDown = true;
+ server.forceTicks = true;
+ if (restart) {
+ RestartCommand.addShutdownHook( SpigotConfig.restartScript );
+ }
+ // try one last chance to safe shutdown on main incase it 'comes back'
+ server.safeShutdown(false, restart);
+ try {
+ Thread.sleep(1000);
+ } catch (InterruptedException e) {
+ e.printStackTrace();
+ }
+ if (!server.hasStopped()) {
+ server.close();
+ }
}
break;
} // Paper end
Fix issues with SIGTERM not shutting down gracefully - Fixes #3062 We had 2 issues. 1) Log4J2 Shutdown hook seemed to be causing issues as it shutdown logger while we still needed it 2) ServerShutdownThread needs to stay alive until server is shutdown to keep jvm open. It appears SIGINT is handled differently than SIGTERM, as SIGINT worked correctly. But this will make both methods work.
2020-06-11 03:40:28 +00:00
diff --git a/src/main/resources/log4j2.xml b/src/main/resources/log4j2.xml
index 476f4a5cbe664ddd05474cb88553018bd334a5b8..8af159abd3d0cc94cf155fec5b384c42f69551bf 100644
--- a/src/main/resources/log4j2.xml
+++ b/src/main/resources/log4j2.xml
@@ -1,5 +1,5 @@
<?xml version="1.0" encoding="UTF-8"?>
-<Configuration status="WARN" packages="com.mojang.util">
+<Configuration status="WARN" packages="com.mojang.util" shutdownHook="disable">
<Appenders>
<Queue name="ServerGuiConsole">
<PatternLayout pattern="[%d{HH:mm:ss} %level]: %msg%n" />