You are viewing a plain text version of this content. The canonical link for it is here.
Posted to notifications@seata.apache.org by "john1337 (via GitHub)" <gi...@apache.org> on 2025/12/04 07:53:06 UTC
[I] seata raft模式下500并发时GlobalBeginRequest、GlobalCommitRequest耗时100-500多毫秒 [incubator-seata]
john1337 opened a new issue, #7835:
URL: https://github.com/apache/incubator-seata/issues/7835
### Check Ahead
- [x] I have searched the [issues](https://github.com/seata/seata/issues) of this repository and believe that this is not a duplicate.
- [ ] I am willing to try to implement this feature myself.
### Why you need it?
按照seata官网部署了seata raft高可用集群，并发量为500TPS，发现GlobalBeginRequest、GlobalCommitRequest这两个阶段耗时很长，长达100-500多毫秒，seata-server给到的8C/8G 硬盘为ssd，
max-append-bufferSize: 524288 ，比官网增加1倍
disruptor-buffer-size: 32768 比官网大1倍
不知还可以调整哪些哪些参数，降低请求处理时长
### How it could be?
_No response_
### Other related information
_No response_
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: notifications-unsubscribe@seata.apache.org.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
---------------------------------------------------------------------
To unsubscribe, e-mail: notifications-unsubscribe@seata.apache.org
For additional commands, e-mail: notifications-help@seata.apache.org
Re: [I] seata raft模式下500并发时GlobalBeginRequest、GlobalCommitRequest耗时100-500多毫秒 [incubator-seata]
Posted by "john1337 (via GitHub)" <gi...@apache.org>.
john1337 commented on issue #7835:
URL: https://github.com/apache/incubator-seata/issues/7835#issuecomment-3619258356
还有一个就是我目前用的是AT模式，这种模式在高并发下性能比TCC模式要差不少吧，我准备换成TCC模式验证下
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: notifications-unsubscribe@seata.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
---------------------------------------------------------------------
To unsubscribe, e-mail: notifications-unsubscribe@seata.apache.org
For additional commands, e-mail: notifications-help@seata.apache.org
Re: [I] seata raft模式下500并发时GlobalBeginRequest、GlobalCommitRequest耗时100-500多毫秒 [incubator-seata]
Posted by "john1337 (via GitHub)" <gi...@apache.org>.
john1337 commented on issue #7835:
URL: https://github.com/apache/incubator-seata/issues/7835#issuecomment-3611623925
arthas追踪了下seata server端commit部分代码，结果如下：
trace org.apache.seata.server.coordinator.DefaultCore commit '#cost>200'
`---ts=2025-12-04 19:04:40.284;thread_name=ServerHandlerThread_1_20_500;id=151;is_daemon=true;priority=5;TCCL=jdk.internal.loader.ClassLoaders$AppClassLoader@6438a396
`---[232.067017ms] org.apache.seata.server.coordinator.DefaultCore:commit()
+---[2.07% 4.79429ms ] org.apache.seata.server.session.SessionHolder:findGlobalSession() #238
+---[0.00% 0.00749ms ] org.apache.seata.server.session.GlobalSession:isTimeout() #243
+---[91.55% 212.46275ms ] org.apache.seata.server.session.SessionHolder:lockAndExecute() #249
+---[3.57% 8.290511ms ] org.apache.seata.server.coordinator.DefaultCore:doGlobalCommit() #268
+---[0.00% 0.00624ms ] org.apache.seata.server.session.GlobalSession:hasBranch() #270
+---[0.00% 0.00512ms ] org.apache.seata.server.session.GlobalSession:canBeCommittedAsync() #270
`---[2.74% 6.361747ms ] org.apache.seata.server.session.GlobalSession:asyncCommit() #271
`---ts=2025-12-04 19:04:42.540;thread_name=ServerHandlerThread_1_47_500;id=185;is_daemon=true;priority=5;TCCL=jdk.internal.loader.ClassLoaders$AppClassLoader@6438a396
`---[230.315432ms] org.apache.seata.server.coordinator.DefaultCore:commit()
+---[2.93% 6.756012ms ] org.apache.seata.server.session.SessionHolder:findGlobalSession() #238
+---[0.00% 0.00296ms ] org.apache.seata.server.session.GlobalSession:isTimeout() #243
+---[93.30% 214.890715ms ] org.apache.seata.server.session.SessionHolder:lockAndExecute() #249
+---[1.64% 3.782465ms ] org.apache.seata.server.coordinator.DefaultCore:doGlobalCommit() #268
+---[0.00% 0.00142ms ] org.apache.seata.server.session.GlobalSession:hasBranch() #270
+---[0.00% 0.00121ms ] org.apache.seata.server.session.GlobalSession:canBeCommittedAsync() #270
`---[2.08% 4.79145ms ] org.apache.seata.server.session.GlobalSession:asyncCommit() #271
`---ts=2025-12-04 19:04:42.543;thread_name=ServerHandlerThread_1_1_500;id=87;is_daemon=true;priority=5;TCCL=jdk.internal.loader.ClassLoaders$AppClassLoader@6438a396
`---[229.206927ms] org.apache.seata.server.coordinator.DefaultCore:commit()
+---[1.96% 4.492525ms ] org.apache.seata.server.session.SessionHolder:findGlobalSession() #238
+---[0.00% 0.00227ms ] org.apache.seata.server.session.GlobalSession:isTimeout() #243
+---[93.86% 215.129822ms ] org.apache.seata.server.session.SessionHolder:lockAndExecute() #249
+---[1.73% 3.967832ms ] org.apache.seata.server.coordinator.DefaultCore:doGlobalCommit() #268
+---[0.00% 0.00152ms ] org.apache.seata.server.session.GlobalSession:hasBranch() #270
+---[0.00% 0.0012ms ] org.apache.seata.server.session.GlobalSession:canBeCommittedAsync() #270
`---[2.42% 5.55572ms ] org.apache.seata.server.session.GlobalSession:asyncCommit() #271
`---ts=2025-12-04 19:04:42.734;thread_name=ServerHandlerThread_1_31_500;id=164;is_daemon=true;priority=5;TCCL=jdk.internal.loader.ClassLoaders$AppClassLoader@6438a396
`---[227.575881ms] org.apache.seata.server.coordinator.DefaultCore:commit()
+---[2.07% 4.709012ms ] org.apache.seata.server.session.SessionHolder:findGlobalSession() #238
+---[0.00% 0.0019ms ] org.apache.seata.server.session.GlobalSession:isTimeout() #243
+---[94.53% 215.136442ms ] org.apache.seata.server.session.SessionHolder:lockAndExecute() #249
+---[1.61% 3.671706ms ] org.apache.seata.server.coordinator.DefaultCore:doGlobalCommit() #268
+---[0.00% 0.00192ms ] org.apache.seata.server.session.GlobalSession:hasBranch() #270
+---[0.00% 9.8E-4ms ] org.apache.seata.server.session.GlobalSession:canBeCommittedAsync() #270
`---[1.76% 3.994172ms ] org.apache.seata.server.session.GlobalSession:asyncCommit() #271
`---ts=2025-12-04 19:04:42.707;thread_name=ServerHandlerThread_1_49_500;id=187;is_daemon=true;priority=5;TCCL=jdk.internal.loader.ClassLoaders$AppClassLoader@6438a396
`---[303.130777ms] org.apache.seata.server.coordinator.DefaultCore:commit()
+---[1.17% 3.55253ms ] org.apache.seata.server.session.SessionHolder:findGlobalSession() #238
+---[0.00% 0.002069ms ] org.apache.seata.server.session.GlobalSession:isTimeout() #243
+---[2.29% 6.932369ms ] org.apache.seata.server.session.SessionHolder:lockAndExecute() #249
+---[94.23% 285.627591ms ] org.apache.seata.server.coordinator.DefaultCore:doGlobalCommit() #268
+---[0.00% 0.00183ms ] org.apache.seata.server.session.GlobalSession:hasBranch() #270
+---[0.00% 0.00113ms ] org.apache.seata.server.session.GlobalSession:canBeCommittedAsync() #270
`---[2.29% 6.949568ms ] org.apache.seata.server.session.GlobalSession:asyncCommit() #271
`---ts=2025-12-04 19:04:42.756;thread_name=ServerHandlerThread_1_43_500;id=179;is_daemon=true;priority=5;TCCL=jdk.internal.loader.ClassLoaders$AppClassLoader@6438a396
`---[257.186733ms] org.apache.seata.server.coordinator.DefaultCore:commit()
+---[1.22% 3.129915ms ] org.apache.seata.server.session.SessionHolder:findGlobalSession() #238
+---[0.00% 0.0032ms ] org.apache.seata.server.session.GlobalSession:isTimeout() #243
+---[2.69% 6.917289ms ] org.apache.seata.server.session.SessionHolder:lockAndExecute() #249
+---[93.15% 239.573008ms ] org.apache.seata.server.coordinator.DefaultCore:doGlobalCommit() #268
+---[0.00% 0.00193ms ] org.apache.seata.server.session.GlobalSession:hasBranch() #270
+---[0.00% 0.00128ms ] org.apache.seata.server.session.GlobalSession:canBeCommittedAsync() #270
`---[2.91% 7.472612ms ] org.apache.seata.server.session.GlobalSession:asyncCommit() #271
`---ts=2025-12-04 19:04:42.739;thread_name=ServerHandlerThread_1_48_500;id=186;is_daemon=true;priority=5;TCCL=jdk.internal.loader.ClassLoaders$AppClassLoader@6438a396
`---[274.188536ms] org.apache.seata.server.coordinator.DefaultCore:commit()
+---[77.30% 211.958157ms ] org.apache.seata.server.session.SessionHolder:findGlobalSession() #238
+---[0.00% 0.00192ms ] org.apache.seata.server.session.GlobalSession:isTimeout() #243
+---[1.88% 5.144406ms ] org.apache.seata.server.session.SessionHolder:lockAndExecute() #249
+---[17.95% 49.213956ms ] org.apache.seata.server.coordinator.DefaultCore:doGlobalCommit() #268
+---[0.01% 0.016399ms ] org.apache.seata.server.session.GlobalSession:hasBranch() #270
+---[0.00% 0.00118ms ] org.apache.seata.server.session.GlobalSession:canBeCommittedAsync() #270
`---[2.84% 7.776698ms ] org.apache.seata.server.session.GlobalSession:asyncCommit() #271
`---ts=2025-12-04 19:04:43.362;thread_name=ServerHandlerThread_1_19_500;id=150;is_daemon=true;priority=5;TCCL=jdk.internal.loader.ClassLoaders$AppClassLoader@6438a396
`---[248.884783ms] org.apache.seata.server.coordinator.DefaultCore:commit()
+---[1.47% 3.667406ms ] org.apache.seata.server.session.SessionHolder:findGlobalSession() #238
+---[0.00% 0.00301ms ] org.apache.seata.server.session.GlobalSession:isTimeout() #243
+---[2.64% 6.582485ms ] org.apache.seata.server.session.SessionHolder:lockAndExecute() #249
+---[9.12% 22.687412ms ] org.apache.seata.server.coordinator.DefaultCore:doGlobalCommit() #268
+---[0.00% 0.003039ms ] org.apache.seata.server.session.GlobalSession:hasBranch() #270
+---[0.00% 0.00182ms ] org.apache.seata.server.session.GlobalSession:canBeCommittedAsync() #270
`---[86.72% 215.838152ms ] org.apache.seata.server.session.GlobalSession:asyncCommit() #271
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: notifications-unsubscribe@seata.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
---------------------------------------------------------------------
To unsubscribe, e-mail: notifications-unsubscribe@seata.apache.org
For additional commands, e-mail: notifications-help@seata.apache.org
Re: [I] seata raft模式下500并发时GlobalBeginRequest、GlobalCommitRequest耗时100-500多毫秒 [incubator-seata]
Posted by "john1337 (via GitHub)" <gi...@apache.org>.
john1337 commented on issue #7835:
URL: https://github.com/apache/incubator-seata/issues/7835#issuecomment-3615829123
GlobalBeginRequest 与GlobalCommitRequest消息的处理最终都会路由到AbstractNettyRemoting类的processMessage方法处理，
而这个方法的执行最终是由一个传统线程池处理掉的，这个线程池的配置默认是corePoolSize为50，maximumPoolSize为500，但是 queue队列长度长达20000，也就是说如果如果有些接口处理超时的话可能会导致后续所有GlobalBeginRequest 等类似请求处理进入线程池等待队列中，这个就会增加对应请求处理时长，所以我准备增加corePoolSize、适当减少queue队列长度，GlobalBeginRequest 实际上也会涉及到写redis、或者本地file的操作（这些写操作也会加锁，所以如果操作处理慢也会增加锁的时长） ，我考虑将Server start方法中的线程池更换成jdk21的虚拟线程，这个是不是更好，不知有没有什么隐患 @funky-eyes
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: notifications-unsubscribe@seata.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
---------------------------------------------------------------------
To unsubscribe, e-mail: notifications-unsubscribe@seata.apache.org
For additional commands, e-mail: notifications-help@seata.apache.org
Re: [I] seata raft模式下500并发时GlobalBeginRequest、GlobalCommitRequest耗时100-500多毫秒 [incubator-seata]
Posted by "john1337 (via GitHub)" <gi...@apache.org>.
john1337 commented on issue #7835:
URL: https://github.com/apache/incubator-seata/issues/7835#issuecomment-3610820277
<img width="1359" height="275" alt="Image" src="https://github.com/user-attachments/assets/12252d61-8004-4ed5-9712-51f271660ff3" /> 这是seata 3节点的CPU是用情况，看上去并不高
下面是seata 3节点与另外一个k8s节点的ping延迟情况，整体看上去也不高 @funky-eyes ：
worker1：
64 bytes from 172.16.8.239: icmp_seq=1 ttl=64 time=0.483 ms
64 bytes from 172.16.8.239: icmp_seq=2 ttl=64 time=0.230 ms
64 bytes from 172.16.8.239: icmp_seq=3 ttl=64 time=0.161 ms
64 bytes from 172.16.8.239: icmp_seq=4 ttl=64 time=0.139 ms
64 bytes from 172.16.8.239: icmp_seq=5 ttl=64 time=0.290 ms
64 bytes from 172.16.8.239: icmp_seq=6 ttl=64 time=0.196 ms
64 bytes from 172.16.8.239: icmp_seq=7 ttl=64 time=0.524 ms
worker2：
64 bytes from 172.16.8.239: icmp_seq=1 ttl=64 time=11.7 ms
64 bytes from 172.16.8.239: icmp_seq=2 ttl=64 time=0.502 ms
64 bytes from 172.16.8.239: icmp_seq=3 ttl=64 time=0.140 ms
64 bytes from 172.16.8.239: icmp_seq=4 ttl=64 time=0.171 ms
64 bytes from 172.16.8.239: icmp_seq=5 ttl=64 time=0.146 ms
64 bytes from 172.16.8.239: icmp_seq=6 ttl=64 time=0.192 ms
64 bytes from 172.16.8.239: icmp_seq=7 ttl=64 time=0.234 ms
64 bytes from 172.16.8.239: icmp_seq=8 ttl=64 time=0.192 ms
worker3：
64 bytes from 172.16.8.239: icmp_seq=1 ttl=64 time=0.311 ms
64 bytes from 172.16.8.239: icmp_seq=2 ttl=64 time=0.168 ms
64 bytes from 172.16.8.239: icmp_seq=3 ttl=64 time=0.139 ms
64 bytes from 172.16.8.239: icmp_seq=4 ttl=64 time=0.242 ms
64 bytes from 172.16.8.239: icmp_seq=5 ttl=64 time=0.211 ms
64 bytes from 172.16.8.239: icmp_seq=6 ttl=64 time=0.269 ms
64 bytes from 172.16.8.239: icmp_seq=7 ttl=64 time=0.373 ms
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: notifications-unsubscribe@seata.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
---------------------------------------------------------------------
To unsubscribe, e-mail: notifications-unsubscribe@seata.apache.org
For additional commands, e-mail: notifications-help@seata.apache.org
Re: [I] seata raft模式下500并发时GlobalBeginRequest、GlobalCommitRequest耗时100-500多毫秒 [incubator-seata]
Posted by "funky-eyes (via GitHub)" <gi...@apache.org>.
funky-eyes commented on issue #7835:
URL: https://github.com/apache/incubator-seata/issues/7835#issuecomment-3611011490
In theory, GlobalBeginRequest shouldn't take that long. If you're using TCC branches, the latency of GlobalCommitRequest depends on the number of branches and the speed of each branch's second‑phase processing.
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: notifications-unsubscribe@seata.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
---------------------------------------------------------------------
To unsubscribe, e-mail: notifications-unsubscribe@seata.apache.org
For additional commands, e-mail: notifications-help@seata.apache.org
Re: [I] seata raft模式下500并发时GlobalBeginRequest、GlobalCommitRequest耗时100-500多毫秒 [incubator-seata]
Posted by "funky-eyes (via GitHub)" <gi...@apache.org>.
funky-eyes commented on issue #7835:
URL: https://github.com/apache/incubator-seata/issues/7835#issuecomment-3616085234
When you mentioned 500 concurrent TPS, are you referring to concurrent threads or actual TPS? If it's concurrent threads, I suggest setting it to 2-3 times the number of cores on your load testing machine. Regarding virtual threads, it's not viable in JDK 21 at the moment, as Apache Commons Pool (used in Jedis) and the MySQL driver will pin virtual threads. Therefore, you should prioritize increasing the corePoolSize instead.
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: notifications-unsubscribe@seata.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
---------------------------------------------------------------------
To unsubscribe, e-mail: notifications-unsubscribe@seata.apache.org
For additional commands, e-mail: notifications-help@seata.apache.org
Re: [I] seata raft模式下500并发时GlobalBeginRequest、GlobalCommitRequest耗时100-500多毫秒 [incubator-seata]
Posted by "john1337 (via GitHub)" <gi...@apache.org>.
john1337 commented on issue #7835:
URL: https://github.com/apache/incubator-seata/issues/7835#issuecomment-3611050641
发起全局事务的服务用的AT模式，即服务A，服务A会调用服务B（TCC模式）、服务C（AT模式）的相关接口，基本上就是这么个调用链路，我确实对GlobalBeginRequest 耗时在100-300毫秒左右感觉有些诧异
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: notifications-unsubscribe@seata.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
---------------------------------------------------------------------
To unsubscribe, e-mail: notifications-unsubscribe@seata.apache.org
For additional commands, e-mail: notifications-help@seata.apache.org
Re: [I] seata raft模式下500并发时GlobalBeginRequest、GlobalCommitRequest耗时100-500多毫秒 [incubator-seata]
Posted by "john1337 (via GitHub)" <gi...@apache.org>.
john1337 commented on issue #7835:
URL: https://github.com/apache/incubator-seata/issues/7835#issuecomment-3611059683
另外GlobalCommitRequest应该是服务A发起的吧，seata-server会阻塞等待服务B、服务C执行完commit操作之后才返回么，我现在监测到的是服务A发起GlobalCommitRequest 这个请求的耗时 @funky-eyes
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: notifications-unsubscribe@seata.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
---------------------------------------------------------------------
To unsubscribe, e-mail: notifications-unsubscribe@seata.apache.org
For additional commands, e-mail: notifications-help@seata.apache.org
Re: [I] seata raft模式下500并发时GlobalBeginRequest、GlobalCommitRequest耗时100-500多毫秒 [incubator-seata]
Posted by "funky-eyes (via GitHub)" <gi...@apache.org>.
funky-eyes commented on issue #7835:
URL: https://github.com/apache/incubator-seata/issues/7835#issuecomment-3619710068
I also recommend deploying the Raft cluster on virtual machines (VMs) instead of using Kubernetes pods — this will give you more accurate test results with fewer sources of interference.
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: notifications-unsubscribe@seata.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
---------------------------------------------------------------------
To unsubscribe, e-mail: notifications-unsubscribe@seata.apache.org
For additional commands, e-mail: notifications-help@seata.apache.org
Re: [I] seata raft模式下500并发时GlobalBeginRequest、GlobalCommitRequest耗时100-500多毫秒 [incubator-seata]
Posted by "john1337 (via GitHub)" <gi...@apache.org>.
john1337 commented on issue #7835:
URL: https://github.com/apache/incubator-seata/issues/7835#issuecomment-3650964891
暂时还没有，过几天等服务器资源充足了，我再验证下 @funky-eyes
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: notifications-unsubscribe@seata.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
---------------------------------------------------------------------
To unsubscribe, e-mail: notifications-unsubscribe@seata.apache.org
For additional commands, e-mail: notifications-help@seata.apache.org
Re: [I] seata raft模式下500并发时GlobalBeginRequest、GlobalCommitRequest耗时100-500多毫秒 [incubator-seata]
Posted by "john1337 (via GitHub)" <gi...@apache.org>.
john1337 commented on issue #7835:
URL: https://github.com/apache/incubator-seata/issues/7835#issuecomment-3619107183
jdk21在虚拟线程使用场景确实存在问题，但是我用jdk25可以解决jdk21在虚拟线程上的问题，我尝试增加corePoolSize 的值，但是给到的CPU是16核，增加到100也没有什么效果，所以我想尝试用jdk25的虚拟线程来优化传统线程池 @funky-eyes
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: notifications-unsubscribe@seata.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
---------------------------------------------------------------------
To unsubscribe, e-mail: notifications-unsubscribe@seata.apache.org
For additional commands, e-mail: notifications-help@seata.apache.org
Re: [I] seata raft模式下500并发时GlobalBeginRequest、GlobalCommitRequest耗时100-500多毫秒 [incubator-seata]
Posted by "funky-eyes (via GitHub)" <gi...@apache.org>.
funky-eyes commented on issue #7835:
URL: https://github.com/apache/incubator-seata/issues/7835#issuecomment-3614964957
Please review the GC logs to determine if GC pauses are impacting performance, as the stack traces suggest that the prolonged execution times seem to be randomly occurring in certain methods.
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: notifications-unsubscribe@seata.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
---------------------------------------------------------------------
To unsubscribe, e-mail: notifications-unsubscribe@seata.apache.org
For additional commands, e-mail: notifications-help@seata.apache.org
Re: [I] seata raft模式下500并发时GlobalBeginRequest、GlobalCommitRequest耗时100-500多毫秒 [incubator-seata]
Posted by "john1337 (via GitHub)" <gi...@apache.org>.
john1337 commented on issue #7835:
URL: https://github.com/apache/incubator-seata/issues/7835#issuecomment-3611076990
我也测试了redis作为存储的高可用部署模式，GlobalCommitRequest与GlobalCommitRequest 也是耗时100-500多毫秒
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: notifications-unsubscribe@seata.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
---------------------------------------------------------------------
To unsubscribe, e-mail: notifications-unsubscribe@seata.apache.org
For additional commands, e-mail: notifications-help@seata.apache.org
Re: [I] seata raft模式下500并发时GlobalBeginRequest、GlobalCommitRequest耗时100-500多毫秒 [incubator-seata]
Posted by "funky-eyes (via GitHub)" <gi...@apache.org>.
funky-eyes commented on issue #7835:
URL: https://github.com/apache/incubator-seata/issues/7835#issuecomment-3615101833
By default, Seata sets both Xmx and Xms to 2g. To override this, configure the JVM_XMX and JVM_XMS environment variables, for example: export JVM_XMX=6g.
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: notifications-unsubscribe@seata.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
---------------------------------------------------------------------
To unsubscribe, e-mail: notifications-unsubscribe@seata.apache.org
For additional commands, e-mail: notifications-help@seata.apache.org
Re: [I] seata raft模式下500并发时GlobalBeginRequest、GlobalCommitRequest耗时100-500多毫秒 [incubator-seata]
Posted by "john1337 (via GitHub)" <gi...@apache.org>.
john1337 commented on issue #7835:
URL: https://github.com/apache/incubator-seata/issues/7835#issuecomment-3616444909
500并发线程，但是我现在了吞吐量在500，我给到的CPU是16C，seata默认的线程池最小为50，最大为500，但是基本上就不会达到500，因为线程池的队列长度为2万
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: notifications-unsubscribe@seata.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
---------------------------------------------------------------------
To unsubscribe, e-mail: notifications-unsubscribe@seata.apache.org
For additional commands, e-mail: notifications-help@seata.apache.org
Re: [I] seata raft模式下500并发时GlobalBeginRequest、GlobalCommitRequest耗时100-500多毫秒 [incubator-seata]
Posted by "funky-eyes (via GitHub)" <gi...@apache.org>.
funky-eyes commented on issue #7835:
URL: https://github.com/apache/incubator-seata/issues/7835#issuecomment-3645587929
Has there been any progress on this issue?
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: notifications-unsubscribe@seata.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
---------------------------------------------------------------------
To unsubscribe, e-mail: notifications-unsubscribe@seata.apache.org
For additional commands, e-mail: notifications-help@seata.apache.org
Re: [I] seata raft模式下500并发时GlobalBeginRequest、GlobalCommitRequest耗时100-500多毫秒 [incubator-seata]
Posted by "funky-eyes (via GitHub)" <gi...@apache.org>.
funky-eyes commented on issue #7835:
URL: https://github.com/apache/incubator-seata/issues/7835#issuecomment-3611005433
Use Arthas (or a similar tool) to capture the stack traces that show where time is actually being spent, and send them over.
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: notifications-unsubscribe@seata.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
---------------------------------------------------------------------
To unsubscribe, e-mail: notifications-unsubscribe@seata.apache.org
For additional commands, e-mail: notifications-help@seata.apache.org
Re: [I] seata raft模式下500并发时GlobalBeginRequest、GlobalCommitRequest耗时100-500多毫秒 [incubator-seata]
Posted by "john1337 (via GitHub)" <gi...@apache.org>.
john1337 commented on issue #7835:
URL: https://github.com/apache/incubator-seata/issues/7835#issuecomment-3611444685
是用arthas对commit操作进行了trace，结果如下：
`---ts=2025-12-04 18:39:47.032;thread_name=tomcat-handler-164682;id=169286;is_daemon=true;priority=5;TCCL=org.springframework.boot.web.embedded.tomcat.TomcatEmbeddedWebappClassLoader@2f079444
`---[452.981685ms] org.apache.seata.core.rpc.netty.AbstractNettyRemoting:sendSync()
+---[0.00% 0.00254ms ] org.apache.seata.core.rpc.netty.AbstractNettyRemoting$$sw$auxiliary$ln5r360:<init>()
`---[100.00% 452.965185ms ] org.apache.skywalking.apm.agent.core.plugin.interceptor.enhance.InstMethodsInter:intercept()
99%以上的耗时在InstMethodsInter:intercept方法上，客户端我添加了seata的apm追踪插件，我准备关闭skywalking追踪再测试下 @funky-eyes
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: notifications-unsubscribe@seata.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
---------------------------------------------------------------------
To unsubscribe, e-mail: notifications-unsubscribe@seata.apache.org
For additional commands, e-mail: notifications-help@seata.apache.org
Re: [I] seata raft模式下500并发时GlobalBeginRequest、GlobalCommitRequest耗时100-500多毫秒 [incubator-seata]
Posted by "funky-eyes (via GitHub)" <gi...@apache.org>.
funky-eyes commented on issue #7835:
URL: https://github.com/apache/incubator-seata/issues/7835#issuecomment-3610755477
What's the ping latency (round‑trip time) from the client to the server? Also, can you provide (or expose here) the server's CPU and I/O monitoring metrics?
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: notifications-unsubscribe@seata.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
---------------------------------------------------------------------
To unsubscribe, e-mail: notifications-unsubscribe@seata.apache.org
For additional commands, e-mail: notifications-help@seata.apache.org
Re: [I] seata raft模式下500并发时GlobalBeginRequest、GlobalCommitRequest耗时100-500多毫秒 [incubator-seata]
Posted by "john1337 (via GitHub)" <gi...@apache.org>.
john1337 commented on issue #7835:
URL: https://github.com/apache/incubator-seata/issues/7835#issuecomment-3615124154
我通过JAVA_OPT设置了-Xmx8G -Xms8g -Xmn4g，原生的镜像里面并没有skywalking-agent插件，我准备新弄一个docker镜像，把sky-agent放进去，打开skywalking再看下
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: notifications-unsubscribe@seata.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
---------------------------------------------------------------------
To unsubscribe, e-mail: notifications-unsubscribe@seata.apache.org
For additional commands, e-mail: notifications-help@seata.apache.org
Re: [I] seata raft模式下500并发时GlobalBeginRequest、GlobalCommitRequest耗时100-500多毫秒 [incubator-seata]
Posted by "funky-eyes (via GitHub)" <gi...@apache.org>.
funky-eyes commented on issue #7835:
URL: https://github.com/apache/incubator-seata/issues/7835#issuecomment-3619692490
Based on the stack traces you shared, requests that take over 100 ms are getting blocked inside different methods. Some of those methods are just a simple map.get call and shouldn’t be that expensive, and this isn’t related to the thread count you inferred because the request is already running on a thread. You should check the GC logs for any full GCs and also verify whether the Kubernetes pod is being CPU‑throttled.
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: notifications-unsubscribe@seata.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
---------------------------------------------------------------------
To unsubscribe, e-mail: notifications-unsubscribe@seata.apache.org
For additional commands, e-mail: notifications-help@seata.apache.org