Kubernetes 分布式测试 - Jmeter 不会启动； sun.rmi.transport.tcp.TCPConnection 关闭答案

【问题标题】：Kubernetes Distributed testing - Jmeter won't start; sun.rmi.transport.tcp.TCPConnection closeKubernetes 分布式测试 - Jmeter 不会启动； sun.rmi.transport.tcp.TCPConnection 关闭
【发布时间】：2022-10-14 13:05:12
【问题描述】：

我的架构：kubernetes 中的命名空间位于代理、防火墙和其他交换机后面。我的主人向奴隶发送任务，当脚本结束时，它会关闭奴隶和主人。我已经配置并运行了 netpols、入口和出口。我在 3 端口范围内有开放端口，例如60000-60002、1099-1101。

我的问题：通常，当我开始测试时，主人应该发送日志，例如：

"Oct 12, 2022 11:31:39 AM java.util.prefs.FileSystemPreferences$1 run
INFO: Created user preferences directory.
Creating summariser <summary>
Created the tree successfully using /mount/10_2022/scenario.jmx
Configuring remote engine: slave-1
Using local port: 60000
Starting distributed test with remote engines: [slave-1] @ October 12, 2022 11:31:41 AM CEST (1665567101036)
Remote engines have been started:[opl-jmeter-distributed-slave-1]
Waiting for possible Shutdown/StopTestNow/HeapDump/ThreadDump message on port 4445"

当主服务器发送日志时：

"Oct 12, 2022 11:31:39 AM java.util.prefs.FileSystemPreferences$1 run
INFO: Created user preferences directory.
Creating summariser <summary>
Created the tree successfully using /mount/10_2022/scenario.jmx
Configuring remote engine: slave-1
Using local port: 60000
Starting distributed test with remote engines: [slave-1] @ October 12, 2022 11:31:41 AM CEST (1665567101036)

<这行之间大约有7-10分钟的休息时间。>

"Remote engines have been started:[opl-jmeter-distributed-slave-1]
Waiting for possible Shutdown/StopTestNow/HeapDump/ThreadDump message on port 4445"

之后什么都没有发生，除了日志错误，例如：

"2022-10-12 11:40:30,274 ERROR o.a.j.s.RemoteListenerWrapper: testStarted(host) on slave-1
java.rmi.ConnectException: Connection refused to host: <MASTER'S IP>; nested exception is: java.net.ConnectException: Connection timed out (Connection timed out)"

和

"2022-10-12 11:49:18,668 ERROR o.a.j.s.BatchSampleSender: sampleOccurred
java.rmi.ConnectException: Connection refused to host: <MASTER'S IP>; nested exception is:java.net.ConnectException: Connection timed out (Connection timed out)"

我开始调试我的操作，这是我从主日志中得到的：

"Oct 13, 2022 10:45:37 AM sun.rmi.transport.tcp.TCPChannel free
FINE: RMI RenewClean-[172.16.28.15:60000,SSLRMIClientSocketFactory(keyStoreLocation=/mount/keyStore/rmi_keystore.jks, type=JKS, trustStoreLocation=/mount/keyStore/rmi_keystore.jks, type=JKS, alias=rmi)]: create reaper
Oct 13, 2022 10:45:52 AM sun.rmi.transport.tcp.TCPChannel$1 run
FINER: RMI Scheduler(0): wake up
Oct 13, 2022 10:45:52 AM sun.rmi.transport.tcp.TCPChannel freeCachedConnections
FINER: RMI Scheduler(0): connection timeout expired
Oct 13, 2022 10:45:52 AM sun.rmi.transport.tcp.TCPConnection close
FINE: RMI Scheduler(0): close connection"

我开始调试我的操作，这是我从主日志中得到的：

Oct 13, 2022 10:45:37 AM sun.rmi.transport.tcp.TCPChannel free
FINE: RMI RenewClean-[172.16.28.15:60000,SSLRMIClientSocketFactory(keyStoreLocation=/mount/keyStore/rmi_keystore.jks, type=JKS, trustStoreLocation=/mount/keyStore/rmi_keystore.jks, type=JKS, alias=rmi)]: create reaper
Oct 13, 2022 10:45:52 AM sun.rmi.transport.tcp.TCPChannel$1 run
FINER: RMI Scheduler(0): wake up
Oct 13, 2022 10:45:52 AM sun.rmi.transport.tcp.TCPChannel freeCachedConnections
FINER: RMI Scheduler(0): connection timeout expired
Oct 13, 2022 10:45:52 AM sun.rmi.transport.tcp.TCPConnection close
FINE: RMI Scheduler(0): close connection

请帮忙。有什么线索吗？为什么我的测试没有开始？

【问题讨论】：

标签： kubernetes jmeter distributed

【解决方案1】：

很可能您的RMI configuration 不正确

您需要创建一个Service 并公开以下端口：

1099 或您的 server_port
4000 或您的 server.rmi.localport

就像在从站的 Dockerfile 中一样：

EXPOSE 1099 4000

在 k8s 服务 YAML

spec:
  clusterIP: None
  ports:
    - port: 1099
      name: server_port
      targetPort: 1099
    - port: 4000
      name: server_rmi_localport
      targetPort: 4000

并对 JMeter 主端的client.rmi.localport 执行相同的操作。

【讨论】：