PeopleSoft Web 服务器崩溃,失去与 AppServer 的连接
在我们的 Web 服务器上,我们看到大量以下错误:
应用程序服务器上次连接 //psoftapp.company.net_8850 bea.jolt.ServiceException:bea.jolt.JoltRemoteService(GetCertificate)call():超时\ nbea.jolt.SessionException:连接接收错误\ nbea.jolt.JoltException:[3] NwHdlr.recv():超时
错误应用程序服务器:
PSPUBDSP_dflt.27505 (0) 07/20/11 08:13:33 (JNIUTIL): Java 异常抛出: java.net.SocketException: 连接重置
我正在阅读 PeopleSoft & 的一些调整文档。我发现了一个在几个地方看到过的建议——将 tcp_wait_time_interval 减少到 60 秒。我想我有点明白这是在做什么 - 似乎不再使用的网络(或套接字?)连接被“回收”或变得可用?有人可以证实这一点吗?另外,为什么这些连接未使用/过时?是否是由于人们没有正确退出应用程序(并且只是关闭浏览器)引起的?
谢谢!
On our Webserver, we're seeing a ton of these errors:
Application Server last connected //psoftapp.company.net_8850
bea.jolt.ServiceException: bea.jolt.JoltRemoteService(GetCertificate)call(): Timeout\nbea.jolt.SessionException: Connection recv error\nbea.jolt.JoltException: [3] NwHdlr.recv(): Timeout Error
and on our Appserver:
PSPUBDSP_dflt.27505 (0) 07/20/11 08:13:33 (JNIUTIL): Java exception thrown: java.net.SocketException: Connection reset
I'm reading some tuning documents from PeopleSoft & I found a suggestion that I've seen in a couple of places -- Reducing the tcp_wait_time_interval to 60 seconds. I think I sort of understand what this is doing - It seems that network (or socket?) connections that are no longer being used are "recycled" or made available? Can someone confirm this? Also, why are these connections unused/stale? Is it caused by people not properly logging out of the app (and just closing the browser)?
Thanks!
如果你对这篇内容有疑问,欢迎到本站社区发帖提问 参与讨论,获取更多帮助,或者扫码二维码加入 Web 技术交流群。
绑定邮箱获取回复消息
由于您还没有绑定你的真实邮箱,如果其他用户或者作者回复了您的评论,将不能在第一时间通知您!
发布评论
评论(1)
PSPUBDP 是 Integration Broker 应用程序消息传递框架的一部分。您也可以查看 Tuxedo 日志或 Integration Broker Monitor,了解发生了什么情况。您可能正在运行大量消息并使服务器超载,或者您可能收到一条包含错误的消息,从而以某种方式导致崩溃。
PSPUBDP is part of the Integration Broker application messaging framework. You could look at the Tuxedo logs or the Integration Broker Monitor too see what is going on. You may be running a high number of messages and overloading the server or possibly you have a message with errors that is somehow causing the crashes.