引用:

昨天在測試幫助文件時發(fā)現(xiàn)一直打不開，好像是一直再刷新，于是就到服務(wù)器上查看Tomcat的日志，就發(fā)現(xiàn)了下面的信息：
Apr 21, 2008 9:27:34 AM org.apache.tomcat.util.net.TcpWorkerThread runIt
SEVERE: Remote Host /172.16.128.248 SocketException: Connection reset
并且這兩條信息幾乎是每個1-2秒就出現(xiàn)一次，很是頻繁。于是就搜索了一下，就找到了下面的內(nèi)容：
The usual cause is that the browser''s stopped a HTTP request part-way through,
generally because the user''s navigated to another page before the first page has
completely downloaded. The browser closes the connection, with the result that
Tomcat gets an exception when it next tries to write data to the socket.
Arguably it shouldn''t be logged as a SEVERE error on production systems, but
getting that log message during development and stress testing can be very
handy! 于是就查看它引用的JavaScript文件，發(fā)現(xiàn)好多不存在。終于明白，是因?yàn)橛行┳兞坎淮嬖冢瑢?dǎo)致了這里的反復(fù)重載。重新加入這些JavaScript文件后，問題解決。

Fix it by any of the following (in increasing order of difficulty):

1) Ignore it (and change any log-reading scripts that look for SEVERE errors to
ignore this one);

2) Grab the source for 5.0.28, find the line where this is logged and change the
logging level in the case of a connection reset error, then recompile Tomcat;

3) Educate your users that they are supposed to sit on their hands until the
page has completely loaded (by far the hardest ;-) ).

后面還有別人的補(bǔ)充：
If it happens that often and you don''t have a high traffic site, it might be a monitor script, that does only check if it can do a TCP connect and drop the connection before a full HTTP request-response cycle has finished.

不過我的情況和上面所說的都不符合，于是就查看了源文件，發(fā)現(xiàn)了如下代碼：



if (!window.gbWhVer||!window.gbWhUtil||!window.gbWhMsg)
    document.location.reload();