https://bz.apache.org/bugzilla/show_bug.cgi?id=65455
--- Comment #2 from 13998197...@163.com --- (In reply to Mark Thomas from comment #1) > It looks like Tomcat is reaching the maxConnections limit. Given the low > usage of the system, it looks like some connection closures are being > missed. If this analysis is correct, the finding the root cause is going to > be a lot easier if we can recreate the issue. > > The default value for maxConnections is 10000. Can you try reducing, the > lower the better, and then try reproducing the issue? That should trigger > the issue faster and may provide some clues as to which connection closures > are being missed. > > What is the nature of the web application where this is observed? Does is > use the Servlet async API? Does it use the Servlet non-blocking API? Does it > use WebSocket? > > Try switching to the NIO connector. The performance of NIO and NIO2 is very > similar is most circumstances. > > Testing with the latest 8.5.x release would also be helpful. First of all, THANK YOU for your help!! I failed to recreate the issue by using jmeter and hping3, even if use hping3 mock flood attack. I use command 'netstat -nat|grep 8080|wc -l' to stat connections when issue happens, it's 108, and server load is very light, 3-5 req/second, so I think the reason is not maxConnections limit. By the way, client lib is using httpClient, it's timeout param set to 10s. I think, client and server's time out param cause the issue. -- You are receiving this mail because: You are the assignee for the bug. --------------------------------------------------------------------- To unsubscribe, e-mail: dev-unsubscr...@tomcat.apache.org For additional commands, e-mail: dev-h...@tomcat.apache.org
