> You are right there is something wrong here as well, especially since the
> time gets worse on the second attempt. What is the actual roundtrip time
> with your server ?
>
> Rémy
Based on ping, average time is 83ms
Some more details that allow to get the resets: the test code from
above should
> Since 9.0.12 and 16 do the same, I wouldn't look at that at all. Something
> simple like this works in the general case, there must be something
> specific here. So it's Windows, which some unspecified OpenSSL version.
>
> Rémy
That's not right. After many tests I've found out that 9.0.12 build
9.0.17 was used from
http://tomcat.10.x6.nabble.com/VOTE-Release-Apache-Tomcat-9-0-17-td5083815.html
I've created a new totally clean AWS instance with new domain name and
new Lets Encrypt certificates.
It's available here:
temp-test-01.tk
Configuration:
OS: Windows Server 2012 R2 Base
Prerequisites:
OS: Windows Server 2012 R2
Java: checked on both jdk1.8.0_162 jdk1.8.0_181
Tomcat: windows x64 builds checked on 9.0.12, 9.0.16, 9.0.17-dev
Valid SSL certificates
Content of file located at webapp/ROOT/1.txt: []
Tomcat's connector settings:
This configuration leads to 50%
I'm experiencing many "Socket accept failed: The specified network
name is no longer available." exceptions in the Tomcat error log.
These exceptions happen both on port 80 and 443. Both connectors are
using Http11Nio2Protocol.
I can't reproduce this exception on the local PC, but on the
