RFR: 8348309: MultiNST tests need more debugging and timing

[Anthony Scarpino]

On Tue, 18 Feb 2025 15:48:26 GMT, Matthew Donovan <mdonovan at openjdk.org> wrote:

>> I need a review of this change that adds new timing controls for the initial server setup.  On rare occasions, more so on certain architectures, the server may not fully start before the client tries to connect.  Additional debugging is added to help identify if there are other timing issues.
>> 
>> Thanks
>> 
>> Tony
>
> test/jdk/sun/security/ssl/SSLSessionImpl/MultiNSTParallel.java line 76:
> 
>> 74:             String name = Thread.currentThread().getName();
>> 75:             SSLSession r;
>> 76:             System.err.println(name + " is ready");
> 
> This test is printing some things to System.out and some to System.err. It's also using TLSBase which is printing everything to System.err. This will make the logs difficult to read and understand.

With the threading though, some of the errors need to be printed right away.  I understand your concern, but my attempt is to make stdout display obvious results without searching all the debugging info.  In some cases catching an exception is a valid result, which may look like a test failure if they were all in one output buffer.

-------------

PR Review Comment: https://git.openjdk.org/jdk/pull/23407#discussion_r1960495565