On Thu, 25 Jun 2026 07:19:19 GMT, Alan Bateman <[email protected]> wrote:
> Does the logging by code in the httpserver include something to identity the > instance? Nope, it doesn't. For tests, it doesn't matter much — since most HTTP server & client tests run sequentially. While skimming through logs, you can reliably guess which log belongs to which server. That said, for production applications containing multiple HTTP servers, an identifier in the log would be super helpful while troubleshooting. It'd even help with the cognitive load in test logs. @kieran-farrell, @dfuch, do you agree? > I'm wondering how the log messages look when there is more than one instance > in the VM. They get interleaved. Consider the following JTreg log snippet from a `test/jdk/com/sun/net/httpserver/ClearTextServerSSL` execution: [18:28:00.715] STARTED ClearTextServerSSL::test 'test()' Jun 24, 2026 6:28:00 PM sun.net.httpserver.ServerImpl <init> FINE: HttpServer created http localhost/127.0.0.1:0 Jun 24, 2026 6:28:00 PM sun.net.httpserver.ServerImpl createContext FINE: context created: /ClearTextServerSSL/ Jun 24, 2026 6:28:01 PM sun.net.httpserver.ServerImpl$Exchange run FINER: exchange started Jun 24, 2026 6:28:01 PM sun.net.httpserver.ServerImpl$Exchange run FINE: Exchange request line: GET /ClearTextServerSSL/clear HTTP/1.1 Jun 24, 2026 6:28:01 PM sun.net.httpserver.ExchangeImpl sendResponseHeaders FINER: Sent headers: noContentToSend=true ------------- PR Review Comment: https://git.openjdk.org/jdk/pull/31670#discussion_r3472668542
