Written for the person debugging at three in the morning

Everything here is about the moment when the system is already broken and the only tools are what you left behind: the log line you wrote for yourself, the stack trace the JVM decided to stop printing, the retry semantics nobody wrote down when the queue went in, the cause hidden behind the failure that closed the resource. Operability is a design decision made months before it is needed.

• Most arguments about adding a queue are arguments about retries, not throughput
  19 September 2026 · The load numbers get quoted, the failure semantics get assumed, and the assumption is what breaks.
• The exception that closed your resource hides the one that broke your code
  19 September 2026 · try-with-resources suppresses the close() failure and most logging throws it away. getSuppressed() is where the real cause went.
• The JVM stops printing stack traces on repeated exceptions, and one flag restores them
  14 September 2026 · An older note, kept here to show what falls out of the news window.
• Logs are a product, and your user is you at three in the morning
  10 September 2026 · Write the line you will need, not the line that is easy to write.