HttpClient thread stuck using 100% cpu on JDK17

Issue

• The following sequence of events leaves the HttpClient thread stuck using 100% cpu:

  • After a successful HTTPS request, the connection in HttpClient pool remains open (keep-alive).

  • 75 seconds later, the remote server initiates closing the connection.

  • The HttpClient worker thread with the stack trace below uses 100% CPU:

  "HttpClient-1-Worker-2" #17 daemon prio=5 os_prio=0 cpu=27904.71ms elapsed=28.16s tid=0x00007fb10c00e860 nid=0x105e76 runnable  [0x00007fb1959fd000]
  java.lang.Thread.State: RUNNABLE
  at sun.security.ssl.SSLEngineImpl.unwrap(java.base@17.0.17/SSLEngineImpl.java:506)
  at sun.security.ssl.SSLEngineImpl.unwrap(java.base@17.0.17/SSLEngineImpl.java:482)
  at javax.net.ssl.SSLEngine.unwrap(java.base@17.0.17/SSLEngine.java:679)
  at jdk.internal.net.http.common.SSLFlowDelegate$Reader.unwrapBuffer(java.net.http@17.0.17/SSLFlowDelegate.java:543)
  at jdk.internal.net.http.common.SSLFlowDelegate$Reader.processData(java.net.http@17.0.17/SSLFlowDelegate.java:447) 
  locked <0x000000060f40c140> (a java.lang.Object)
  at jdk.internal.net.http.common.SSLFlowDelegate$Reader$ReaderDownstreamPusher.run(java.net.http@17.0.17/SSLFlowDelegate.java:282)
  at jdk.internal.net.http.common.SequentialScheduler$LockingRestartableTask.run(java.net.http@17.0.17/SequentialScheduler.java:205)
  at jdk.internal.net.http.common.SequentialScheduler$CompleteRestartableTask.run(java.net.http@17.0.17/SequentialScheduler.java:149)
  at jdk.internal.net.http.common.SequentialScheduler$SchedulableTask.run(java.net.http@17.0.17/SequentialScheduler.java:230)
  at java.util.concurrent.ThreadPoolExecutor.runWorker(java.base@17.0.17/ThreadPoolExecutor.java:1136)
  at java.util.concurrent.ThreadPoolExecutor$Worker.run(java.base@17.0.17/ThreadPoolExecutor.java:635)
  at java.lang.Thread.run(java.base@17.0.17/Thread.java:840)
  

• Only restarting resolves the issue.