[GEODE-8681] peer-to-peer message loss due to sending connection closing with TLS enabled - ASF JIRA

XML

Word

Printable

JSON

Details

Type: Bug
Status: Closed
Priority: Major
Resolution: Fixed
Affects Version/s: 1.10.0, 1.11.0, 1.12.0, 1.13.0
Fix Version/s: 1.12.1, 1.13.1, 1.14.0
Component/s: membership, messaging
Labels:
- pull-request-available

Description

We have observed message loss when TLS is enabled and a distributed lock is released right after sending a message that doesn't require acknowledgement if the sending socket is immediately closed. The closing of sockets immediately after sending a message is frequently seen in function execution threads or server-side application threads that use this pattern:

 try {
    DistributedSystem.setThreadsSocketPolicy(false);
    acquireDistributedLock(lockName);
    (perform one or more cache operations)
  } finally {
    distLockService.unlock(lockName);
    DistributedSystem.releaseThreadsSockets(); // closes the socket
  }

The fault seems to be in NioSSLEngine.unwrap(), which throws an SSLException() if it finds the SSLEngine is closed even though there is valid data in its decrypt buffer. It shouldn't throw an exception in that case.