[logback-dev] [JIRA] Updates for LOGBACK-1711: Deadlock when using Virtual Threads

[QOS.CH (JIRA)]

logback / LOGBACK-1711 [Open]
Deadlock when using Virtual Threads

==============================

Here's what changed in this issue in the last few minutes.
This issue has been created
This issue is now assigned to you.

View or comment on issue using this link
https://jira.qos.ch/browse/LOGBACK-1711

==============================
 Issue created
------------------------------

Gili created this issue on 03/Dec/22 9:52 AM
Summary:              Deadlock when using Virtual Threads
Issue Type:           Bug
Affects Versions:     1.4.5
Assignee:             Logback dev list
Components:           logback-core
Created:              03/Dec/22 9:52 AM
Priority:             Blocker
Reporter:             Gili
Description:
  Repro steps:
   # Unpack testcase.zip
   # Run the Testcase class
   # Application should deadlock at around the time it outputs "Progress: 340000"
  
  This issue is 100% reproducible and only occurs when using JDK 19 virtual threads.
   
  The bug seems to be triggered by *RollingFileAppender.rollover()*.
   
  Please note that using I/O inside synchronized blocks pins a carrier threads and this substantially reduces the performance of virtual threads. Please consider replacing such constructs with ReentrantLock or StampedLock as you have done elsewhere.
  
  The deadlock might have something to do with all carrier threads waiting to acquire a synchronized block, while the thread holding the lock is blocked waiting for an I/O response. I'm not sure why the latter never comes back though...


==============================
 This message was sent by Atlassian Jira (v8.8.0#808000-sha1:e2c7e59)