[SLING-11824] Deadlock during repository restart

• Type: Bug
• Status: Closed
• Priority: Major
• Resolution: Fixed
• Affects Version/s: Engine 2.14.0
• Fix Version/s: Engine 2.15.0
• Component/s: Engine
• Assignee: Carsten Ziegeler
• Reporter: Julian Sedding
• Created: 11/Apr/23 15:50
• Updated: 26/May/23 07:31
• Resolved: 12/Apr/23 05:35

Description

A deadlock can occur when the repository is restarted in a running system. This may be due to a configuration change of a low level component, but has also been reported to occur after refreshing the commons-io bundle.

From the thread dump

"VM Thread" os_prio=0 cpu=106005.20ms elapsed=72363.91s tid=0x000056165220f800 nid=0x19ef runnable  
"ParGC Thread#0" os_prio=0 cpu=11616.68ms elapsed=72363.93s tid=0x00005616520cb800 nid=0x19ee runnable  
"ParGC Thread#1" os_prio=0 cpu=11624.23ms elapsed=72362.83s tid=0x00005616534fd800 nid=0x19f8 runnable  
"ParGC Thread#2" os_prio=0 cpu=11621.66ms elapsed=72362.83s tid=0x0000561653518000 nid=0x19f9 runnable  
"VM Periodic Task Thread" os_prio=0 cpu=32588.40ms elapsed=72360.26s tid=0x0000561655593800 nid=0x1a13 waiting on condition  
JNI global refs: 98, weak refs: 3
Found one Java-level deadlock:
=============================
"CM Event Dispatcher (Fire ConfigurationEvent: pid=org.apache.sling.jcr.resource.internal.JcrResourceResolverFactoryImpl)":
  waiting to lock monitor 0x000056166302ce80 (object 0x00000007679e6638, a org.apache.sling.engine.impl.helper.SlingServletContext),
  which is held by "SlingServletContext registration"
"SlingServletContext registration":
  waiting to lock monitor 0x00005616631e3480 (object 0x000000074f987900, a java.util.HashMap),
  which is held by "CM Event Dispatcher (Fire ConfigurationEvent: pid=org.apache.sling.jcr.resource.internal.JcrResourceResolverFactoryImpl) 

Issue Links

• links to: GitHub Pull Request #32