Topics

Glassfish slow undeploy of a web-app with redisson


simon.cigoj@...
 

I'm have a webapp deployed on Glassfish 4.1.2 which includes Redisson 3.5.4 and on the undeploy of the app it takes about 5 minutes.

I did some analyse with jprofiler and all the threads of Redisson are stoped ok.
After the servlet ServletContextListener.contextDestroyed is called the undeploy waits for around 5 minutes.

If I disable Redisson the app undeploys much faster...around 1 minute.

Here is the thread dump made during the long undeploy https://pastebin.com/Sv4VW4pT

In the long undeploy period there is no CPU activity or GC activity visible.

How can I debug this is there any appropriate package where should I put a higher (FINER/FINEST) logging level to se what is going on, or is there a time-out setting that could affect some undeploy time, because the time taken seems constant around 5 minutes. 

My stack:
Glassfish 4.1.2 (build 1)
java version "1.8.0_144" 
Redisson 3.5.4
Netty 4.1.15.FInal
Windows10

I have tried to upgrade Redisson also to 3.5.5 but the same problem persists 

This was what I could get from the logs:

[2017-11-08T15:28:28.927+0100] [glassfish 4.1] [FINE] [] [javax.enterprise.system.core] [tid: _ThreadID=70 _ThreadName=DynamicReloader] [timeMillis: 1510151308927] [levelValue: 500] [CLASSNAME: com.sun.enterprise.v3.server.DynamicReloader] [METHODNAME: chooseAppsToReload] [[
  [Reloader] Removing undeployed app web-test-1.0-SNAPSHOT from reload info]]
 
[2017-11-08T15:31:55.717+0100] [glassfish 4.1] [FINE] [] [javax.enterprise.system.core] [tid: _ThreadID=61 _ThreadName=admin-listener(3)] [timeMillis: 1510151515717] [levelValue: 500] [CLASSNAME: com.sun.enterprise.v3.admin.CommandRunnerImpl] [METHODNAME: doCommand] [[
  Command execution stage 4 : Call post supplemental commands for undeploy]]