We're employing the 'hot deploy' feature in jetty to redeploy an
updated webapp.

What we're observing is that the redeployment doesn't discard the
disk resources of the prior deployment.

Using 'lsof' as such:

  lsof -nP +L1 /var | grep /var | grep '(deleted)' | less

We can see dozens of old deployments (old extracted war files?), like this:

  java      26734        root   82r   REG    8,6   575389     0  32470
  
/var/tmp/webapp_tmp/jetty-0.0.0.0-80-webapp-_idp-any-4794158382650079166.dir/webinf/WEB-INF/lib/commons-collections-3.2.1.jar
  (deleted)
  java      26734        root  110r   REG    8,6   575389     0  32561
  
/var/tmp/webapp_tmp/jetty-0.0.0.0-80-webapp-_idp-any-2567933012053298590.dir/webinf/WEB-INF/lib/commons-collections-3.2.1.jar
  (deleted)
  java      26734        root  318r   REG    8,6   575389     0  40227
  
/var/tmp/webapp_tmp/jetty-0.0.0.0-80-webapp-_idp-any-4949209254558295977.dir/webinf/WEB-INF/lib/commons-collections-3.2.1.jar
  (deleted)
  ...

As the JVM is still holding a reference, the disk space never frees, and our
/var/ partition eventually fills. :/

Do other people see this?

Are we doing this wrong?  All we did was modify the scanInterval setting to
a nonzero value.

Is there some configuration switch we can throw to coerce the JVM
to really release these resources, or are we stuck waiting for the
GC to eventually cleanup?

Details of our environment:

  jetty-distribution-9.3.8.v20160314

  OpenJDK Runtime Environment (build 1.8.0_71-b15)

  CentOS release 6.7 (Final)

  # uname -r
  2.6.32-573.18.1.el6.x86_64

-- 
Brian Reichert                          <[email protected]>
BSD admin/developer at large    
_______________________________________________
jetty-users mailing list
[email protected]
To change your delivery options, retrieve your password, or unsubscribe from 
this list, visit
https://dev.eclipse.org/mailman/listinfo/jetty-users

Reply via email to