[GLASSFISH-3935] Orderly ORB shutdown behaves wrong Created: 19/Dec/07  Updated: 08/Feb/12

Status: Open
Project: glassfish
Component/s: orb
Affects Version/s: 9.1pe
Fix Version/s: 4.0

Type: Bug Priority: Minor
Reporter: fmeili Assignee: Harshad Vilekar
Resolution: Unresolved Votes: 3
Labels: None
Remaining Estimate: Not Specified
Time Spent: Not Specified
Original Estimate: Not Specified
Environment:

Operating System: All
Platform: All


Issuezilla Id: 3,935
Tags: 3_1-exclude

 Description   

In case of an orderly shutdown of GlassFish v2 it should block new incoming
requests and throw a CORBA exception with COMPLETE_NO while the shutdown is in
progress. This is documented in section "Impact of orderly shutdown" in this
document https://glassfish-corba.dev.java.net/design/orbdArchitecture.html

But the client receives an java.rmi.ServerException with a notice "server is in
STOPPING". But this exception is not a CORBA exception and so it's impossible
for the client to initiate a failover. Only if it's a CORBA exception with
COMPLETE_NO the client tries to failover, as described in this document.

The result is that all rmi-iiop calls to a server instance which is in a
shutdown in progress state will fail and will not failover to other instances.

More details are in the forum Threads
http://forums.java.net/jive/thread.jspa?threadID=34528&tstart=15
and
https://glassfish-corba.dev.java.net/servlets/ProjectForumMessageView?forumID=2940&messageID=22494



 Comments   
Comment by fmeili [ 14/Jan/08 ]
      • Issue 3935 has been confirmed by votes. ***
Comment by Ken Cavanaugh [ 06/Feb/08 ]

Changed target to V3.

Comment by sanandal [ 11/Jan/09 ]

"Reclassifying as P4 because this issue is not deemed "must fix" for this v2.1
release whose primary release driver is SailFin.
This issue will be scrubbed after this release and will be given the right
priority for the next release."

Comment by Ken Cavanaugh [ 23/Mar/10 ]

Moved to v3.1.

Comment by Ken Cavanaugh [ 23/Mar/10 ]

Moving to P3: we should fix this in V3.1.

Comment by Ken Cavanaugh [ 01/Oct/10 ]

I decided to leave this in 3.1 for now, to remind me to look at it.

Comment by Ken Cavanaugh [ 08/Nov/10 ]

This issue is likely to have changed behavior significantly in GF 3.1.
But I cannot address this in GF 3.1, because for many other reasons,
we don't have a completely clean orderly shutdown. The asadmin command
stop-instance of course shuts down the instance, but it forces the shutdown
with a System.exit or equivalent mechanism due to some unresolved issues
with non-daemon threads.

There are also problems in the EJB container, because the EJB contained
does not shutdown the ORB. It should, and it should to this BEFORE it
undeploys EJBs and marks the container as STOPPED (or whatever the equivalent
is in GF 3.1). Consequently, I am moving this issue to GF 3.2 for further
consideration in the next release.

Comment by Nazrul [ 18/Nov/10 ]

Excluding from 3.1 query

Comment by Ken Cavanaugh [ 23/Feb/11 ]

Moving to orb component (where it belongs).

Generated at Sat Aug 27 21:36:52 UTC 2016 using JIRA 6.2.3#6260-sha1:63ef1d6dac3f4f4d7db4c1effd405ba38ccdc558.