This Bugzilla instance is a read-only archive of historic NetBeans bug reports. To report a bug in NetBeans please follow the project's instructions for reporting issues.

Bug 27594 - Reloading RMI registry crash JVM
Summary: Reloading RMI registry crash JVM
Status: CLOSED FIXED
Alias: None
Product: ide
Classification: Unclassified
Component: Internal Server (show other bugs)
Version: 3.x
Hardware: PC Linux
: P1 blocker (vote)
Assignee: issues@obsolete
URL:
Keywords:
Depends on:
Blocks:
 
Reported: 2002-09-25 18:25 UTC by pknakal
Modified: 2003-06-10 17:42 UTC (History)
1 user (show)

See Also:
Issue Type: DEFECT
Exception Reporter:

Attachments
JVM Crash Log (16.84 KB, text/plain)
2002-09-25 19:33 UTC, Tomas Zezula 		Details
Add an attachment (proposed patch, testcase, etc.)

Note You need to log in before you can comment on or make changes to this bug.
Description pknakal 2002-09-25 18:25:23 UTC 
JDK 1.4.1_fcs, NetBeans_dev_2002090250100

How to reproduce
1) start NetBeans
2) create RMI server
3) start local RMI registry
4) start RMI server
5) stop local RMI registry
6) Terminate RMI server
7) start local RMI registry

now JVM crash with internal JVM error

Something similar happend on Solaris 8 (same
version of JDK and same NB build). Difference is
that JVM not crash but is inpossible to browse RMI
registry.

There is not difference between using RMI Executor
with codebase as URL or file:///
Comment 1 Tomas Zezula 2002-09-25 19:33:31 UTC 
Created attachment 7518 [details]
JVM Crash Log
Comment 2 Tomas Zezula 2002-09-25 19:39:10 UTC 
Indeed this in an JVM Error.
JVM crashes in the BSD Socket's accept JNI call in one of the Tomcat
threads.

See attached JVM Crash Log.
When RMI Executor is changed to use file protocol instead of http
protocol everything works fine.

The workaround is to use local files for java.rmi.server.codebase and
java.security.policy.
Comment 3 Martin Balin 2002-09-26 08:24:49 UTC 
We already had such a problem in the past reported under Issuezilla
#21022. There was also bug reported against JVM #4623152 related to
this problem. This JVM one is marked as fixed for J2SE 1.4.1. Now we
are experiencing this problem again, so I recommend Tomas to fill a
new JVM bug or reopen the older one.
Comment 4 _ rkubacki 2002-09-30 13:44:57 UTC 
Bug 4623152 mentioned by Martin was problem in classloading that hid
another crash in sockets - I reported part related to sockets as
4680160. Tomas, can you look if it is our case?
Comment 5 Tomas Zezula 2002-10-08 10:46:59 UTC 
The crash is caused by lazy initialization of http server from
executor, threads of the http server belongs to the executor
ThreadGroup. When the process is killed, the core part of execution
engine calls depricated stop method on the ThreadGroup. Implies that
the http threads are killed. Some of the threads are in JNI
socketAccept call, which was not ThreadDeath aware in JDK 1.4.1 and
1.4.0 and ThreadDeath causes JVM crash.
Both parts of this bug (JDK and NetBeabs) are fixed.
Comment 6 pknakal 2003-02-07 08:05:48 UTC 
Verified.
Comment 7 pknakal 2003-06-10 17:42:44 UTC 
Closed. Hope this is OK ;-)