Is new InitialContext supposed to be thread safe? It doesn't seem to be...
I have a client test program that creates multiple threads, each of which
does a getInitialContext. This fails about 2 times out of three, usually
with a:
ObjectIdMgrTest:-- Getting Initial Context --
javax.naming.NamingException: Error reading application-client descriptor:
Error
looking up EJBHome: Lookup error: java.io.EOFException; nested exception is:
java.io.EOFException
at
com.evermind.server.ApplicationClientInitialContextFactory.getInitial
Context(JAX)
at
javax.naming.spi.NamingManager.getInitialContext(NamingManager.java:6
68)
at
javax.naming.InitialContext.getDefaultInitCtx(InitialContext.java:246
)
at javax.naming.InitialContext.init(InitialContext.java:222)
at javax.naming.InitialContext.<init>(InitialContext.java:178)
at
sbd.sd.ejb.sb.objectidmgr.ObjectIdMgrTest.getInitialContext(ObjectIdM
grTest.java:305)
at
sbd.sd.ejb.sb.objectidmgr.ObjectIdMgrTest.run(ObjectIdMgrTest.java:16
1)
at java.lang.Thread.run(Thread.java:484)
occasionally it will hang instead. This is on W98, SDK 1.3.0, Orion 1.4.4.
The test is straightforward; in pseudo code...
private InitialContext iCtx;
private final static String testName = "name of test";
main()
loop: create new thread, start
run()
this.getInitialContext()
hammer away...
getInitialContext()
this.iCtx = new InitialContext()
to get around this I modified getInitialContext to
synchronized (testName)
this.iCtx = new InitialContext()
This works when tried a dozen or so times.
Is this an Orion bug? If not, how should it be handled?
Kirk Yarina
[EMAIL PROTECTED]
