Problem with cluster stateful session bean example

Hi,

I'm trying to get the cluster stateful session bean example to work on my

cluster with iAS 6.0SP3 on Solaris.

I installed a clean cluster environment. The first sample-app, the

cluster-servlet works fine. The other, the cluster-sfsb appears to work

fine, too. But after sending about 20 requests, I get no more respons from

iAS. Not even from a browser on another PC. I increased the logging and was

able to see in kjs.log that the request does arrive at iAS, but no response

is generated. The only remedy is to restart iAS. The last lines in kjs.log

are (with DSync logging options enabled):

CUT-

***** GetRootToken: *****

***** GetNode(/gxsessroot) *****

ReaderStart: Thr. 0xb60758, for /gxsessroot(pNode=0)

ReaderStart: Thr. 0xb60758 succeeded; path=/gxsessroot(pNode=1039058),

rcnt=1, wcnt=0

ReaderStart: Thr. 0xb60758 exiting; path=/gxsessroot(pNode=1039058), rcnt=1,

wcnt=0, hr=0, gotIt=1

RdrWrFin: thread=0xb60758, path=/gxsessroot, mode=0

***** GetNode(/gxsessroot/) *****

ReaderStart: Thr. 0xb60758, for /gxsessroot/(pNode=0)

ReaderStart: Thr. 0xb60758 succeeded; path=/gxsessroot/(pNode=101b5a0),

rcnt=1,wcnt=0

ReaderStart: Thr. 0xb60758 exiting; path=/gxsessroot/(pNode=101b5a0),

rcnt=1, wcnt=0, hr=0, gotIt=0

RdrWrFin: thread=0xb60758, path=/gxsessroot/, mode=0

***** GetNode(/gxsessroot//6f334808835cbefc) *****

ReaderStart: Thr. 0xb60758, for /gxsessroot//6f334808835cbefc(pNode=0)

ReaderStart: Thr. 0xb60758 blocking for

/gxsessroot//6f334808835cbefc(pNode=116ff48, status=0x0); rcnt=0, wcnt=1,

owner=b60ee0

CUT-

I especially noticed the last line saying ".. blocking ..", where other

similar lines say something like ".. succeeded ..".

Any help, please?

Peter.

[1919 byte] By [709281] at [2007-11-25 4:58:10]
# 1

I don't know the root cause but I've seen the problem as well on sp3.

Distributed access to SFSBs works fine on sp3. However, failover in a multi-KXS cluster seems to hang the cluster. For me, the hand occurred on the next request for the SFSB after the failover.

Failover with a single KXS and multiple KJSs works fine with sp3 - I show it in the iAS classes I teach for Sun.

This problem seems to be fixed in sp4. I have a small sample that fails-over just fine in a multi server cluster with sp4.

709282 at 2007-6-29 9:13:37 > top of Java-index,Application & Integration Servers,Application Servers...