[omniORB] LocateRequest gets dropped on RHEL 6.1 system

[Brad Fawcett]

Hello,

I am seeing a problem with running my code using CORBA (omniORB - 4.1.4) on
a RHEL 6.1 system.

Problem scenario is this:

1.) invoke the CORBA request
2.) this causes a LOCATE_REQUEST to be sent out
3.) TCP ACK is sent back
	Note:  looking through the CORBA server traces.  this LOCATE_REQUEST
is never received there.   However, a suspicious looking
                                  one is received in its place.
4.) but no LOCATE_REPLY is sent back
5.)  then after a while (looks like a 3 minute time-out), the
LOCATE_REQUEST is sent out again.  (on a different port)
6.) this time the LOCATE_REPLY is sent back.

=======================  relevant TCPdump results ==================
349	18.856754	9.5.167.92	9.5.167.92	GIOP	GIOP 1.2 LocateRequest s=26
id=20 op=LocateRequest
0000  00 00 00 00 00 00 00 00  00 00 00 00 08 00 45 00   ........ ......E.
0010  00 5a 68 92 40 00 40 06  71 49 09 05 a7 5c 09 05   .Zh. at .@. qI...\..
0020  a7 5c cf d6 8c d5 3c 86  94 47 3c 63 0b b9 80 18   .\....<. .G<c....
0030  01 09 61 0f 00 00 01 01  08 0a 14 1a e9 34 14 1a   ..a..... .....4..
0040  e7 3b 47 49 4f 50 01 02  01 03 1a 00 00 00 14 00   .;GIOP.. ........
0050  00 00 00 00 00 00 0e 00  00 00 fe ea 91 66 4e 00   ........ .....fN.
0060  00 37 53 00 00 00 00 00                            .7S.....

350	18.856783	9.5.167.92	9.5.167.92	TCP	36053 > 53206 [ACK] Seq=299
Ack=1469 Win=38144 Len=0 TSV=337307956 TSER=337307956
0000  00 00 00 00 00 00 00 00  00 00 00 00 08 00 45 00   ........ ......E.
0010  00 34 ec c5 40 00 40 06  ed 3b 09 05 a7 5c 09 05   .4.. at .@. .;...\..
0020  a7 5c 8c d5 cf d6 3c 63  0b b9 3c 86 94 6d 80 10   .\....<c ..<..m..
0030  01 2a a4 76 00 00 01 01  08 0a 14 1a e9 34 14 1a   .*.v.... .....4..
0040  e9 34                                              .4

// Locate_Reply never is sent.

******  Also, the suspect GIOP message shown below in the CORBA trace never
appears in the TCP tracelog.

============== CORBA server-side daemon trace ===================
taken around the time that the Locate_Request is expected to be received.

omniORB: 2011-09-06 16:34:52.667839: SocketCollection idle. Sleeping.
omniORB: 2011-09-06 16:34:52.670531: inputMessage: from
giop:tcp:[::ffff:9.5.167.92]:53206 38 bytes
omniORB: 2011-09-06 16:34:52.670551:
4749 4f50 0102 0100 0501 0000 1200 0000      GIOP............
0000 0000 0000 0000 0e00 0000 feea 9166      ...............f
4e00 0037 5300
N..7S.
omniORB: 2011-09-06 16:34:55.919786: Scan for idle connections
(1315344895,919689000)
omniORB: 2011-09-06 16:34:55.919816: Scavenger reduce idle count for strand
0x7f70f80009b0 to 35
omniORB: 2011-09-06 16:34:55.919840: Scavenger reduce idle count for strand
0x7f70f80016b0 to 35

Does this Message look suspicious?

This problem is very consistent.  Full traces are easily available.

This test runs successfully everytime on RHEL 5.5 system, but fails
everytime on a RHEL 6.1 system.   (Should qualify this some, in that
machine type also is important here.   It  fails everytime on a RHEL 6.1 /
HS21 IBM bladesystem  but works everytime on RHEL 6.1 / HS22 IBM
bladesystem.  This behavior has been verified on several machines of the
various types, & was very consistent.)

Best Regards,

Brad Fawcett
bfawcett at us.ibm.com
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://www.omniorb-support.com/pipermail/omniorb-list/attachments/20110907/cadd144d/attachment.htm