[Ace-users] [tao-users] LOCATION FORWARD and _is_a() loop

Christian Fromme christian at ast.dfs.de
Tue Aug 14 05:22:41 CDT 2007


Hi List,

TAO VERSION: 1.5
ACE VERSION: 5.5

HOST MACHINE and OPERATING SYSTEM:
Intel(R) Xeon(R) CPU 5130  @ 2.00GHz
Linux 2.6.16.21-0.8-smp #1 SMP i386 GNU/Linux

TARGET MACHINE and OPERATING SYSTEM, if different from HOST:
COMPILER NAME AND VERSION (AND PATCHLEVEL):
gcc version 3.4.3

DOES THE PROBLEM AFFECT:
    COMPILATION?
        No.
    LINKING?
        No.
    EXECUTION?
        Yes.

SYNOPSIS:
Location Forward and _is_a seem to loop over two hosts

DESCRIPTION:
Here is the setup:

Two hosts, both basically running the same set of applications (A
bunch of CORBA-servers and clients). One of the CORBA-servers
running on both hosts is a management and control application (from
now on: "mac-server") which clients can use to retrieve status-info 
about the whole system.

The mac-server is implemented in a way that lets you retrieve
status info about any of the two physical machines via any of the
two mac-servers. (That means you can ask mac-server on host A about
the status of the application on host B, and vice versa) Both
mac-servers offer the same CORBA interface. If you ask one about the
state of the other, it will internally try to resolve the object
reference to its peer mac-server and ask him using the same
interface that someone just used to call him. (Hope it is clear what
I mean here)

Now, on certain conditions (which I not isolated completely yet), I
see a loop between the two mac-servers:
I ask server A about the status of server B and the ORB of
mac-server on server A starts the request by sending a _is_a()
request to server B. Server B then -- for some reason I don't
understand -- sends a LOCATION FORWARD back to server A, which again
sends _is_a() to server B, and so on and on and on.

After about 4500 requests in about 3 seconds time, one of the 
mac-servers dies with segmentation fault.

My question: Is there a way to completely disable LOCATION FORWARDs
and/or do you have any idea where this might come from?

SAMPLE FIX/WORKAROUND:
None.

I can send the 3.5MB pcap logfile if anyone is interested.

Thanks in advance,
Christian Fromme



More information about the Ace-users mailing list