Help please!!! osfind stuck???
SZ
I am currently using ORBeline 2.0. I defined two hosts (host A and B
in different subnet) as one domain by assigning a unique OSAGENT_PORT
number to these two hosts. What I want to do is to use "osfind" in
host A to find the object which is registered in host B with its local
oad. I did everything as root in both host A and B.
Host network hardware config: host A has one NIC and host B has 2 NICs.
It appears to me that the osagent in host A and B can talk each other, but
once the client (osfind) on host A got the ip of host B and tried to connect
the oad on host B, it stuck for 2 minutes and then came up some error message
as shown in the following. Do you know if there is a trick because host B
has two NICs and two IPs? If I telnet to host B and do a local osfind, it does
show its osagent running on a different IP which host A can not reach. Does
that matter? Do you know what UDP/TCP port oad is listening to? Does oad
listen to all the interfaces?
Since I am not quite familar with Corba, Please, please help me with this
problem.
Thanks a lot.
-SZ
Here is what I did:
I started osagent and oad in host B as normal:
#setenv OSAGENT_PORT 12345
#osagent&
#oad&
Then I have one of the object (if name: obj1) register with the oad using
regobj on host B.
In host A, since it is not in the same net with host B, I started the osagent
as following:
osagent -ORBAdress host_B_IP1 -p 12345
Note, that host_B_IP1 (101.44.2.134) is one of the two IP addresses
that host B has. Host A can reach host B only with host_B_IP1. I also
set OSAGENT_PORT=12345 ORBELINE=/tmp on host A.
On host A, I have problem on osfind:
----------------------------------------
host_A:/tmp> osfind -i obj1 -a
osfind: Found 2 agents at port 12345
HOST: sms032
HOST: 101.44.2.134 (Note: this is host_B_IP1 )
(here it waits 2 minutes...)
osfind: There are no implementations currently running/ registered with
OAD
host_A:/tmp> osfind -i obj1 -o
(here it waits 2 minutes...)
osfind: There are no OADs running on in your domain.
osfind: There are no implementations currently running/ registered with
OAD
host_A:/tmp> listimpl
listimpl: Unable to bind to OAD.
Exception: CORBA::NO_IMPLEMENT
Minor: 0
Completion Status: NO
TO prove host B did have osagent and oad running properly, I telnet into host
B and used osfind. It came up fine:
--------------------------------------
host_B:/> osfind -i obj1
osfind: Following are the list of Implementations registered with OADs.
HOST: 101.45.100.100 (note: this is the other IP address host B has
and host A can not reach this IP)
INTERFACE NAME: obj1
OBJECT NAME: obj_n1
host_B:/>
host_B:/> listimpl
Interface Name:"obj1"
Object Name:"obj_n1"
Path:"/pace/dbase/ce_bin/sdServer.exe"
Policy:UNSHARED_SERVER
ReferenceData:"NONE"
Args:"SD_APPS=PM+CCD, SD_APPS=PM+CCDTRACE_LEVEL=1,
TRACE_LEVEL=1DB_APPL_ID=100, DB_APPL_ID=100NODE_TYPE=SCP,
NODE_TYPE=SCPBROADCAST_RECVR_CE=SQLIM0,
BROADCAST_RECVR_CE=SQLIM0SLP_START_PATH=/pace/sli/StartupProcess.exe,
SLP_START_PATH=/pace/sli/StartupProcess.exeSLP_TYPE=SLP,
SLP_TYPE=SLPREMOVE_SP=SCPConfig..sdsp_removeservice,
REMOVE_SP=SCPConfig..sdsp_removeserviceSLI_APPL_NAME=SLI,
SLI_APPL_NAME=SLI"
Env:"NONE"
host_B:/>
John Liptak
SZ <shan...@yahoo.com> wrote:
>Hi, all,
>
>I am currently using ORBeline 2.0. I defined two hosts (host A and B
>in different subnet) as one domain by assigning a unique OSAGENT_PORT
>number to these two hosts. What I want to do is to use "osfind" in
>host A to find the object which is registered in host B with its local
>oad. I did everything as root in both host A and B.
Visibroker 3 (the successor to ORBeline 2) had a problem with multi-homed
HP boxes (and perhaps others as well) because it performed a connect on
the UDP connection between the client and server.
This means that if the osagent host sent it's responce back to the osfind
process from an IP that was different than was expected, it will fail.
This was fixed in Visibroker 3.3 or 3.4 (I don't remember, it been a
few years).
You may have to upgrade to a newer version of the software.
John