0

I'm having an issue with getting a device to connect to a Redhawk SDR domain. I've tried following answer questions to similar questions like this one issues-with-running-a-second-node-on-another-processor but can't work it out. When I enter the command nameclt list on the device machine I get the following response:

Caught a TRANSIENT exception when trying to validate the type of the 
NamingContext. Is the naming service running?

I'm running the domain manager on a VM with CentOS 6.8 x64 with Redhawk 2.0.2. The device is also on a VM, cloned from the first one. The two VMs are on separate hosts connected by Ethernet LAN. I've configured the hosts and VMs to bridge to their respective Ethernet port and can ping all 4 machines. The device VM is 192.168.1.2. The device host is 192.168.1.5. The domain host is 192.168.1.6 and the domain VM is 192.168.1.3.

My omniORB.cfg on the device VM is as follows:

InitRef = NameService=corbaname::192.168.1.3
InitRef = EventService=corbaloc::192.168.1.3:11169/omniEvents
supportBootstrapAgent = 1
endPoint = giop:unix:

My omniORB.cfg on the domain VM is as follows:

InitRef = NameService=corbaname::127.0.0.1
InitRef = EventService=corbaloc::127.0.0.1:11169/omniEvents
supportBootstrapAgent = 1
endPoint = giop:unix:

Edit: For more information, the two hosts are connected directly together between their two Ethernet ports. There is no router or switch. The IPs for the hosts are statically assigned.

Community
  • 1
  • 1
ARob
  • 11
  • 2
  • Could be more of a network problem than REDHAWK configuration issue. Can you confirm you can ping from one host to another? Can you confirm that you can run "telnet 192.168.1.3 11169" and do not get a "Connection refused" on either host? (If you do it's likely a firewall issue) – Youssef Bagoulla Dec 12 '16 at 12:36
  • Also, you have listed the giop:unix endpoint which will cause omniORB to only listen on the unix pipe endpoint, remove that line from both configuration files and it will fall back to using the default TCP/IP endpoint. If you have more than 1 network interface, you may have to specify the IP address in the endpoint like this: endPoint = giop:tcp:1.2.3.4: where you would replace 1.2.3.4 with the IP of the host. – Youssef Bagoulla Dec 12 '16 at 16:14
  • I had a similar problem and it WAS a firewall issues as you suggest. I could ping and could verify that the message from nameclt to the Naming Service was received but the reply could not get back to the requesting machine due to its firewall not permitting corbaloc, corba-iiop and corba-iiop-ssl (ports 2809, 683 and 684). Either turn off the firewall or configure it to accept messages to those ports. – Terry L Anderson Dec 12 '16 at 19:36
  • Terry, the firewall issues was the trick! Thanks. – ARob Dec 13 '16 at 05:56
  • There should be some clarification added to manual about this. It currently states: Edit: code block `Note: omniORB may have trouble automatically resolving its location. In this case, it may be necessary to set the endpoints in the omniORB.cfg les by adding the following to each omniORB.cfg le: endpoint = giop:tcp:. You must restart omniEvents and omniNames for these changes to take effect.` I've also seen on similar questions to mine where it was stated that adding the :unix: endPoint would improve performance. – ARob Dec 13 '16 at 06:03

0 Answers0