Why does Host program accesses DNS server

[Tom_Sz]

I have used the communication wizard to generate the TCL, NPL and HOST part. I am behind a VPN Gateway which I can unplug from the outside world. The RT and the HOST system are plugged in to the VPN. I can start up the RT part (TCL &
NPL) from the host through the RT Project interface. But, when I try to start up the Host part and the internet is busy, I get a 6-8 second time out (a pause of all activity on the host computer). The HOST loop is a one shot which gets restarted in another VI. Thus a possibility of a pause gets greater. The pause is always there if I disconnect the VPN from the outside world. Error Message "Waiting for RT engine to respond", which will time out after 6-8 seconds and the data displayed. On one of the forums it was indicated that the Host program goes to 127.0.0.1 and 192.168.1.1 using DNS lookup. Why is there a DNS lookup?
By the way the VPN's gateway address is 192.168.1.1. The glass pulling process can not afford 6-8 second delays. Is there a fix? File attached may not be complete.

Thanks for your help.

Thomas Szebenyi
Research Technician
Cornell University

[altenbach]

Just create an appropriate entry in the local hosts file.

[Tom_Sz]

I have tried that in both the hosts and lmhosts tables - no luck.

Tom

[Matthew_Kelton]

You can use a program called ethereal to capture the traffic to confirm that the addresses and such that you are assuming is correct.

[TreyB_NI]

As I responded in your email service request, I don't understand the context of VPN here.  You seem to be describing a local network consisting of a router with a host PC and a RT target connected to it.  If this is the case, the RT target should be setup to automatically acquire it's networking address using DHCP.  If it is a hub or switch, you shouldn't need a DNS, in which case you should fill it as 0.0.0.0.

As to why you are getting lag when you're router is busy, I am not sure.  I need to understand your network more thoroughly to troubleshoot this issue.  What platform is your RT target?  If by VPN you are trying to remotely control it, platforms such as cRIO and FieldPoint allow you to supply IP addresses in MAX under remote systems that are allowed add them as remote subnet targets.

If we are not able to get rid of this lag, I would recommend constructing code for an embedded network watchdog.

To other people viewing the forum, TCL and NPL are VIs on the target and the host part is the host part, which have not been attached to this forum topic.

Trey B | Applications Engineering | National Instruments

[Tom_Sz · Accepted solution]

Via Trey B. of NI Support:

"I believe the DNS lookup occurs because the Phar Lap ETS OS on the Real-Time PC likes to reverse DNS lookup things that communicate with it.
You say that your IP addresses are fixed/static.  Have you tried filling out 0.0.0.0 for the RT target?  You can do this in MAX under remote systems, as seen in the attachment.

If you're using TCP VIs, you can force the OS to not resolve remote addresses, as seen here:

http://digital.ni.com/public.nsf/allkb/011F5615859F5694862572580080C501?OpenDocument"

I have set the DNS and gateway on the RT to 0.0.0.0 and most (99%) of the hang up is gone. even with the ethernet to the outside world disconnected. The Host still has a few hangups (ie. something is trying to access the ethernet (call home)). This I still need to investigate.

Thank for your help

Tom