Strange Problem

glr-fti

A system that has been running without issues for years....2 NI3000 masters connected via M2M. System works fine until a command is issued that requires devices or communication from System 2 then the ICSP communication comes to a halt and the buffer never empties. I can login via Telnet and the CPU is up to the 95% range, plenty of memory. I can replicate continuously. I tried switching masters and it doesn't matter. I've tried 3100 and 2100 and get the same result. I also isolated the network to just the two masters and my PC and it still happens so I don't think it is a network issues. Anyone run into such an ordeal?

TurnipTruck

Try watching internal diagnostic messages while running the event that causes the crash. Sometimes the messages will give a hint at the problem.

jjames

Just curious - have you checked all the masters' URL tables?

glr-fti

There are no messages in diag and I have checked URL Just one entry from one master to the other nothing in the 2nd.

mush

Are either of the units trying to communicate on a serial port?

glr-fti

No.

(pad to get enough chars for msg)

mpullin

glr-fti wrote: »

A system that has been running without issues for years....2 NI3000 masters connected via M2M. System works fine until a command is issued that requires devices or communication from System 2 then the ICSP communication comes to a halt and the buffer never empties. I can login via Telnet and the CPU is up to the 95% range, plenty of memory. I can replicate continuously. I tried switching masters and it doesn't matter. I've tried 3100 and 2100 and get the same result. I also isolated the network to just the two masters and my PC and it still happens so I don't think it is a network issues. Anyone run into such an ordeal?

I think I am seeing this too, on a larger system with more than 6 masters. It's like the code locks up, I can see network events in diagnostics but the code clearly isn't running anymore, I have a SEND_STRING 0 every time a button is pushed that tells me what TP and button were pushed, and I don't see the SEND_STRING 0 when I push a button, but I do see it on Push Message Status.

What devices do you have on the 2nd master? What command exactly are you using to reproduce it?

glr-fti

Worked with support this morning and they feel my project was probably just on the line previously and I lucked out. What makes no sense to me though is that I tried using some X100 masters and had the same problem so there is something strange happening still. I had device definitions in the second master that were also in the first and they had me remove from the second system and that has allowed me to continue running. I remember someone mentioning on the forum that you didn't need the devices defined on both but just never cleaned up any old systems. I surely will now!
I sent the workspace to them for review and I hope I learn something new. I may just have to move some of the code off of the main master and put it on the second. But I know I have a lot of larger systems than this running on one master without issue.

vining

glr-fti wrote: »

I had device definitions in the second master that were also in the first and they had me remove from the second system and that has allowed me to continue running. I remember someone mentioning on the forum that you didn't need the devices defined on both but just never cleaned up any old systems.

You had all your devices defined in both systems/master or just the devices that are common to both systems defined in both?

With out defining common devices in both systems M2M can't work.

TurnipTruck

Double check that master URLs are only in one URL list.

On most systems, I typically consider one of the masters to be the "Main." I put the URLs of all of the other masters in its URL list. Nothing is in the URL lists of the other masters.

Also, by default, masters are set to ROUTE_MODE_NORMAL. in this mode, a remote system can reroute to others through others. In most cases, I find this unnesscessary. I set my systems to "ROUTE_MODE_DIRECT." It seems to cause less buffering if the connection to a master is poor.

ozgun

mpullin wrote: »

I think I am seeing this too, on a larger system with more than 6 masters. It's like the code locks up, I can see network events in diagnostics but the code clearly isn't running anymore, I have a SEND_STRING 0 every time a button is pushed that tells me what TP and button were pushed, and I don't see the SEND_STRING 0 when I push a button, but I do see it on Push Message Status.

What devices do you have on the 2nd master? What command exactly are you using to reproduce it?

This is my exact problem for now and I realy hard to find this problem..I have two Ni and 3 panels in system all of them is online and connected to my master ni..After a while our Ni is not working good..and we have to reset it..and it working normally again..

Joe Hebert

vining wrote: »

With out defining common devices in both systems M2M can't work.

That’s true if you are referring to virtual devices. Real devices don’t have to be defined in both systems for M2M to work. For example I can run code in system 1 that controls the IR ports and serial ports in system 2 even if system 2 has a completely blank program. I do it all the time.

TurnipTruck wrote: »

I set my systems to "ROUTE_MODE_DIRECT." It seems to cause less buffering if the connection to a master is poor.

Same here. I also do a *set udp bc rate* and set it to 0.

glr-fti

I only had the real devices connected on system 2 defined in both. Removing them has allowed the program to run but there is still an underlying problem that will show up again if left alone that I am reviewing. This is my test system at my house so there is lots of crap that never gets properly cleaned up and lots of hardware set up that may or may not be attached at any given time.