Home AMX User Forum AMX General Discussion

Problem with NI after reboot

In our home system we have two masters, NI2000 and NI3000. When rebooted, there is no problem connecting to the masters by IP. But after a while you cannot make that contact anymore. And then the whole system neds to be rebooted. It is also a problem with the wireless TP. It does not connect with the master anymore.
Anyone an idea what the problem could be?

Comments

  • You've probably got something eating memory. If your running DUET firmware check that the SET DUET parameter is at its lowest (2M) - done through Telnet. Otherwise you've likely got a loop in the code - which is really easy to do.
  • ChristelChristel Posts: 9
    No source code

    Thanks icraigie. Will have a look later on at home about the Telnet .
    Unfortunately this system was installed in 2005 by another firm ( we were involved with AMX then), and since then it has not been completely finished, and there are still some problems, contradictions etc.
    After doing the installer course, I had a look for the source code, and no luck, only machine code loaded, and we do not have the source code. So I cannot do anything there.
    We have requested that everything will be put up to date and that source code will be give, but they say it is for protection of their "copyright". So am a bit bound by both hands. Still to early for me to start from scratch integrating our whole home. Will probably do that after I have MUCH more experience
  • One more thing which can end in such effects is the URL list in the masters. If the master's IP is in the list of that master itself. This will create a loopback and will lockup that master.
  • ChristelChristel Posts: 9
    Have tried to get in via Telnet but simply cannot get into the masters.So cannot see the duet parameter.
  • Christel wrote:
    Have tried to get in via Telnet but simply cannot get into the masters.So cannot see the duet parameter.
    If you are on site, try the serial programming port. If this also doesn't work, set Switch 1 of the DipSwitches to ON and reboot the system. This will launch the operating system only, but not the code written.
  • ericmedleyericmedley Posts: 4,177
    Here are some things to check. If the two masters are meant to communicate with each other (Master to Master) then you can check the URL table in each one. Only one should be linked to the other. For example:

    NI-2000s URL table contains the IP address of the NI-3000. However the NI-3000 should not be connected back to the NI-2000. The two masters don't need to be looped back. The one connection will suffice. Two starts causing collisions that begin to clog the error handling and event logs.

    another clue is look at the output light. If it is on solid red (not occasionally blinking) you may have some programming issues. You probably have some statements in your Define_program section that are being updated on every pass. that eats up processor time.

    Hope that helps
  • Spire_JeffSpire_Jeff Posts: 1,917
    A few more things to check:

    Verify that the network wires are terminated using the TIA-568A or TIA-568B. It is very important especially when running 100bt network speeds. This is even more important to verify since you inherited this job and you don't know how the other people terminated their wiring. I have seen a lot of wires terminated as follows:
    Pin / Color
    1/Blue
    2/White/Blue(or Striped Blue)
    3/Orange
    4/White/Orange(or Striped Orange)
    5/Green
    6/White/Green(or Striped Green)
    7/Brown
    8/White/Brown(or Striped Brown)

    1/Brown
    2/Green
    3/Orange
    4/Blue
    5/White/Blue(or Striped Blue)
    6/White/Orange(or Striped Orange)
    7/White/Green(or Striped Green)
    8/White/Brown(or Striped Brown)


    Neither of these standards maintain communications across twisted pairs. The network lights will light up but the link will have problems based on distance and interference. These bad links will often slow a network to a grinding halt with retransmissions of packets. Also, verify that the twist is maintained until the wires reach the terminator.

    Another thing to check is verify that your masters are set to ETHERNET MODE 10 HALF. I'm not sure that this is still an issue, but I knew that it caused problems in the past if processors weren't limited to this.

    Also, you can telnet in (or use the serial port) and watch the free memory. (SHOW MEM will work) see if the free memory is slowly counting down to nothing. This would indicate a problem in the code.

    Hope some of these suggestions help.

    Jeff
  • ChristelChristel Posts: 9
    One step further

    After all suggestions here, we have tried them out.
    The situation is now as follows:

    After rebooting both masters we can follow them up in Telnet. Gradually their memory is decreasing into telnet until access to the master blocks.
    This is a recent event. (Because everything was working !)
    Via Comm-Tec( AMX reps in Spain) I have been informed that we should upgrade the firmware of the masters to the following version
    2105-04-NI-X100-MASTER-V-21-348ipstartupHfix.kit
    (Sounded a bit weird, since our masters are NI 2000 and NI 3000, not NI2100, NI3100)
    So tried this but no succes, get a message that there is not enough memory.
    The current firmware in the Masters is version 3.21.343

    Suggestions are welcome!!! ::))

    Another thing is that after reboot and going in via Netlinx Studio we try to get the IP info and have the master listen to get the URL list. One Master cnnot find the URL listing and is always attempting connection. Is this normal?
  • There are also hotfixes available for NI-x000 and NI-x00.

    Can you get connect on programming port when telnet is not working any longer? If yes, please try the terminal command SHOW MEM. I'm interested in the Non Volatile memory. Do you have large arrays which are not defined VOLATILE?

    Will this memory loss also happen if you disable runmode on both masters? (set Dipswitch 1 to ON and reboot the master. Only operating system is started now, but not the program you wrote)
  • ChristelChristel Posts: 9
    Nonvolatile Mem

    Marc,
    We have tried what you suggested
    The following shows up on telnet when command SHOW MEM is given:

    NI 2000 :
    Volatile free - 8188328
    Non Volatile free : 1047144
    Disk free : 26324992
    Duet Mem free : 981256

    When you re-enter the SHOW MEM command on this master , the volatile mem decreases ( it does not on the NI 3000)

    NI 3000:
    Volatile free - 9223964
    Non Volatile free : 1040256
    Disk free : 26304512
    Duet Mem free : 938468

    Before we try setting the dipswitch 1 to ON position, please know we do not have source code ( will this be a problem ?)

    Already thanks for all help on this forum!!!
  • DHawthorneDHawthorne Posts: 4,584
    Switch 1 only disables the program, it doesn't wipe it or delete it. However, if you don't have the source code, fixing a memory leak is going to be a problem. Another thing you can try is to use an FTP client to connect to the problem master fresh after a reboot and look what files are on the flash drive. The program may be storing something on the drive that is getting bigger and bigger. You can also just type "list" from the telnet session, but that won't let you delete anything that obviously shouldn't be there.
  • ericmedleyericmedley Posts: 4,177
    Before we try setting the dipswitch 1 to ON position, please know we do not have source code ( will this be a problem ?)

    Already thanks for all help on this forum!!!

    Since you do not have the source code I would assume that you don't have the original programming files. I would then argue that you definitely have something going on in the program that's causing the memory leak or overrun in the error logs.

    There are almost limitless ways to write code that creates huge cascading errors.

    Not having the code is going to make it nearly impossible to correct the problem. Otherwise you'll just be poking around a big black box while being blind as a bat.

    You should probably take steps to get to the original code.
Sign In or Register to comment.