Attention

You are viewing an older version of the documentation. The latest version is v3.3.

NGSF Troubleshooting

Clock Synchronization

  1. Make sure the clocks are close to each other before starting the synchronization.

In the setups with switches, set the Industrial PCs (IPCs) clock to align with the switch clock.

To do so, get the switch time, log in remotely on the switch, and run the following command:

date

The result will be similar to:

Tue Oct 10 22:55:01 PDT 2017

Use this string in the connection to the IPCs and set their clocks as:

date -s “Tue Oct 10 22:55:01 PDT 2017”

  1. In Ground 0 – NGSF Foundation, you might need to edit the gptp.conf file for the Primary controller to change the priority of the clock. The algorithm that chooses the best clock on the network will be the one with the lower priority setting.

For example, if the gptp.conf file looks similar to the following figure, it is recommended that you change both priorities to 124 for the clock that is to be the main clock.

../../_images/gptp-conf.png

SSH and Time Sensitive Configuration

  1. During the configuration of Time Sensitive Networking, it can happen that after the configuration, the SSH is closed and cannot be opened anymore.

    If this happens, the alternative is to have a second connection for the operation and control of the IPC as a workaround. There is no final solution for the issue at the time.

  2. Interface names change after disconnecting and connecting the cable or after reboot.

    New name has to be updated in the ngsf.json file and in the config file of the network.

    Keeping notes of the interface names helps in resolving these issues.

Large Oscillating Numbers Appear on End-to-end Latency on Dashboard

  1. If the setup has a switch, it is possible that the switch is not configured for the default “pass all” setting.

  2. If there is no switch, it is possible that the Time Sensitive Network is not properly configured on Publisher or Subscriber. Review the tcpdump output and look for the URLs used by the ngsf.json file for the communication that is expected to be seen on the given interface. Start with the publisher interface on the publisher side and then review the subscriber interface. If the configuration was done correctly, and still the messages do not arrive, reconfigure the network interface, including stopping the clock synchronization. If the error persists, reboot the problematic component.

Negative Numbers Appear on End-to-end Latency on Dashboard

This issue is typically linked to clock synchronization. Try the following:

  • Check the ptp4l and phc2sys clocks.

  • If a very short cable is used, it might be possible that the algorithm for synchronization is unable to do the right correction. A longer cable (about 1m in length) should not have this issue.

Connmanctrl is unavailable in Edge Controls for Industrial Installation

For the Debian-based ECI flavors, the configuration of the network is done via /etc/network/interfaces.

A typical entry for configuring a device in this file is:

auto eth0
iface eth0 inet static
    address 10.0.0.101/24