Configure Tx2 on linux server

Hi,

I am using Jetson Tx2 board in my application, I want to Mount Tx2 on linux server.
Can you help me in connecting my linux server with Tx2 board, like seequence of ssh commands to start communicating with tx2 from
linux terminal.

Thanks,
Nagendar

It isn’t too difficult, but here is some background on the topic since I don’t know what you actually know already.

Every networked computer has an address. The raw form is “dotted-decimal”. This form looks something like “192.168.1.2”.

Optionally a name may be assigned to a dotted-decimal address. This is be done by a DNS server (such as public sites known all over, e.g., “google.com” or “nvidia.com”), or can be known only to a few machines on your local network. If used only on a few machines on the local network, then it is the router which provides this name-to-dotted-decimal conversion. Under Linux you can also assign a name-to-dotted-decimal conversion in the file “/etc/hosts”, e.g., you could name your desktop PC “myhost”, and then either ping it by dotted-decimal address or “ping myhost”.

By default the Jetson gets its dotted-decimal address from a router by broadcasting what is known as a DHCP query. The router gives a DHCP response. The DHCP query/response could assign a named address for machines on the local network. This tends to be “linux-tegra”. If this works, then you can “ping linux-tegra”. This is customizable and not guaranteed. Working with the dotted-decimal address is universal and guaranteed.

So your desktop PC can act as a router, or you can have a hardware router appliance. Either way there will be some sort of log of the DHCP request, along with the dotted-decimal address assigned. The Jetson itself will see its own address by running the command “ifconfig”. ifconfig will list a number of network devices, but only one will be the integrated hard-wired ethernet port. You can ignore any “lo” or “wlan” interface if you are looking at the wired port. Usually this is “eth0”, although naming can change. The second line of the ifconfig for eth0 should show “inet addr:…”. If this exists and shows a dotted-decimal address, then any machine on your local network can use ssh and talk to that address. Write that address down.

Now on your other computer which you run Linux on and wish to use to talk to the Jetson, type in the following command (I’m using “192.168.1.2” as an example, you have to substitute with the actual address or a named address):

ssh 192.168.1.2

You’ll be asked for a login name, then a password. By default Jetsons have accounts “nvidia” with password “nvidia”, and account “ubuntu” with password “ubuntu”. These are administrative accounts and can be used for some privileged commands. If your network has computers on it other than ones you control, then you should change the passwords of these accounts very quickly to avoid a security issue. You can change those passwords with a local keyboard/monitor prior to connecting to the network if you want.

An abbreviated command under ssh to log in specifically to the ubuntu account:

ssh nvidia@192.168.1.2

Under a Linux host you can forward graphical commands to display on your local computer instead of on the Jetson with the “-X” or “-Y” option, e.g.:

ssh -Y nvidia@192.168.1.2
# ...after login run a GUI command, e.g., "xterm":
xterm

You can run non-privileged commands without logging in, e.g., this runs “ls” to list the login directory files:

ssh ubuntu@192.168.1.2 ls

…or this opens an xterm window without going to a shell:

ssh -Y ubuntu@192.168.1.2 xterm &

NOTE: If you use “-X” or “-Y” to display a remote graphical application, then the GPU work is offloaded to the PC and won’t really be running on the Jetson. If you need to display remote graphical content on the local PC while the Jetson’s GPU is used you’ll need to use a virtual desktop (which is a whole new topic).