How-to
Configure IP networking on an edge PC with no SSH access
Short answer
A new edge PC that leaves the factory with DHCP and static addressing disabled has no IP address, so SSH cannot connect to it. Do the first configuration at a keyboard and monitor: set X1 to DHCP and the plant-side port to the planned static address, restart networking, then confirm both addresses with nmcli device status and ip a before opening SSH or browser management on port 9090.
On this page
- Why does SSH fail on a brand-new edge computer?
- What do you need at the bench before you start?
- Which port on the back is which interface name?
- How do you choose a static address that works the first time?
- What is the exact sequence at the console?
- How do you prove the address took before you enable remote access?
- When is it safe to walk away from the console?
- Questions asked
An edge PC that refuses SSH almost always has no address yet: this class of device leaves the factory with DHCP and static IP addressing both disabled, so no interface can answer a connection until you configure one locally. Attach a keyboard and monitor, log in at the console, set one port to DHCP and the plant-side port to the static address from your site plan in /etc/network/interfaces, restart networking, and confirm the result with nmcli device status and ip a. The specifics below — the port mapping X1 to enp2s0 and X2 to enp3s0, and browser-based management on port 9090 — are documented for the WAGO 752-9800 Edge Computer; on any other box, run the same sequence but read the interface names from the console instead of assuming them.
Why does SSH fail on a brand-new edge computer? #
The unit leaves the factory with DHCP and static IP addressing both disabled, so no interface holds an address and there is nothing for SSH to connect to; the first configuration has to happen at a keyboard and monitor attached to the device itself.
Ping, SSH and browser-based management all fail for the same reason: with both addressing modes off, the device cannot reply on any network it is cabled to. That is why the vendor's procedure starts at the local console rather than at your desk, and why rebooting the box changes nothing — the address it needs does not exist until you write it.
What do you need at the bench before you start? #
Four things: a monitor on HDMI or DisplayPort, a USB keyboard — the 752-9800 carries four USB ports — the site-approved IP plan in writing, and the factory console account from the vendor's quick-start documentation, which forces a password change at first login.
The IP plan has to state four values for the static side: address, prefix length written as /24 or /25, default gateway, and DNS servers. Collect them from whoever runs the plant network before the device goes on the bench. The vendor's product page for the 752-9800 lists four Ethernet ports alongside HDMI and DisplayPort video, so an ordinary office monitor and any USB keyboard cover the console. This article deliberately does not reprint the factory credential: it travels with the unit's own documentation, and the console forces the change at first login anyway.
Which port on the back is which interface name? #
On the 752-9800 the port labelled X1 is enp2s0 and the port labelled X2 is enp3s0; write those exact names into the configuration, because a stanza that names an interface that does not exist is the most common reason an address never appears.
That mapping is documented for this model, not as a general Linux rule. The unit has four Ethernet ports and the documented pair covers X1 and X2 only; if your plan uses another port, read its name from ip a at the console while a cable is plugged into that port alone. Names come from the operating system image rather than from the metal, so rebuilding the box on a different image can change them — the walkthrough on choosing an operating system image for an industrial PC covers what else that decision moves.
How do you choose a static address that works the first time? #
Pick an address inside the interface's subnet that is not the network address, not the broadcast address, not the gateway, and outside the DHCP pool; in a 192.168.10.0/24 network that leaves 192.168.10.2 through 192.168.10.254, minus the pool and minus 192.168.10.1.
Decide the mode per interface rather than per box: DHCP where the upstream network already runs a DHCP server you do not manage, static on the machine-side network, where PLCs, historians and engineering stations have to find this box at one fixed address.
Worked example — the numbers are illustrative plan values, so substitute your own. Suppose the machine network is 192.168.10.0/24 with the gateway at 192.168.10.1 and a DHCP pool of 192.168.10.100 through 192.168.10.199, and the plan assigns the edge PC 192.168.10.42. The prefix sets the space: 32 - 24 = 8 host bits, and 2^8 = 256 addresses, 192.168.10.0 through 192.168.10.255. The first address names the network and the last is the broadcast address, leaving .1 through .254 for hosts. The gateway takes .1 and the pool takes .100 to .199, so 192.168.10.42 clears every constraint: same subnet as the gateway, unique, outside the pool.
The prefix is what moves the answer, by exactly a factor of two per bit: re-subnet to /25 and the space halves to 2^7 = 128 addresses, with broadcast at .127 and usable hosts .1 through .126. Address .42 still fits, but any pool plan built on 256 addresses does not. Close the loop with whoever runs the DHCP server: the static address must be excluded from the pool or reserved to it. That single entry is what prevents the duplicate-address outage that surfaces weeks later, when the server finally leases your address to another device.
What is the exact sequence at the console? #
The job is six console steps: log in and accept the forced password change, open /etc/network/interfaces in nano, give X1 (enp2s0) a DHCP stanza, give X2 (enp3s0) the static plan values, save with Ctrl-X then Y then Enter, and apply it with systemctl restart networking.
If this box already feeds a running line, do the restart inside a planned stop: systemctl restart networking drops every interface, all at once, for a moment.
- Connect the monitor to HDMI or DisplayPort and the keyboard to USB, power up, and log in as
rootwith the factory password from the unit's documentation. The console forces a password change at first login — set one that meets the plant password policy and record it before going further. - Open the configuration with
nano /etc/network/interfaces. The file included in the image already contains the editable lines for X1 and X2, so change those lines instead of adding stanzas for names you have not confirmed. - Set X1 (
enp2s0) to DHCP, so the office or plant side takes whatever address the upstream DHCP server hands out. - Set X2 (
enp3s0) to static with the values from your plan: the address, a netmask that matches the prefix, and the gateway. - Save with
Ctrl-X, thenY, then Enter to confirm the filename. - Apply the change with
systemctl restart networking.
The static lines use the standard Debian ifupdown keywords, with the worked-example values written out:
# Illustrative values from the worked example above, not a template
iface enp3s0 inet static
address 192.168.10.42
netmask 255.255.255.0
gateway 192.168.10.1
X1 takes the same stanza shape with dhcp in place of static and no address lines under it. If the plan includes DNS servers, the standard ifupdown place for them is a dns-nameservers line in the same stanza; run a name lookup once networking is back to confirm the resolver picked them up.
How do you prove the address took before you enable remote access? #
Two commands decide it: nmcli device status must show both interfaces connected, and ip a must show a lease on enp2s0 and exactly the static address you wrote on enp3s0 — any other result means the file, not the hardware, needs another look.
Read them in that order. An empty enp3s0 almost always means a stanza problem: an interface name that does not exist or a wrong netmask. Reopen the file, correct it, and run systemctl restart networking again. An empty enp2s0 means the console side is fine but the upstream network gave nothing — check that a DHCP server actually serves the segment that port is cabled to. Finish from your workstation by pinging the static address: if it answers at the console but not from the station, the configuration is done and the fault has moved into the network path of cabling, VLANs or firewalls, which is the case the guide to diagnosing an edge gateway that is online while the PLC is offline walks through.
When is it safe to walk away from the console? #
Only after ip a shows both addresses: from that moment you can SSH to the verified address from a terminal emulator and open the device's web-based management in a browser on port 9090, and the keyboard and monitor can go back on the shelf.
Before the cabinet door closes, decide who may reach that management traffic. SSH and a browser page on 9090 are exactly the paths that layered OT security controls exist to segment, so name the VLAN or station list allowed to reach the device and leave every other network without a route to it. The password change from step 1 retired the factory login; segmentation keeps a compromised desk PC from reaching the device at all.
Questions asked
What is the default console password on a WAGO 752-9800 Edge Computer?
The factory credential comes in the vendor's getting-started documentation packed with the unit, and this article does not reprint it. Log in as root with that factory password once; the console forces a password change at first login. Complete the change before editing the network file, and record the new value in the plant password vault so the next engineer finds it.
Can the first network configuration be done over SSH instead of the console?
No. SSH needs an interface with a working address, and until the console session sets one the device has nothing to connect to, so there is no shortcut for the first configuration. Some other hardware offers a USB console or a vendor setup tool, but the keyboard-and-monitor route works on any edge PC with a video output and needs no extra software.
Why does the static address disappear after a reboot?
Because the change was made only in the running system, not in the persistent configuration. Addresses assigned with one-off commands at the shell last until the next boot, while /etc/network/interfaces is what Debian re-applies at startup. The stanza with your static address, netmask and gateway has to be saved in that file, and systemctl restart networking confirms it took.
How do I find the interface names on an edge PC that is not a 752-9800?
Run ip a at the local console: it lists every interface with its name and shows which ones already carry an address. Then cable one port at a time and run it again; the interface that gains a link while its cable is connected is that port. Match names to silkscreen labels by this test rather than by guessing from the name pattern.