Network troubleshooting: the step-by-step method with commands
Good support engineers don't guess, they narrow things down. The order below works through the network from the bottom up. Every step halves the possible causes.
1. Scope the problem
One user or everyone? One website or all of them? Since when? Two questions often save half the work: if everyone is affected, the fault is central (router, DNS, firewall), not the single PC.
2. IP configuration
Does the machine have a valid address, the right subnet mask, a gateway and a DNS server? A 169.254 address means DHCP is missing.
$ ipconfig /allWindows$ ip -br addr && ip routeLinux3. Gateway and internet
Ping your own gateway first, then an address on the internet. Where the chain breaks, that's where the fault is. tracert or traceroute shows at which hop the connection ends.
$ ping 192.168.10.1your own gateway$ tracert 8.8.8.8Windows; Linux: traceroute -n 8.8.8.84. Name resolution
If IP addresses work but names don't, DNS is the cause. Check the configured server and query it directly.
$ nslookup example.comanswer or “timed out”?5. Port, firewall, service
If the server is reachable but the application isn't, check the port. If it's closed, a firewall is blocking it or the service isn't running.
$ Test-NetConnection intranet.firma.local -Port 443Windows: TcpTestSucceeded$ sudo ss -tulpnLinux: is the service listening on the port?6. Fix, verify, document
Change one thing at a time, then re-test the original symptom and record cause and fix in the ticket. That's how a team spots recurring faults.
FAQ
- Why bottom-up (OSI model)?
- Higher layers depend on lower ones. Without a valid IP address DNS can't work; without a gateway no website on the internet loads.
- What if ping is blocked?
- Some servers and firewalls drop ICMP. Test with a port check (Test-NetConnection, curl) instead of ping.
Last updated: 2026-11-08