How to check traceroute / MTR

Quick answer: traceroute shows the route packets take hop by hop; MTR combines traceroute with continuous ping statistics and is the tool support actually wants. Run mtr -rwbc 100 TARGET for a shareable report, and remember: only loss that persists to the final hop is real.

Overview

When "the network is slow/broken", the first question is where. Traceroute/MTR answer it by mapping every router between two points with its latency and loss. Internet paths are asymmetric, so a complete picture always needs a run from each side.

Before you start

• Install:

  sudo apt install -y mtr-tiny traceroute   # Debian/Ubuntu
  sudo dnf install -y mtr traceroute        # AlmaLinux/Rocky

  Windows: tracert is built in; WinMTR is the MTR equivalent.
• Know both endpoints: server IP and your public IP (curl -4 ifconfig.me).

Step-by-step guide

1. Quick route view:

   traceroute -n TARGET_IP     # -n skips DNS for speed

2. Proper measurement with MTR:

   mtr -rwbc 100 TARGET_IP

   -r report mode, -w wide (full hostnames), -b show IPs and names, -c 100 = 100 cycles (~2 min).
3. Read the columns: Loss% (see the rule below), Avg (typical RTT), StDev (jitter — high values mean unstable latency).
4. Find the story: a big Avg jump between two hops = the long geographic leg; loss starting at hop N and continuing to the end = problem at/after hop N.
5. Run the reverse direction from the server towards your IP — attach both to any support ticket.

Common issues

• Middle-hop loss panic: routers rate-limit ICMP about themselves; if the final hop is clean, the path is clean.
• Hops showing ???: some routers don't reply to traceroute at all — normal, judge by what surrounds them.
• Too few cycles: 10 pings prove nothing; use -c 100 for statistics worth sharing.

When to contact support

Attach MTR reports (both directions, 100 cycles, timestamps) whenever you report a network issue — it's the single most useful artifact for our network team.

Frequently asked questions

Traceroute or MTR — which should I use?

MTR for anything you plan to share or analyze: it adds continuous statistics per hop. Traceroute is fine for a quick one-off route view.

What does ??? mean in the output?

It is a router that does not answer traceroute probes at all — completely normal on the Internet. Judge the path by the responding hops around it instead.

Which MTR flags does support expect?

mtr -rwbc 100 TARGET — report mode, wide output, names plus IPs, 100 cycles, run from both directions.

Related articles

• How to troubleshoot packet loss
• How to check latency to Cloud2Y locations
• How to report network issues to support
• How to use a Looking Glass

Need a hand? Contact Cloud2Y support →