What it does
mtr (My Traceroute) continuously probes the network path
to a destination, combining the hop discovery of traceroute
with the ongoing latency and loss measurements of ping.
How it works (mechanical)
mtr sends probe packets with increasing TTL values to discover
each hop. Unlike traceroute, it keeps sending probes, maintaining rolling
statistics (loss, latency, jitter) for every hop in the path.
- TTL-based hop discovery
- Repeated probes per hop
- Rolling statistics over time
- ICMP, UDP, or TCP probes (depending on options)
10 Practical Examples
# 1) Interactive trace to a host mtr example.com
# 2) Numeric output only (no DNS) mtr -n example.com
# 3) Report mode (one-shot summary) mtr -r example.com
# 4) Report mode with 100 probes mtr -r -c 100 example.com
# 5) TCP-based probes (firewall-friendly) mtr -T example.com
# 6) Use a specific port for TCP probes mtr -T -P 443 example.com
# 7) Wide output (better alignment) mtr -w example.com
# 8) IPv6 path testing mtr -6 ipv6.google.com
# 9) JSON output (automation) mtr -r -j example.com
# 10) Diagnose intermittent packet loss mtr -n -c 200 example.com
Notes & Gotchas
- Loss shown at intermediate hops may be ICMP rate-limiting.
- End-to-end loss is the most important metric.
- TCP mode often succeeds where ICMP is blocked.
- Interactive mode is best for live diagnosis.
- Report mode is ideal for sharing with providers.
Historical Context
mtr was created to overcome the snapshot nature of traceroute,
giving operators a live, statistical view of path quality rather than
a single moment in time.
Modern Equivalent / Related Tools
- traceroute — single-pass hop discovery
- tracepath — MTU-focused path tracing
- ping — reachability and latency
- smokeping — long-term latency monitoring
- tcpdump — packet-level confirmation