The mtr command has two default output modes that aren't ideal for scripting:

• Graphical GTK window interface (default)
• Interactive ncurses terminal interface (-t flag)

Both modes create issues for developers who need to:

mtr example.com | grep "packet loss"
mtr -c 5 google.com > trace_results.txt

Use the --report or -r flag for script-friendly output:

mtr --report example.com
mtr -rw -c 10 google.com

The key flags to combine:

• -r/--report: Enables report mode
• -w/--report-wide: Shows full hostnames
• -c COUNT: Sets number of pings per hop

This command provides clean, parseable output:

mtr -rw -c 5 8.8.8.8 | tee dns_trace.log

For JSON output (requires mtr 0.90+):

mtr --json example.com | jq '.report.hubs[] | select(.LossPct > 0)'

CSV output alternative:

mtr -rwc 1 example.com | awk '!/Start|HOST/ {print $1","$2","$3"%"}' > trace.csv

If you see terminal control characters:

mtr -r example.com | col -b > clean_output.txt

For older mtr versions (pre-0.82):

mtr --no-dns -r example.com

When running mtr (My Traceroute) on Linux systems, the default behavior often frustrates sysadmins and developers who want clean, pipeable output. The two main interface options - GTK and NCURSES - both interfere with standard UNIX output redirection.

# What happens when you try basic redirection
$ mtr google.com > output.txt
$ cat output.txt  # Contains terminal control chars like ESC[?1049h

MTR actually has built-in options for raw output, but they're not well documented. Here's the combination that works:

# Report mode with CSV output (best for parsing)
$ mtr --report --csv google.com

# Alternative: Standard report format
$ mtr --report google.com

# Force single cycle and show IPs (good for quick checks)
$ mtr --report -c1 -b google.com

To avoid typing flags repeatedly, configure defaults in /etc/mtr.conf:

# Sample configuration for script-friendly output
report-mode=1
displaymode=1  # 0=GTK, 1=NCURSES, 2=RAW
interval=1.0
timeout=10

Here's how to process MTR output in shell scripts:

# Extract latency stats per hop
mtr --report -c5 google.com | awk '/^h/ {print $2,$3,$4}'

# Continuous monitoring with alerting
while true; do
  mtr --report -c1 api.service.com | \
  grep -E '^[0-9]+\.[0-9]+\.[0-9]+\.[0-9]+' | \
  awk '$7 > 100 {system("send-alert.sh High latency at " $1)}'
  sleep 300
done

• Missing hostnames: Use -z flag for DNS resolution
• IPv6 problems: Force IPv4 with -4 or IPv6 with -6
• Timeout adjustments: -u for UDP, -T for TCP, -P for specific ports