CN2 GIA Route Test: How to Test Your Server's Route to Mainland China with MTR and Traceroute

A CN2 GIA route test tells you which path traffic really takes between your server and users in mainland China, and it is the first thing to run when pages feel slow or latency jumps in the evening. This guide shows how to run an MTR route test and traceroute on Debian/Ubuntu and CentOS/Rocky/AlmaLinux, how to test from the China side as well, how to read the hops (59.43.x.x for CN2, 202.97.x.x for the ordinary 163 backbone) and what information to give JUSTG support. It applies to JUSTG cloud servers and dedicated servers in Johannesburg and Moscow (China Telecom CN2 GIA), Tokyo (China Telecom premium routing) and Seoul (Asia-optimized routing).

Step 1: Install MTR and traceroute on the server

MTR combines ping and traceroute: it sends many probes to every hop and reports loss and latency per hop, which is far more useful than a single traceroute. Log in over SSH as root and install both tools.

# Debian / Ubuntu
apt update && apt install -y mtr-tiny traceroute

# CentOS / Rocky Linux / AlmaLinux
dnf install -y mtr traceroute
# (CentOS 7: yum install -y mtr traceroute)

Step 2: Run an MTR route test from the server to China

Pick a target in mainland China that represents your users, ideally the public IP of your own office or home connection, or an IP on the same carrier (China Telecom, China Unicom or China Mobile) as most of your visitors. Send at least 100 probes so short spikes are averaged out. The flags mean: -r report mode, -w wide output, -z show AS numbers, -b show hostnames and IPs, -c 100 probe count.

# replace 198.51.100.20 with a public IP inside mainland China
# (for example your office or home connection)
mtr -rwzbc 100 198.51.100.20

# TCP mode, useful when ICMP is filtered somewhere on the path
mtr -rwzbc 100 -T -P 443 198.51.100.20

If you only need a quick look at the path, a classic traceroute works too:

traceroute -n 198.51.100.20
# TCP traceroute to port 443
traceroute -T -p 443 -n 198.51.100.20

Step 3: Test the return route from a China-side machine

Internet routing is often asymmetric: the path from your server to China can differ from the return route from China back to your server, so test both directions. Run the same test from a computer inside mainland China towards your JUSTG server IP.

# Linux machine in mainland China -> your JUSTG server
mtr -rwzbc 100 203.0.113.10

# Windows PC in mainland China (Command Prompt)
tracert -d 203.0.113.10
pathping -n 203.0.113.10

If you do not have a machine in China, public web-based tools can help. ITDOG offers ping and route tracing from many Chinese provinces and carriers at once, and ping.pe runs ping and MTR from probes around the world. Use them to check whether a problem affects one carrier or one region only, and keep the result pages or copy the text output.

Step 4: Read the hops (59.43.x.x CN2 vs 202.97.x.x 163)

Below is an illustrative MTR report from a CN2 GIA server. Your own hop count and latency will differ.

HOST: justg-vps                    Loss%   Snt   Last   Avg  Best  Wrst StDev
  1. AS???    203.0.113.1           0.0%   100    0.5   0.6   0.3   2.1   0.2
  2. AS???    203.0.113.254         0.0%   100    1.1   1.3   0.9   3.0   0.3
  3. AS4809   59.43.x.x             0.0%   100   12.4  12.6  12.1  14.0   0.4
  4. AS4809   59.43.x.x            40.0%   100  148.2 148.9 147.6 153.1   1.0
  5. AS4809   59.43.x.x             0.0%   100  150.3 150.8 149.9 156.2   1.1
  6. AS4134   202.97.x.x            0.0%   100  152.0 152.6 151.4 158.9   1.3
  7. ???      198.51.100.20         0.0%   100  155.7 156.1 155.0 161.4   1.2

The backbone IP ranges tell you which network carries the traffic:

Pay attention to three things. First, the biggest latency jump usually appears on the international segment; that is distance, not a fault. Second, loss on an intermediate hop that does not continue to the following hops (like hop 4 above) is just a router limiting ICMP replies and can be ignored; only loss that persists to the final hop matters. Third, compare results at quiet hours with results between 20:00 and 23:00 Beijing time, when cross-border links are busiest.

Step 5: Check end-to-end latency and loss

MTR shows the path; a long ping gives a clean summary of the final result. Run it at the same time as the MTR tests.

ping -c 100 198.51.100.20 | tail -n 3

Stable latency with low jitter and near zero loss is a healthy result. Steady loss at the final hop, or latency that is much higher at peak times than at night, is worth reporting.

Step 6: What to send JUSTG support when latency is high

A complete report lets the network team find the problem quickly. Include:

• Your server IP and location (Johannesburg, Moscow, Tokyo or Seoul).
• Your public IP in China, your carrier and your province or city.
• MTR text output in both directions (server to China and China to server), copied as text rather than screenshots.
• The time of each test with time zone, ideally one at peak and one off-peak.
• The ping summary and a short description of the symptom (slow website, game lag, SSH delays).

FAQ

Why do I see 202.97.x.x hops on a CN2 GIA server?

Many home broadband users sit on the ChinaNet 163 access network, so a few 202.97.x.x hops near the destination are normal. What matters is whether the international segment is carried on 59.43.x.x.

MTR shows 30% loss on one hop. Is the line bad?

Not necessarily. If the following hops and the final destination show 0% loss, the router is simply rate-limiting ICMP. Only loss that continues to the last hop affects your traffic.

Which JUSTG location has the lowest latency to China?

Distance plays a large role, so Tokyo and Seoul are geographically closest, while Johannesburg and Moscow use China Telecom CN2 GIA. Run the tests in this guide from your users' carriers to compare.

If you still cannot resolve the issue, please submit a ticket to contact JUSTG technical support with the test results above.