MA-NET1 · Module 14
Packet reading — ICMP echo and path probes
Topics: ICMP echo · TTL expiry · Traceroute
Overview
Session 14 — ICMP echo and path probes
Most traffic you have studied so far carries someone's data: a forecast, a DNS answer, an icon. Networks also need a small stream of messages about themselves: "is that machine alive?", "your packet died because its hop budget ran out", "nobody is listening on that port". These travel in ICMP, a companion of IP that rides directly inside IP datagrams, with no TCP or UDP in between.
In this session you will read two invented capture panes. In the first, Mira pings a Northline Relay server and you decode echo request and echo reply frames, learn the type and code idea, and read round-trip times. In the second, a traceroute-style probe sequence crosses BayLink Access and SpineNet on the way to the relay, and you will see how TTL expiry and time-exceeded messages let the sender discover each router in turn.
Every address (in 198.51.100.x and 203.0.113.x) and every millisecond figure is invented for practice, and nothing needs to be installed.
Plan: Concepts → Examples → Quiz.
Time: about 55–70 minutes, including a few hop-by-hop timing comparisons.
Concepts
Core ideas — ICMP messages, ping, and how path probes discover routers
Everything in the last three sessions was about carrying application data. This session is about the network talking about itself. When a router has to discard a datagram, when an operator wants to know whether a machine is reachable, or when a host receives a UDP datagram for a port nobody is listening on, something has to be able to say so. That something is the Internet Control Message Protocol, ICMP. It is small, it is everywhere, and two famous tools, ping and traceroute, are built entirely out of it. We will read both from invented capture panes, following Mira, whose Skylark Forecast app has started to feel slow when she is on a BayLink Access connection.
ICMP rides inside IP
ICMP is a little unusual. It is not a transport protocol like TCP or UDP, and it has no ports. An ICMP message is placed directly inside an IP datagram, and the IP header's Protocol field says so: the value is 1 (compare 6 for TCP and 17 for UDP). Here is how the layers stack for an ICMP frame:
Link header (14) | IP header (20), Protocol = 1 | ICMP message (8 + data)
Because there are no ports, ICMP messages are not addressed to a program; they are handled by the operating system's network code itself. That is why you can ping a machine that runs no servers at all.
Every ICMP message begins with the same three small fields:
| Field | Size | Purpose | |---|---|---| | Type | 1 byte | The broad category of message | | Code | 1 byte | A finer detail within that category | | Checksum | 2 bytes | Error check over the ICMP message |
After those come four more bytes whose meaning depends on the type (for echo messages, an identifier and a sequence number), followed by any data.
Common confusion: It is tempting to say ICMP is "above" IP the way TCP is, because it sits in the IP payload. But architecturally it is part of the network layer, a helper for IP itself. The practical consequence for reading captures: you will see
Protocol: ICMP (1)in the IP header, and no port numbers anywhere.
Type and code: a category and a detail
You do not need to memorise the full list. You need an intuition: the type says what kind of message this is, and the code refines it. The few you will meet constantly are these.
| Type | Code | Name | What it means | |---|---|---|---| | 8 | 0 | Echo request | "Are you there?" (what ping sends) | | 0 | 0 | Echo reply | "Yes, I am here." (the answer to ping) | | 3 | 0 | Destination unreachable, network | A router has no route toward that network | | 3 | 1 | Destination unreachable, host | The last router could not reach that host | | 3 | 3 | Destination unreachable, port | The host is there, but nothing listens on that UDP port | | 11 | 0 | Time exceeded, TTL expired in transit | A router discarded a datagram whose TTL reached zero |
Reading one is a two-step habit: first the type to learn the category (reply? error? expired?), then the code to learn the reason. Type 3 is the "something could not be delivered" family, and its code tells you whether the problem was the network, the host, or the port. Type 11 is the "ran out of hops" message.
Notice the pattern: types 3 and 11 are errors reported back to the original sender. Types 8 and 0 are a deliberate question and its answer.
Ping: reading an echo exchange
Mira runs a ping to the relay server 203.0.113.40 from a BayLink-connected computer at 198.51.100.77. This time she captures at that computer. Here is the packet list:
No. Time(s) Source Destination Proto Len Info
1 0.000000 198.51.100.77 203.0.113.40 ICMP 74 Echo request id=0x2b6e seq=1 ttl=64
2 0.015290 203.0.113.40 198.51.100.77 ICMP 74 Echo reply id=0x2b6e seq=1 ttl=58
3 1.000000 198.51.100.77 203.0.113.40 ICMP 74 Echo request id=0x2b6e seq=2 ttl=64
4 1.015740 203.0.113.40 198.51.100.77 ICMP 74 Echo reply id=0x2b6e seq=2 ttl=58
5 2.000000 198.51.100.77 203.0.113.40 ICMP 74 Echo request id=0x2b6e seq=3 ttl=64
6 3.000000 198.51.100.77 203.0.113.40 ICMP 74 Echo request id=0x2b6e seq=4 ttl=64
7 3.015230 203.0.113.40 198.51.100.77 ICMP 74 Echo reply id=0x2b6e seq=4 ttl=58
And the detail pane for frame 1:
Internet Protocol Src 198.51.100.77 Dst 203.0.113.40 TTL 64 Protocol ICMP (1) Total length 60
Internet Control Message Protocol
Type: 8 (Echo request) Code: 0 Checksum: 0x4f6a [correct]
Identifier: 0x2b6e Sequence: 1
Data (32 bytes): "skylark-probe-0123456789-abcdefg"
Walk through it. The identifier (0x2b6e) is picked once per ping run so the sender can tell its own replies from anyone else's; the sequence number counts up by one per request, and the reply copies both back. That is how the sender pairs each reply with its request, and how it notices a missing one: frame 5 (seq=3) has no matching reply anywhere in the capture, so that probe, or its reply, was lost.
Round-trip time (RTT) is the gap between a request and its reply: frame 2 minus frame 1 is 0.015290 s = 15.29 ms; frame 4 minus frame 3 is 15.74 ms; frame 7 minus frame 6 is 15.23 ms. The summary a ping tool prints would read: 4 sent, 3 received, 25 % loss, RTT min/avg/max = 15.23 / 15.42 / 15.74 ms.
The size chain still works: 14 (link) + 20 (IP) + 8 (ICMP header) + 32 (data) = 74, matching Len.
One more clue hides in the TTL column. The request leaves with TTL 64, while the reply arrives with TTL 58. The server probably started its reply at 64 as well, and every router on the way back subtracted 1, so six routers sat on the return path. You cannot be sure of the starting value (some systems use 128 or 255), but the guess is a handy sanity check on hop counts.
Common confusion: A missing echo reply does not prove the machine is down. The request may have been lost, the reply may have been lost, a firewall may be configured to drop echo requests, or the target may rate-limit them. Silence from ping is a clue, not a verdict. Conversely, a reply proves the path works for this kind of packet, not that the web server on the machine is healthy.
TTL expiry: the idea that makes path probes possible
From the network-layer session you know that each router subtracts 1 from the IP TTL and discards a datagram whose TTL reaches 0. TTL is a hop budget, designed to stop datagrams circulating forever in a routing loop. What we did not discuss then is that the router which discards a datagram this way does not do so silently. It sends an ICMP Time Exceeded message (type 11, code 0) back to the datagram's source.
That message has a very useful shape. Its IP source address is the router's own address. Its payload quotes the beginning of the discarded datagram: the original IP header plus the first 8 bytes of what followed. For an echo request, those 8 bytes contain the ICMP type, code, checksum, identifier and sequence number, which is exactly what the sender needs to work out which of its probes died. Here is such a message, frame 2 of the next capture:
Internet Protocol Src 198.51.100.130 Dst 198.51.100.77 TTL 253 Protocol ICMP (1) Total length 56
Internet Control Message Protocol
Type: 11 (Time exceeded) Code: 0 (TTL expired in transit)
Quoted original datagram:
IP: Src 198.51.100.77 Dst 203.0.113.40 TTL 1 Protocol ICMP (1) Total length 60
ICMP: Type 8 Code 0 Identifier 0x31a0 Sequence 7 (first 8 bytes only)
Read the quoted TTL: 1. That is the TTL the discarded datagram had on arrival at that router; the router subtracted one, got 0, and gave up. It also tells you the probe started with a bigger TTL (3) and was reduced by two earlier routers before it reached this one.
Traceroute: using the expiry on purpose
Here is the trick. If you send a datagram toward a destination with TTL = 1, the very first router will expire it and reveal itself through a Time Exceeded message. With TTL = 2, the first router passes it on and the second router reveals itself. With TTL = 3 the third router speaks, and so on. By sending probes with TTL 1, 2, 3, ... and listening for the replies, the sender collects the address of each router in order. When a probe finally has a TTL large enough to reach the destination, the destination answers differently (here, with an echo reply), and the search is over.
Real tools send three probes per TTL value so that one lost probe does not leave a hole, and they print the three RTTs. Here is the middle of Mira's traceroute capture. The tool uses ICMP echo requests as probes, with identifier 0x31a0, and the third group of probes (TTL 3) is shown:
No. Time(s) Source Destination Proto Len Info
1 0.000000 198.51.100.77 203.0.113.40 ICMP 74 Echo request id=0x31a0 seq=7 ttl=3
2 0.009810 198.51.100.130 198.51.100.77 ICMP 70 Time-to-live exceeded (in transit) [quotes seq=7]
3 0.010050 198.51.100.77 203.0.113.40 ICMP 74 Echo request id=0x31a0 seq=8 ttl=3
4 0.019750 198.51.100.130 198.51.100.77 ICMP 70 Time-to-live exceeded (in transit) [quotes seq=8]
5 0.019900 198.51.100.77 203.0.113.40 ICMP 74 Echo request id=0x31a0 seq=9 ttl=3
6 0.030000 198.51.100.130 198.51.100.77 ICMP 70 Time-to-live exceeded (in transit) [quotes seq=9]
All three probes were addressed to the final destination, 203.0.113.40, yet every answer came from 198.51.100.130, the third router. That is the whole mechanism in one picture: the destination address of the probe stayed the same while the TTL decided who answered. The RTTs here are 9.81, 9.70 and 10.10 ms.
Putting the whole run together, the tool prints a table like this:
1 198.51.100.1 1.4 ms 1.2 ms 1.3 ms BayLink: customer gateway
2 198.51.100.9 3.1 ms 2.9 ms 3.0 ms BayLink: core
3 198.51.100.130 9.8 ms 9.7 ms 10.1 ms SpineNet: entry
4 198.51.100.134 11.2 ms 11.0 ms 11.4 ms SpineNet
5 198.51.100.138 12.1 ms * 12.0 ms SpineNet: exit
6 203.0.113.1 14.6 ms 14.5 ms 14.9 ms Northline Relay: edge
7 203.0.113.40 15.3 ms 15.2 ms 15.6 ms the destination itself
(The labels on the right are our annotations; a real tool would print only the addresses and times.) The last row is the destination, which answered with echo replies, so the path has six routers and the destination is the seventh hop, consistent with the TTL 58 guess from ping. The final RTT, about 15.3 ms, agrees with the ping RTT of about 15.3 ms.
Reading the quirks
Traceroute output often looks slightly strange, and each strangeness has a calm explanation.
- A
*instead of a time means a probe got no reply within the timeout. Some routers rate-limit or deprioritise the generation of ICMP messages, and some filters block them. One star among three probes, like hop 5 above, is common and harmless. A whole row of stars is more interesting, but traffic may still pass through that router normally; it just does not announce itself. - RTTs do not have to grow smoothly. Each hop's RTT includes the trip to that router, the time the router takes to build the ICMP message, and the trip back, which may even follow a different path. A router can answer a probe slowly while forwarding real traffic quickly, so one hop's RTT can be larger than the next's.
- Jumps hint at long links. The RTT goes from 3.0 to 9.8 ms between hops 2 and 3: about 6.7 ms of extra round trip, which suggests the BayLink-to-SpineNet link is physically longer than the others. Differences between hops are more informative than any single number.
- The address shown is an interface address. A router has several interfaces, and the one that appears is normally the one that received the probe.
Common confusion: Traceroute does not ask each router "what is your address?" and does not read a route table anywhere. It infers the path by watching which routers happen to complain about expiring probes. If the network sends different probes along different routes (some balanced paths do), the table can even mix routers from two different paths.
What to carry forward
ICMP lives directly inside IP (Protocol 1), has no ports, and begins with type and code: echo request is 8/0, echo reply is 0/0, the 3-family says "could not deliver" with the code naming the reason, and 11/0 says "TTL expired in transit". Ping measures reachability and RTT from matching identifier and sequence numbers. Traceroute deliberately sends probes with growing TTL, so that each router along the path reveals itself by sending a Time Exceeded message from its own address. The same reading habits of the previous sessions apply: unwrap the layers, check the length chain, and place each fact where it actually lives.
Examples
Session 14 — Worked examples
Every example has a Scene, a Capture, a Reading, and a Takeaway. The captures are invented, and the numbers are chosen so that each step can be checked by hand. Try covering the Reading first.
Example A — Reading a ping run
Scene. Mira's app feels slow on the BayLink connection, so she pings the Northline Relay server 203.0.113.40 from her BayLink-connected computer (198.51.100.77) and captures the result. She wants three answers: how fast is the round trip, was anything lost, and roughly how many routers are in between?
Capture.
No. Time(s) Source Destination Proto Len Info
1 0.000000 198.51.100.77 203.0.113.40 ICMP 74 Echo request id=0x2b6e seq=1 ttl=64
2 0.015290 203.0.113.40 198.51.100.77 ICMP 74 Echo reply id=0x2b6e seq=1 ttl=58
3 1.000000 198.51.100.77 203.0.113.40 ICMP 74 Echo request id=0x2b6e seq=2 ttl=64
4 1.015740 203.0.113.40 198.51.100.77 ICMP 74 Echo reply id=0x2b6e seq=2 ttl=58
5 2.000000 198.51.100.77 203.0.113.40 ICMP 74 Echo request id=0x2b6e seq=3 ttl=64
6 3.000000 198.51.100.77 203.0.113.40 ICMP 74 Echo request id=0x2b6e seq=4 ttl=64
7 3.015230 203.0.113.40 198.51.100.77 ICMP 74 Echo reply id=0x2b6e seq=4 ttl=58
Reading.
- Pair the frames. Each reply copies the identifier and sequence number of its request, so seq 1 pairs frames 1 and 2, seq 2 pairs frames 3 and 4, and seq 4 pairs frames 6 and 7.
- Round-trip times. Seq 1: 0.015290 − 0.000000 = 15.29 ms. Seq 2: 1.015740 − 1.000000 = 15.74 ms. Seq 4: 3.015230 − 3.000000 = 15.23 ms.
- Loss. Seq 3 (frame 5) has no reply. Four requests, three replies: one lost, which is 1 ÷ 4 = 25 % loss for this short run.
- Average RTT. (15.29 + 15.74 + 15.23) ÷ 3 = 46.26 ÷ 3 = 15.42 ms, with minimum 15.23 ms and maximum 15.74 ms.
- Routers on the path. The replies arrive with TTL 58. If the server started them at 64 (a common default), then 64 − 58 = 6 routers decremented the TTL on the way back. That is only a hint, because the start value is an assumption.
- Size check. 14 + 20 + 8 + 32 = 74 matches the
Lencolumn: a 32-byte data area plus an 8-byte ICMP header.
Takeaway. Ping gives you pairing (identifier and sequence), round-trip time (reply time minus request time), loss (unanswered sequence numbers), and a rough hop-count hint (the reply's TTL).
Example B — Sorting ICMP messages by type and code
Scene. Skylark's support engineer sends Mira four ICMP lines from different incidents and asks her to say, in plain words, what each one means.
Capture.
(a) ICMP 198.51.100.77 → 203.0.113.40 Type 8 Code 0
(b) ICMP 203.0.113.40 → 198.51.100.77 Type 0 Code 0
(c) ICMP 198.51.100.134 → 198.51.100.77 Type 11 Code 0
(d) ICMP 203.0.113.1 → 198.51.100.77 Type 3 Code 1
Reading.
- (a) Type 8, code 0 is an echo request: the sender is asking "are you there?". It travels from Mira's machine to the server.
- (b) Type 0, code 0 is an echo reply, the answer to (a). It travels back from the server, with source and destination swapped.
- (c) Type 11, code 0 is time exceeded, TTL expired in transit. It comes from
198.51.100.134, a router that discarded one of Mira's datagrams because its TTL reached zero. The source address tells you which router, and it is not the destination. - (d) Type 3 is the "destination unreachable" family, and code 1 says host unreachable. It comes from
203.0.113.1, the Northline edge router, which is telling Mira's machine that it could not deliver a datagram to the final host on its own network. This is a different story from (c): the datagram did not run out of hops; it arrived at the last router and found no way to the machine. - Direction clue. (a) goes from Mira; (b), (c) and (d) all come to her, because replies and errors are always sent back to the source of the datagram that caused them.
Takeaway. Read type first to get the category (question, answer, error), then code for the reason. The source address of an error tells you which device noticed the problem.
Example C — Anatomy of one Time Exceeded message
Scene. During a traceroute, one frame arrives from 198.51.100.130. Mira wants to know which of her probes died, why it died there, and how far the message itself travelled.
Capture.
Frame 2: 70 bytes
Internet Protocol Src 198.51.100.130 Dst 198.51.100.77 TTL 253 Protocol ICMP (1) Total length 56
Internet Control Message Protocol
Type: 11 (Time exceeded) Code: 0 (TTL expired in transit)
Quoted original datagram:
IP: Src 198.51.100.77 Dst 203.0.113.40 TTL 1 Protocol ICMP (1) Total length 60
ICMP: Type 8 Code 0 Identifier 0x31a0 Sequence 7
Reading.
- Who spoke? The IP source is
198.51.100.130. In a time-exceeded message, the source is the router that discarded the datagram. So the third router on the path is198.51.100.130. - Which probe? Look inside the quote: identifier
0x31a0, sequence 7. That is the probe sent as seq 7 (the first with TTL 3). The sender can therefore match this error to the exact request, even though the error came from a stranger and not from the destination. - Why did it die? The quoted TTL is 1. That is the TTL the probe had when this router received it. The router subtracted 1, reached 0, and discarded it. The probe left Mira's computer with TTL 3, lost one at router 1 and one at router 2, and arrived here with 1.
- Where is the destination? The quote shows the original destination
203.0.113.40, not the router. The probe was never addressed to the router at all; the router merely happened to be where its budget ended. - How far did the error itself travel? The Time Exceeded message arrives with TTL 253. If the router started it at 255, it lost 2 on the way back, through router 2 and router 1. That matches the router being third on the path.
- Sizes. The message is 20 (IP header) + 8 (ICMP header) + 28 (quoted 20-byte IP header and first 8 bytes) = 56 bytes of IP datagram, and with the 14-byte link header 70 bytes on the wire. The router quotes only the first 8 bytes of the probe, which is enough to include the identifier and sequence number.
Takeaway. A time-exceeded frame answers three questions at once. The source says who, the quote says which probe, and the quoted TTL says why.
Example D — Reading a complete traceroute table
Scene. Mira runs the traceroute to 203.0.113.40 and wants to describe the path to BayLink's support desk: which network owns each hop, where the long link is, and what the star means.
Capture.
1 198.51.100.1 1.4 ms 1.2 ms 1.3 ms
2 198.51.100.9 3.1 ms 2.9 ms 3.0 ms
3 198.51.100.130 9.8 ms 9.7 ms 10.1 ms
4 198.51.100.134 11.2 ms 11.0 ms 11.4 ms
5 198.51.100.138 12.1 ms * 12.0 ms
6 203.0.113.1 14.6 ms 14.5 ms 14.9 ms
7 203.0.113.40 15.3 ms 15.2 ms 15.6 ms
Reading.
- One row per TTL value, three probes per row. Row n shows the replies for TTL = n. Seven rows mean the destination was reached with TTL 7, so there are 7 − 1 = 6 routers between Mira and the server.
- Who owns what? BayLink's own addresses are in rows 1 and 2 (the customer gateway and the core). SpineNet's backbone appears in rows 3 to 5, and Northline Relay's edge router is row 6. Row 7 is the destination, which answered with echo replies and not time-exceeded messages. (Ownership comes from knowing the networks; the tool only prints addresses.)
- Hop-by-hop differences using the first column: 1.4 → 3.1 (+1.7), 3.1 → 9.8 (+6.7), 9.8 → 11.2 (+1.4), 11.2 → 12.1 (+0.9), 12.1 → 14.6 (+2.5), 14.6 → 15.3 (+0.7). The largest jump is between rows 2 and 3, the BayLink-to-SpineNet crossing, so that is the link to ask about if the delay matters.
- The star. In row 5 the second probe got no answer. The other two did, and the following rows reply normally, so traffic clearly passes through that router. It likely rate-limited its ICMP replies. A single star is no cause for concern.
- Consistency check. The destination's RTT of about 15.3 ms agrees with the ping RTT of 15.29 ms in Example A, and the six routers agree with the TTL-58 hint. Two independent tools telling the same story is reassuring.
Takeaway. Count rows for the hop number, read addresses for ownership, and read differences (not raw values) for where time is spent. A star among replies is usually silence, not breakage.
Example E — When the probes are UDP: port unreachable at the end
Scene. Some traceroute tools send UDP datagrams to an unused high port, instead of echo requests. A teammate runs such a tool against the same server and asks how it knows when to stop, since there is no echo reply to wait for.
Capture.
No. Time(s) Source Destination Proto Len Info
40 0.000000 198.51.100.77 203.0.113.40 UDP 70 50412 → 40123 Len=28 ttl=7
41 0.015500 203.0.113.40 198.51.100.77 ICMP 70 Destination unreachable (Port unreachable)
Detail of frame 41:
IP: Src 203.0.113.40 Dst 198.51.100.77 Protocol ICMP (1)
ICMP: Type 3 Code 3 (Port unreachable)
Quoted: IP Src 198.51.100.77 Dst 203.0.113.40 ... UDP Src port 50412 Dst port 40123
Reading.
- The probe. Frame 40 is a UDP datagram from an ephemeral port (50412) to port 40123, a number chosen because it is very unlikely that any service listens there, with TTL 7.
- The reply. With TTL 7 the probe survives all six routers and reaches the destination host. The host receives a UDP datagram for port 40123, finds nobody listening, and sends back an ICMP message of type 3, code 3: port unreachable.
- Why this works as a stopping signal. Earlier hops answered with type 11 (time exceeded). This one answers with type 3 from the destination's own address (
203.0.113.40). A type 3 message coming from the destination itself means "the probe arrived", so the tool stops. - Matching the probe. The quoted header includes the UDP ports, 50412 and 40123, so the sender can tell exactly which probe this refers to, just as the echo identifier and sequence did earlier.
- Size check. 14 + 20 + 8 + 28 = 70 for the UDP probe: a 28-byte payload plus the 8-byte UDP header gives UDP Length 36, and IP total length 56.
- Contrast with Example B (d). Both are type 3. Code 1 (host unreachable) came from a router that could not reach the machine; code 3 (port unreachable) came from the machine itself, which is a good sign here.
Takeaway. Whatever probe type is used, the logic is the same: routers along the way answer "time exceeded", and the destination answers differently ("echo reply" or "port unreachable"), which tells the tool the path has ended.
Practice check — ICMP echo and path probes
Multiple-choice check from the Matheion app lesson. Pick an answer for each question, then submit.
1.A frame's IP header shows 'Protocol ICMP (1)'. Which statement about the frame is correct?
2.Four ICMP messages are captured: (a) Type 8 Code 0 (b) Type 0 Code 0 (c) Type 3 Code 3 (d) Type 11 Code 0 Which one says 'a router discarded a datagram because its TTL reached zero'?
3.A ping capture contains: 1 0.000000 Echo request id=0x2b6e seq=1 2 0.015290 Echo reply id=0x2b6e seq=1 3 1.000000 Echo request id=0x2b6e seq=2 4 2.000000 Echo request id=0x2b6e seq=3 5 2.014800 Echo reply id=0x2b6e seq=3 What are the round-trip time of seq 1 and the status of seq 2?
4.Why does a time-exceeded message quote the first bytes of the datagram it is about, including the ICMP identifier and sequence number of the original probe?
5.A traceroute-style capture shows probes with these TTL values, all addressed to 203.0.113.40: seq 7 ttl=3 → answered by 198.51.100.130 (Type 11 Code 0) seq 10 ttl=4 → answered by 198.51.100.134 (Type 11 Code 0) What explains the different answering routers?
6.A traceroute table ends with these rows: 6 203.0.113.1 14.6 ms 14.5 ms 14.9 ms 7 203.0.113.40 15.3 ms 15.2 ms 15.6 ms The tool stopped at row 7 because the destination answered with an echo reply. How many routers lie between the sender and the destination?
7.One row of a traceroute table reads: 5 198.51.100.138 12.1 ms * 12.0 ms and the next rows answer normally. What is the most reasonable reading?
8.A UDP-probe traceroute ends with this frame from the destination itself: ICMP 203.0.113.40 → 198.51.100.77 Type 3 Code 3 (Port unreachable) What does this tell the tool?