1tcptop(8) System Manager's Manual tcptop(8)
2
3
4
6 tcptop - Summarize TCP send/recv throughput by host. Top for TCP.
7
9 tcptop [-h] [-C] [-S] [-p PID] [interval] [count]
10
12 This is top for TCP sessions.
13
14 This summarizes TCP send/receive Kbytes by host, and prints a summary
15 that refreshes, along other system-wide metrics.
16
17 This uses dynamic tracing of kernel TCP send/receive functions, and
18 will need to be updated to match kernel changes.
19
20 The traced TCP functions are usually called at a lower rate than per-
21 packet functions, and therefore have lower overhead. The traced data is
22 summarized in-kernel using a BPF map to further reduce overhead. At
23 very high TCP event rates, the overhead may still be measurable. See
24 the OVERHEAD section for more details.
25
26 Since this uses BPF, only the root user can use this tool.
27
29 CONFIG_BPF and bcc.
30
32 -h Print USAGE message.
33
34 -C Don't clear the screen.
35
36 -S Don't print the system summary line (load averages).
37
38 -p PID Trace this PID only.
39
40 interval
41 Interval between updates, seconds (default 1).
42
43 count Number of interval summaries (default is many).
44
46 Summarize TCP throughput by active sessions, 1 second refresh:
47 # tcptop
48
49 Don't clear the screen (rolling output), and 5 second summaries:
50 # tcptop -C 5
51
52 Trace PID 181 only, and don't clear the screen:
53 # tcptop -Cp 181
54
56 loadavg:
57 The contents of /proc/loadavg
58
59 PID Process ID.
60
61 COMM Process name.
62
63 LADDR Local address (IPv4), and TCP port
64
65 RADDR Remote address (IPv4), and TCP port
66
67 LADDR6 Source address (IPv6), and TCP port
68
69 RADDR6 Destination address (IPv6), and TCP port
70
71 RX_KB Received Kbytes
72
73 TX_KB Transmitted Kbytes
74
76 This traces all send/receives in TCP, high in the TCP/IP stack (close
77 to the application) which are usually called at a lower rate than per-
78 packet functions, lowering overhead. It also summarizes data in-kernel
79 to further reduce overhead. These techniques help, but there may still
80 be measurable overhead at high send/receive rates, eg, ~13% of one CPU
81 at 100k events/sec. use funccount to count the kprobes in the tool to
82 find out this rate, as the overhead is relative to the rate. Some sam‐
83 ple production servers tested found total TCP event rates of 4k to 15k
84 per second, and the CPU overhead at these rates ranged from 0.5% to
85 2.0% of one CPU. If your send/receive rate is low (eg, <1000/sec) then
86 the overhead is expected to be negligible; Test in a lab environment
87 first.
88
90 This is from bcc.
91
92 https://github.com/iovisor/bcc
93
94 Also look in the bcc distribution for a companion _examples.txt file
95 containing example usage, output, and commentary for this tool.
96
98 Linux
99
101 Unstable - in development.
102
104 Brendan Gregg
105
107 top(1) by William LeFebvre
108
110 tcpconnect(8), tcpaccept(8)
111
112
113
114USER COMMANDS 2016-09-13 tcptop(8)