How to use tcpdump to capture network packets under CentOS

The following column of centos tutorial will introduce to you how to use tcpdump network packet capture under CentOS. I hope it will be helpful to friends who need it. Helps!

tcpdump is a tool for intercepting and analyzing network data packets under Linux. It has great reference value for optimizing system performance.

Installation

tcpdump is not installed by default. Install under CentOS:

yum install tcpdump

Install under Ubuntu:

apt-get install tcpdump

Start by default

tcpdump

Under normal circumstances, starting tcpdump directly will monitor all data packets flowing on the first network interface.

Monitor the data packets of the specified network interface (be sure to check the network card)

tcpdump -i eth1

If you do not specify a network card, by default tcpdump will only monitor the first network interface, usually eth0. The following examples are No network interface specified.

Monitor the data packets of the specified host

Print all the data packets entering or leaving sundown.

tcpdump host sundown

You can also specify the ip, for example, intercept all 210.27.48.1 hosts receiving and all the data packets sent out

tcpdump host 210.27.48.1

Print the data packets communicated between helios and hot or with ace

tcpdump host helios and \( hot or ace \)

Intercepted host 210.27.48.1 and host 210.27.48.2 or 210.27.48.3 Communication

tcpdump host 210.27.48.1 and \ (210.27.48.2 or 210.27.48.3 \)

Print the IP packets communicated between ace and any other host, but not including the packets with helios.

tcpdump ip host ace and not helios

If you want to get the host 210.27.48.1 except and For IP packets communicated by all hosts other than host 210.27.48.2, use the command:

tcpdump ip host 210.27.48.1 and ! 210.27.48.2

Intercept all data sent by host hostname

tcpdump -i eth0 src host hostname

Monitor all data packets sent to host hostname

tcpdump -i eth0 dst host hostname

Monitor the data packets of the specified host and port

If you want to obtain the telnet packets received or sent by the host 210.27.48.1, use the following command

tcpdump tcp port 23 and host 210.27.48.1

to monitor the udp 123 port of the local machine 123 is the service port of ntp

tcpdump udp port 123

Monitor the packets of the specified network

Print all communication packets between the local host and the host on the Berkeley network (nt: ucb-ether, here It can be understood as the network address of the 'Berkeley network'. The most original meaning of this expression can be expressed as: Print all packets with the network address ucb-ether)

tcpdump net ucb-ether

Print all ftp packets that pass through the gateway snup (Note that the expression is enclosed in single quotes, which prevents the shell from misparsing the parentheses)

tcpdump 'gateway snup and (port ftp or ftp-data)'

Print all IP packets whose source or destination address is the local host

(If the local network is connected to another network through a gateway, the other network cannot be counted as the local network. (nt: The translation of this sentence is confusing and needs to be supplemented). When localnet is actually used, it must be replaced with the name of the local network)

tcpdump ip and not net localnet

Monitor the packets of the specified protocol

Print the start and end packets in the TCP session, and the source or destination of the packet is not a host on the local network. (nt: localnet, actual use It must be replaced with the name of the local network))

tcpdump 'tcp[tcpflags] & (tcp-syn|tcp-fin) != 0 and not src and dst net localnet'

Print all source or destination ports are 80, the network layer protocol is IPv4, and contain data, instead of SYN, FIN and ACK-only, which do not contain data. packet. (The expression of the ipv6 version can be used as an exercise)

tcpdump 'tcp port 80 and (((ip[2:2] - ((ip[0]&0xf)<<2)) - ((tcp[12]&0xf0)>>2)) != 0)'

(nt: can be understood as, ip[2:2] represents the length of the entire ip packet, (ip[0]&0xf) <<2) Indicates the length of the ip data packet header (ip[0]&0xf represents the IHL field in the packet, and the unit of this field is 32 bits. To convert

into bytes, you need to multiply by 4 , that is, shift left by 2. (tcp[12]&0xf0)>>4 represents the length of the tcp header. The unit of this field is also 32 bits. The conversion to the number of bits is ((tcp[12]&0xf0) >> 4) <<２,
That is ((tcp[12]&0xf0)>>2). ((ip[2:2] - ((ip[0]&0xf)<<2)) - ((tcp[12]&0xf0)>>2)) != 0 means: the length of the entire ip packet minus the length of the ip header, and then minus the length of the
tcp header is not 0, which means Yes, there is indeed data in the ip packet. For the ipv6 version, you only need to consider the difference between the 'Payload Length' in the ipv6 header and the 'tcp header length', and the expression 'ip[]' needs to be replaced with 'ip6 []'.)

Print IP packets whose length exceeds 576 bytes, and whose gateway address is snup

tcpdump 'gateway snup and ip[2:2] > 576'

Print all IP layer broadcast or multicast packets, but not physical Ethernet Network layer broadcast or multicast datagram

tcpdump 'ether[0] & 1 = 0 and ip[16] >= 224'

Print ICMP data packets other than 'echo request' or 'echo reply' types (for example, you can use this when you need to print all data packets generated by non-ping programs) Expression.
(nt: 'echo reuqest' and 'echo reply' These two types of ICMP packets are usually generated by the ping program))

tcpdump 'icmp[icmptype] != icmp-echo and icmp[icmptype] != icmp-echoreply'

tcpdump and wireshark

Wireshark (formerly ethereal) is a very simple and easy-to-use packet capture tool under Windows. But it is difficult to find a useful graphical packet capture tool under Linux.
Fortunately, there is Tcpdump. We can use the perfect combination of Tcpdump and Wireshark to achieve this: capture packets in Linux, and then analyze the packets in Windows.

tcpdump tcp -i eth1 -t -s 0 -c 100 and dst port ! 22 and src net 192.168.1.0/24 -w ./target.cap
(1)tcp: ip icmp arp rarp 和 tcp、udp、icmp这些选项等都要放到第一个参数的位置，用来过滤数据报的类型
(2)-i eth1 : 只抓经过接口eth1的包
(3)-t : 不显示时间戳
(4)-s 0 : 抓取数据包时默认抓取长度为68字节。加上-S 0 后可以抓到完整的数据包
(5)-c 100 : 只抓取100个数据包
(6)dst port ! 22 : 不抓取目标端口是22的数据包
(7)src net 192.168.1.0/24 : 数据包的源网络地址为192.168.1.0/24
(8)-w ./target.cap : 保存成cap文件，方便用ethereal(即wireshark)分析

使用tcpdump抓取HTTP包

tcpdump  -XvvennSs 0 -i eth0 tcp[20:2]=0x4745 or tcp[20:2]=0x4854
0x4745 为"GET"前两个字母"GE",0x4854 为"HTTP"前两个字母"HT"。

tcpdump 对截获的数据并没有进行彻底解码，数据包内的大部分内容是使用十六进制的形式直接打印输出的。显然这不利于分析网络故障，通常的解决办法是先使用带-w参数的tcpdump 截获数据并保存到文件中，然后再使用其他程序(如Wireshark)进行解码分析。当然也应该定义过滤规则，以避免捕获的数据包填满整个硬盘。

1、抓取回环网口的包：tcpdump -i lo

2、防止包截断：tcpdump -s0

3、以数字显示主机及端口：tcpdump -n

第一种是关于类型的关键字，主要包括host，net，port, 例如 host 210.27.48.2，指明 210.27.48.2是一台主机，net 202.0.0.0 指明 202.0.0.0是一个网络地址，port 23 指明端口号是23。如果没有指定类型，缺省的类型是host.

第二种是确定传输方向的关键字，主要包括src , dst ,dst or src, dst and src ,这些关键字指明了传输的方向。举例说明，src 210.27.48.2 ,指明ip包中源地址是210.27.48.2 , dst net 202.0.0.0 指明目的网络地址是202.0.0.0 。如果没有指明方向关键字，则缺省是src or dst关键字。

第三种是协议的关键字，主要包括fddi,ip,arp,rarp,tcp,udp等类型。Fddi指明是在FDDI(分布式光纤数据接口网络)上的特定 的网络协议，实际上它是"ether"的别名，fddi和ether具有类似的源地址和目的地址，所以可以将fddi协议包当作ether的包进行处理和 分析。其他的几个关键字就是指明了监听的包的协议内容。如果没有指定任何协议，则tcpdump将会监听所有协议的信息包。

除了这三种类型的关键字之外，其他重要的关键字如下：gateway, broadcast,less,greater,还有三种逻辑运算，取非运算是 'not ' '! ', 与运算是'and','&&;或运算 是'or' ,'||'；这些关键字可以组合起来构成强大的组合条件来满足人们的需要，下面举几个例子来说明。

普通情况下，直接启动tcpdump将监视第一个网络界面上所有流过的数据包。

# tcpdump 
tcpdump: listening on fxp0
11:58:47.873028 202.102.245.40.netbios-ns > 202.102.245.127.netbios-ns: udp 50
11:58:47.974331 0:10:7b:8:3a:56 > 1:80:c2:0:0:0 802.1d ui/C len=43
                       0000 0000 0080 0000 1007 cf08 0900 0000
                       0e80 0000 902b 4695 0980 8701 0014 0002
                       000f 0000 902b 4695 0008 00
11:58:48.373134 0:0:e8:5b:6d:85 > Broadcast sap e0 ui/C len=97
                       ffff 0060 0004 ffff ffff ffff ffff ffff
                       0452 ffff ffff 0000 e85b 6d85 4008 0002
                       0640 4d41 5354 4552 5f57 4542 0000 0000
                       0000 00

使用-i参数指定tcpdump监听的网络界面，这在计算机具有多个网络界面时非常有用，
使用-c参数指定要监听的数据包数量，
使用-w参数指定将监听到的数据包写入文件中保存
A想要截获所有210.27.48.1 的主机收到的和发出的所有的数据包：

#tcpdump host 210.27.48.1

B想要截获主机210.27.48.1 和主机210.27.48.2 或210.27.48.3的通信，使用命令：（在命令行中适用　括号时，一定要

#tcpdump host 210.27.48.1 and / (210.27.48.2 or 210.27.48.3 /)

C如果想要获取主机210.27.48.1除了和主机210.27.48.2之外所有主机通信的ip包，使用命令：

#tcpdump ip host 210.27.48.1 and ! 210.27.48.2

D如果想要获取主机210.27.48.1接收或发出的telnet包，使用如下命令：

#tcpdump tcp port 23 host 210.27.48.1

E 对本机的udp 123 端口进行监视 123 为ntp的服务端口

# tcpdump udp port 123

F 系统将只对名为hostname的主机的通信数据包进行监视。主机名可以是本地主机，也可以是网络上的任何一台计算机。下面的命令可以读取主机hostname发送的所有数据：

#tcpdump -i eth0 src host hostname

G 下面的命令可以监视所有送到主机hostname的数据包：

#tcpdump -i eth0 dst host hostname

H  我们还可以监视通过指定网关的数据包：

#tcpdump -i eth0 gateway Gatewayname

I 如果你还想监视编址到指定端口的TCP或UDP数据包，那么执行以下命令：

#tcpdump -i eth0 host hostname and port 80

J 如果想要获取主机210.27.48.1除了和主机210.27.48.2之外所有主机通信的ip包
，使用命令：

#tcpdump ip host 210.27.48.1 and ! 210.27.48.2

K 想要截获主机210.27.48.1 和主机210.27.48.2 或210.27.48.3的通信，使用命令
：（在命令行中适用　括号时，一定要

#tcpdump host 210.27.48.1 and / (210.27.48.2 or 210.27.48.3 /)

L 如果想要获取主机210.27.48.1除了和主机210.27.48.2之外所有主机通信的ip包，使用命令：

#tcpdump ip host 210.27.48.1 and ! 210.27.48.2

M 如果想要获取主机210.27.48.1接收或发出的telnet包，使用如下命令：

#tcpdump tcp port 23 host 210.27.48.1

第三种是协议的关键字，主要包括fddi,ip ,arp,rarp,tcp,udp等类型
除了这三种类型的关键字之外，其他重要的关键字如下：gateway, broadcast,less,
greater,还有三种逻辑运算，取非运算是 'not ' '! ', 与运算是'and','&&';或运算 是'o
r' ,'||'；
第二种是确定传输方向的关键字，主要包括src , dst ,dst or src, dst and src ,
如果我们只需要列出送到80端口的数据包，用dst port；如果我们只希望看到返回80端口的数据包，用src port。

#tcpdump –i eth0 host hostname and dst port 80  目的端口是80

或者

#tcpdump –i eth0 host hostname and src port 80  源端口是80  一般是提供http的服务的主机

如果条件很多的话  要在条件之前加and 或 or 或 not

#tcpdump -i eth0 host ! 211.161.223.70 and ! 211.161.223.71 and dst port 80

如果在ethernet 使用混杂模式 系统的日志将会记录
May  7 20:03:46 localhost kernel: eth0: Promiscuous mode enabled.
May  7 20:03:46 localhost kernel: device eth0 entered promiscuous mode
May  7 20:03:57 localhost kernel: device eth0 left promiscuous mode
tcpdump对截获的数据并没有进行彻底解码，数据包内的大部分内容是使用十六进制的形式直接打印输出的。显然这不利于分析网络故障，通常的解决办法是先使用带-w参数的tcpdump 截获数据并保存到文件中，然后再使用其他程序进行解码分析。当然也应该定义过滤规则，以避免捕获的数据包填满整个硬盘。

# tcpdump   -i eth1 src  host 211.167.237.199
00:02:03.096713 IP 211.167.237.199.ssh > 221.216.165.189.1467: P 2010208:2010352(144) ack 33377 win 8576
00:02:03.096951 IP 211.167.237.199.ssh > 221.216.165.189.1467: P 2010352:2010496(144) ack 33377 win 8576
00:02:03.100928 IP 211.167.237.199.ssh > 221.216.165.189.1467: P 2010496:2010640(144) ack 33377 win 8576
00:02:03.101165 IP 211.167.237.199.ssh > 221.216.165.189.1467: P 2010640:2010784(144) ack 33377 win 8576
00:02:03.102554 IP 211.167.237.199.ssh > 221.216.165.189.1467: P 2010784:2010928(144) ack 33425 win 8576

表明在00:02:03点的时候，211.167.237.199通过ssh源端口连接到221.216.165.189的1467端口

#tcpdump -i eth1 src host 211.167.237.199 and dst port 1467
00:09:27.603075 IP 211.167.237.199.ssh > 221.216.165.189.1467: P 180400:180544(144) ack 2833 win 8576
00:09:27.605631 IP 211.167.237.199.ssh > 221.216.165.189.1467: P 180544:180688(144) ack 2881 win 8576

截获所有由eth0进入、源地址(src)为192.168.0.5的主机(host)，并且(and)目标(dst)端口(port)为80的数据包

观看网卡传送、接收数据包的状态

$ netstat  -i
Kernel Interface table
Iface MTU Met RX-OK RX-ERR RX-DRP RX-OVR TX-OK TX-ERR TX-DRP TX-OVR Flg
eth0 1500  0  14639   0      0      0    5705    119    0     0   BMRU

Iface:  网卡
RX-OK RX-ERR RX-DRP RX-OVR : 网卡正确接收数据包的数量以及发生错误、流失、碰撞的总数
TX-OK TX-ERR TX-DRP TX-OVR : 网卡正确发送数据包的数量以及发生错误、流失、碰撞的总数

The above is the detailed content of How to use tcpdump to capture network packets under CentOS. For more information, please follow other related articles on the PHP Chinese website!