Pycapa performs network packet capture, both off-the-wire and from a Kafka topic, which is useful for the testing and development of Apache Metron. It is not intended for production use. The tool will capture packets from a specified interface and push them into a Kafka Topic. The tool can also do the reverse. It can consume packets from Kafka and reconstruct each network packet. This can then be used to create a libpcap-compliant file or even to feed directly into a tool like Wireshark to monitor ongoing activity.
General notes on the installation of Pycapa.
- Python 2.7 is required.
- The following package dependencies are required and can be installed automatically with
pip. The requirements are installed as part of step 4
- These instructions can be used directly on CentOS 7+.
- Other Linux distributions that come with Python 2.7 can use these instructions with some minor modifications.
-
Install system dependencies including the core development tools, Python libraries and header files, and Libpcap libraries and header files. On CentOS 7+, you can install these requirements with the following command.
yum -y install "@Development tools" python-devel libpcap-devel -
Install Librdkafka at your chosen $PREFIX.
export PREFIX=/usr wget https://github.com/edenhill/librdkafka/archive/v0.11.5.tar.gz -O - | tar -xz cd librdkafka-0.11.5/ ./configure --prefix=$PREFIX make make install -
Add Librdkafka to the dynamic library load path.
echo "$PREFIX/lib" >> /etc/ld.so.conf.d/pycapa.conf ldconfig -v -
Install Pycapa. This assumes that you already have the Metron source code on the host.
cd metron/metron-sensors/pycapa pip install -r requirements.txt python setup.py install
- These instructions can be used directly on CentOS 6 - useful for developers using the Full Dev Vagrant test box.
- Older distributions, like CentOS 6, that come with Python 2.6 installed, should install Python 2.7 within a virtual environment and then run Pycapa from within the virtual environment.
-
Set up a couple environment variables.
PYCAPA_HOME=/opt/pycapa PYTHON27_HOME=/opt/rh/python27/root -
Install required packages.
for item in epel-release centos-release-scl "@Development tools" python27 python27-scldevel python27-python-virtualenv libpcap-devel libselinux-python; do yum install -y $item; done -
Setup Pycapa directory.
mkdir $PYCAPA_HOME && chmod 755 $PYCAPA_HOME -
Create the virtualenv.
export LD_LIBRARY_PATH="/opt/rh/python27/root/usr/lib64" cd $PYCAPA_HOME ${PYTHON27_HOME}/usr/bin/virtualenv pycapa-venv -
Install Librdkafka at your chosen $PREFIX.
export PREFIX=/usr wget https://github.com/edenhill/librdkafka/archive/v0.11.5.tar.gz -O - | tar -xz cd librdkafka-0.11.5/ ./configure --prefix=$PREFIX make make install -
Add Librdkafka to the dynamic library load path.
echo "$PREFIX/lib" >> /etc/ld.so.conf.d/pycapa.conf ldconfig -v -
Copy the Pycapa source files from the Metron project to your chosen $PYCAPA_HOME (e.g.
/opt/pycapa). You should have pycapa source files in/opt/pycapa/pycapa.scp -r metron-sensors/pycapa root@node1:$PYCAPA_HOME -
Install Pycapa using the
pycapa-venvvirtualenv you created earlier.cd ${PYCAPA_HOME}/pycapa # activate the virtualenv source ${PYCAPA_HOME}/pycapa-venv/bin/activate pip install -r requirements.txt python setup.py install -
Special notes on running pycapa on Centos 6. You should run it using the virtualenv.
cd ${PYCAPA_HOME}/pycapa-venv/bin pycapa --producer --kafka-topic pcap --interface eth1 --kafka-broker $BROKERLIST
Note: To deactivate your virtualenv, simply type "deactivate" and hit enter.
Pycapa has two primary runtime modes.
-
Producer Mode: Pycapa can capture packets from a network interface and forward those packets to a Kafka topic. Pycapa embeds the raw network packet data in the Kafka message body. The message key contains the timestamp indicating when the packet was captured in microseconds from the epoch, in network byte order.
-
Consumer Mode: Pycapa can also perform the reverse operation. It can consume packets from Kafka and reconstruct each network packet. This can then be used to create a libpcap-compliant file or even to feed directly into a tool like Wireshark to monitor activity.
$ pycapa --help
usage: pycapa [-h] [-p] [-c] [-k KAFKA_BROKERS] [-t KAFKA_TOPIC]
[-o {begin,end,stored}] [-i NETWORK_IFACE] [-m MAX_PACKETS]
[-pp PRETTY_PRINT] [-ll LOG_LEVEL] [-X KAFKA_CONFIGS]
[-s SNAPLEN]
optional arguments:
-h, --help show this help message and exit
-p, --producer sniff packets and send to kafka
-c, --consumer read packets from kafka
-k KAFKA_BROKERS, --kafka-broker KAFKA_BROKERS
kafka broker(s) as host:port
-t KAFKA_TOPIC, --kafka-topic KAFKA_TOPIC
kafka topic
-o {begin,end,stored}, --kafka-offset {begin,end,stored}
kafka offset to consume from; default=end
-i NETWORK_IFACE, --interface NETWORK_IFACE
network interface to listen on
-m MAX_PACKETS, --max-packets MAX_PACKETS
stop after this number of packets
-pp PRETTY_PRINT, --pretty-print PRETTY_PRINT
pretty print every X packets
-ll LOG_LEVEL, --log-level LOG_LEVEL
set the log level; DEBUG, INFO, WARN
-X KAFKA_CONFIGS define a kafka client parameter; key=value
-s SNAPLEN, --snaplen SNAPLEN
capture only the first X bytes of each packet;
default=65535
Capture 10 packets from the eth0 network interface and forward those to a Kafka topic called pcap running on localhost:9092. The process will not terminate until all messages have been delivered to Kafka.
$ pycapa --producer \
--interface eth0 \
--kafka-broker localhost:9092 \
--kafka-topic pcap \
--max-packets 10
INFO:root:Connecting to Kafka; {'bootstrap.servers': 'localhost:9092', 'group.id': 'AWBHMIAESAHJ'}
INFO:root:Starting packet capture
INFO:root:Waiting for '6' message(s) to flush
INFO:root:'10' packet(s) in, '10' packet(s) out
Capture packets until SIGINT is received (the interrupt signal sent when entering CTRL-C in the console.) In this example, nothing will be reported as packets are captured and delivered to Kafka. Simply wait a few seconds, then type CTRL-C and the number of packets will be reported.
$ pycapa --producer \
--interface en0 \
--kafka-broker localhost:9092 \
--kafka-topic pcap
INFO:root:Connecting to Kafka; {'bootstrap.servers': 'localhost:9092', 'group.id': 'EULLGDOMZDCT'}
INFO:root:Starting packet capture
^C
INFO:root:Clean shutdown process started
INFO:root:Waiting for '2' message(s) to flush
INFO:root:'21' packet(s) in, '21' packet(s) out
While capturing packets, output diagnostic information every 5 packets. Diagnostics will report when packets have been received from the network interface and when they have been successfully delivered to Kafka.
$ pycapa --producer \
--interface eth0 \
--kafka-broker localhost:9092 \
--kafka-topic pcap \
--pretty-print 5
INFO:root:Connecting to Kafka; {'bootstrap.servers': 'localhost:9092', 'group.id': 'UAWINMBDNQEH'}
INFO:root:Starting packet capture
Packet received[5]
Packet delivered[5]: date=2017-05-08 14:48:54.474031 topic=pcap partition=0 offset=29086 len=42
Packet received[10]
Packet received[15]
Packet delivered[10]: date=2017-05-08 14:48:58.879710 topic=pcap partition=0 offset=0 len=187
Packet delivered[15]: date=2017-05-08 14:48:59.633127 topic=pcap partition=0 offset=0 len=43
Packet received[20]
Packet delivered[20]: date=2017-05-08 14:49:01.949628 topic=pcap partition=0 offset=29101 len=134
Packet received[25]
^C
INFO:root:Clean shutdown process started
Packet delivered[25]: date=2017-05-08 14:49:03.589940 topic=pcap partition=0 offset=0 len=142
INFO:root:Waiting for '1' message(s) to flush
INFO:root:'27' packet(s) in, '27' packet(s) out
Consume 10 packets and create a libpcap-compliant pcap file.
```
$ pycapa --consumer \
--kafka-broker localhost:9092 \
--kafka-topic pcap \
--max-packets 10 \
> out.pcap
$ tshark -r out.pcap
1 0.000000 199.193.204.147 → 192.168.0.3 TLSv1.2 151 Application Data
2 0.000005 199.193.204.147 → 192.168.0.3 TLSv1.2 1191 Application Data
3 0.000088 192.168.0.3 → 199.193.204.147 TCP 66 54788 → 443 [ACK] Seq=1 Ack=86 Win=4093 Len=0 TSval=961284465 TSecr=943744612
4 0.000089 192.168.0.3 → 199.193.204.147 TCP 66 54788 → 443 [ACK] Seq=1 Ack=1211 Win=4058 Len=0 TSval=961284465 TSecr=943744612
5 0.948788 192.168.0.3 → 192.30.253.125 TCP 54 54671 → 443 [ACK] Seq=1 Ack=1 Win=4096 Len=0
6 1.005175 192.30.253.125 → 192.168.0.3 TCP 66 [TCP ACKed unseen segment] 443 → 54671 [ACK] Seq=1 Ack=2 Win=31 Len=0 TSval=2658544467 TSecr=961240339
7 1.636312 fe80::1286:8cff:fe0e:65df → ff02::1 ICMPv6 134 Router Advertisement from 10:86:8c:0e:65:df
8 2.253052 192.175.27.112 → 192.168.0.3 TLSv1.2 928 Application Data
9 2.253140 192.168.0.3 → 192.175.27.112 TCP 66 55078 → 443 [ACK] Seq=1 Ack=863 Win=4069 Len=0 TSval=961286699 TSecr=967172238
10 2.494769 192.168.0.3 → 224.0.0.251 MDNS 82 Standard query 0x0000 PTR _googlecast._tcp.local, "QM" question
```
Consume 10 packets from the Kafka topic pcap running on localhost:9092, then pipe those into Wireshark for DPI.
$ pycapa --consumer \
--kafka-broker localhost:9092 \
--kafka-topic pcap \
--max-packets 10 \
| tshark -i -
Capturing on 'Standard input'
1 0.000000 ArrisGro_0e:65:df → Apple_bf:0d:43 ARP 56 Who has 192.168.0.3? Tell 192.168.0.1
2 0.000044 Apple_bf:0d:43 → ArrisGro_0e:65:df ARP 42 192.168.0.3 is at ac:bc:32:bf:0d:43
3 0.203495 fe80::1286:8cff:fe0e:65df → ff02::1 ICMPv6 134 Router Advertisement from 10:86:8c:0e:65:df
4 2.031988 192.168.0.3 → 96.27.183.249 TCP 54 55110 → 443 [ACK] Seq=1 Ack=1 Win=4108 Len=0
5 2.035816 192.30.253.125 → 192.168.0.3 TLSv1.2 97 Application Data
6 2.035892 192.168.0.3 → 192.30.253.125 TCP 66 54671 → 443 [ACK] Seq=1 Ack=32 Win=4095 Len=0 TSval=961120495 TSecr=2658503052
7 2.035994 192.168.0.3 → 192.30.253.125 TLSv1.2 101 Application Data
8 2.053866 96.27.183.249 → 192.168.0.3 TCP 66 [TCP ACKed unseen segment] 443 → 55110 [ACK] Seq=1 Ack=2 Win=243 Len=0 TSval=728145145 TSecr=961030381
9 2.083872 192.30.253.125 → 192.168.0.3 TCP 66 443 → 54671 [ACK] Seq=32 Ack=36 Win=31 Len=0 TSval=2658503087 TSecr=961120495
10 3.173189 fe80::1286:8cff:fe0e:65df → ff02::1 ICMPv6 134 Router Advertisement from 10:86:8c:0e:65:df
10 packets captured
The probe can be used in a Kerberized environment. The Python client README (https://github.com/confluentinc/confluent-kafka-python) has an important note for Kerberos case that the pre-built Linux wheels do NOT contain SASL Kerberos support. You will need to use the non-binary wheel to install confluent-kafka-python and build/install librdkafka separately. Follow these additional steps to use Pycapa with Kerberos. The following assumptions have been made. These may need altered to fit your environment.
- The Kafka broker is at
kafka1:6667 - Zookeeper is at
zookeeper1:2181 - The Kafka security protocol is
SASL_PLAINTEXT - The keytab used is located at
/etc/security/keytabs/metron.headless.keytab - The service principal is
[email protected]
-
If it is not, ensure that you have
libsaslorlibsasl2installed. On CentOS, this can be installed with the following command.yum install -y cyrus-sasl cyrus-sasl-devel cyrus-sasl-gssapi -
Build Librdkafka with SASL support (
--enable-sasl) and install at your chosen $PREFIX.wget https://github.com/edenhill/librdkafka/archive/v0.11.5.tar.gz -O - | tar -xz cd librdkafka-0.11.5/ ./configure --prefix=$PREFIX --enable-sasl make make install -
Validate Librdkafka does indeed support SASL. Run the following command and ensure that
saslis returned as a built-in feature.$ examples/rdkafka_example -X builtin.features builtin.features = gzip,snappy,ssl,sasl,regex,lz4,sasl_gssapi,sasl_plain,sasl_scram,plugins -
The source install of confluent-kafka.
If you have already installed, remove the binary wheel python client first, repeat until it says no longer installed
pip uninstall -y confluent-kafkapip install --no-binary :all: confluent-kafka -
Grant access to your Kafka topic. In this example the topic is simply named
pcap.${KAFKA_HOME}/bin/kafka-acls.sh \ --authorizer kafka.security.auth.SimpleAclAuthorizer \ --authorizer-properties zookeeper.connect=zookeeper1:2181 \ --add \ --allow-principal User:metron \ --topic pcap ${KAFKA_HOME}/bin/kafka-acls.sh \ --authorizer kafka.security.auth.SimpleAclAuthorizer \ --authorizer-properties zookeeper.connect=zookeeper1:2181 \ --add \ --allow-principal User:metron \ --group pycapa -
Use Pycapa as you normally would, but append the following three additional parameters
-
security.protocol -
sasl.kerberos.keytab -
sasl.kerberos.principal$ pycapa --producer \ --interface eth0 \ --kafka-broker kafka1:6667 \ --kafka-topic pcap --max-packets 10 \ -X security.protocol=SASL_PLAINTEXT \ -X sasl.kerberos.keytab=/etc/security/keytabs/metron.headless .keytab \ -X [email protected] INFO:root:Connecting to Kafka; {'sasl.kerberos.principal': '[email protected]', 'group.id': 'ORNLVWJZZUAA', 'security.protocol': 'SASL_PLAINTEXT', 'sasl.kerberos.keytab': '/etc/security/keytabs/metron.headless.keytab', 'bootstrap.servers': 'kafka1:6667'} INFO:root:Starting packet capture INFO:root:Waiting for '1' message(s) to flush INFO:root:'10' packet(s) in, '10' packet(s) out
Use the following two command-line arguments to get detailed logging.
-X debug=all --log-level DEBUG
When I run Pycapa against a Kafka broker with Kerberos enabled, why do I get an error like "No such configuration property: 'sasl.kerberos.principal'"?
This can be a confusing error message because sasl.kerberos.principal is indeed a valid property for librdkafka as defined here. This is most likely because Pycapa is running against a version of Librdkafka without SASL support enabled. This might happen if you have accidentally installed multiple versions of Librdkafka and Pycapa is unexpectedly using the version without SASL support enabled.
Bottom Line: Make sure that Pycapa is running against a version of Librdkafka with SASL support enabled.