mmt-probe.conf

# This is the configuration file for MMT Probe

# the unique identifier given to the probe
probe-id = 3

#root stack of network protocol
# for Ethernet: 1
# for ieee802154: 800
# for Linux cooked capture: 624
stack-type = 1

# path to the license key file
license = "./license.key"

# enable reporting non-ip traffic, such as, ARP, PPP
enable-proto-without-session-report = false
# enable reporting information about fragmentation if IP packets
enable-ip-fragmentation-report = false

# perform defragemtation of IP packets
enable-ip-defragmentation = false

#perform reassembly of TCP segments. Enabling this option requires to enable the option "enable-ip-defragmentation"
enable-tcp-reassembly = false

# indicates the period of sampling in seconds, e.g., do statistics each 5 seconds
stats-period = 5

# dynamic configuration from outside of probe, such as, start, stop processing, or update some configurations
dynamic-config {
    enable = false
    ## indicate the descriptor of UNIX domain socket to receive control commands,
    descriptor = "/tmp/mmt.sock"
}

# option only when using DPDK to capture packets or to forwarding packets
dpdk-option = "--syslog=syslog --log-level=5 -c 0x55555555555"

input {
    # in case for PCAP the input mode can be ONLINE or OFFLINE, however for DPDK it's only ONLINE
    mode = ONLINE
    
    # input source for PCAP online mode (interface name) and for offline mode (pcap name),
    #   however for DPDK its interface port number
    #   in DPDK mode, MMT supports also multi-port inputs, 
    #    - e.g., source="0,1" will tell MMT to capture packets on port 0 and 1 
    #    - MMT will aggregate traffic on these 2 ports, thus 2 packets of one flow can be received on 2 different ports 
    source = "enp0s3"
    
    # maximal size of a packet
    snap-len = 0        # 0 means the default value 65535, apparently what tcpdump uses for -s 0 (only for PCAP)
}

# Each kind of report can output to different channel, such as, file, kafak, ...
# General setting of output will be in this block, such as, output format, ...
output {
    # output format: either JSON or CSV
    format = CSV
    
    # probe maintains a cache of reports to send them by bulk to output channels (file, mongodb, redis, kafka) to improve its execution performace.
    # The cache will be flushed:
    # - when the cache is full
    # - or periodically designed by the "cace-period" parameter
    cache-max = 100000 # number of messages in the cache

    # number of seconds representing a period to flush cache to destination (file, mongodb)
    #   in case of file-output, a new file will be created each 5 seconds if file-output.sample-file=true
    cache-period = 5
}

# Indicates outputing report to files
file-output {
    enable = true  # set to false to disable file output, true to enable
    output-file  = "data.csv"      # file name where the reports will be dumped:
    output-dir = "/opt/mmt/probe/result/report/online" # Location where files are written:
    # indicate output to multiple sample files or a single file
    #- if sample-file=true, a new sample file is created each x seconds given by output.cache-period
    sample-file = true

    # retains the last x sample files, if sample-file=true
    # - set to 0 to retain all files
    # - note that the value of retain-files must be greater than the value of thread_nb + 1
    retain-files = 80
}

# indicates REDIS output:
redis-output
{
    enable = false # set to false to disable publishing to REDIS , true to enable
                # publishing to REDIS:
    hostname = "localhost" # hostname of REDIS server
    port = 6379            # port number of REDIS server
    channel = "report"     #name of channel to which the messages will be published
}

# indicates KAFKA output:
kafka-output
{
    enable = false # set to false to disable publishing to KAFKA, true to enable
                   #  publishing to KAFKA:
    hostname = "localhost" # hostname of KAFKA server
    port = 9092            # port number of KAFKA server
    topic = "report"       #name of topic to which the message will be published
}

# indicates mongodb output:
mongodb-output {
    enable = false
    hostname = "localhost"
    port = 27017
    database = "mmt-data"

    # collection to store reports
    collection = "reports"

    # limit size (megabytes - MB = 1000*1000 bytes) of collection
    # - set 0 to unlimit
    # - if the size of collection reaches the limit, the oldest reports will be removed to maintain the limit
    limit-size  = 0
}

# socket output
socket-output
{
    enable = false # set to true to enable , false to disable:
    type = BOTH    # type of socket domain: either UNIX or INTERNET or BOTH
    descriptor = "/tmp/mmt-probe-output.sock" # required for UNIX domain.
            # It is a location where socket file descriptor is created.

    port = 5000            # Required for Internet domain.
    hostname = "127.0.0.1" # required for Internet domain.
}


# Indicates where the traffic data will be dumped to pcap files
dump-pcap
{
    enable       = false   # set to true/false to enable/disable dumping packet
    # output-dir = "/opt/mmt/probe/" # Location where files will be located
    output-dir   = "/opt/mmt/probe/pcap/" # Location where files will be located
    protocols    = {"unknown"} # List of protocols appearing in a packet which will be dumped
    period       = 60 # New pcap file is created every x seconds. 0 means default value 3600 seconds
    retain-files = 50   # retains the last x sampled files,
    snap-len     = 65355 # set snaplength in the output pcap files
    
}

forward-packet
{
    enable     = false
    output-nic = "eth0"
    nb-copies  = 1 #number of copies of a packet to be sent
    snap-len   = 0 #specifies the snapshot length to be set on the handle.
    promisc    = 1 #specifies whether the interface is to be put into promiscuous mode. If promisc is non-zero, promiscuous mode will be set, otherwise it will not be set.
    default    = FORWARD #default action when packets are not selected/satisfied by any rule
                         # either FORWARD to forward the packets or DROP to drop the packets
    
    #forward packets to a target using SCTP protocol: MMT will be a SCTP client, 
    # - it connects to the given "sctp-host" at "sctp-port"
    # - the SCTP packets' payload will be sent to the target using this SCTP connection
    target-protocols = { SCTP }
    target-hosts     = { "127.0.0.5" }
    target-ports     = { 38412 }
}


# indicates the threads configuration options
thread-nb    = 0    # this is the number of threads MMT will use for processing packets.
                    # 0: use only one thread to read then process packets
                    # 1: use one thread to process packets, another to read packets
thread-queue = 262144 # 8388608 # 4194304 # this is the maximum number of packets being queued for processing that a single thread accepts (only for PCAP).
            # 0 means MMT decides how many packets to queue (default =1000)
            # If a packet is dispatched for a thread having a full queue, then the packet will be dropped.


# smp-security
security
{
    enable        = false
    thread-nb     = 0   # the number of security threads per one probe thread , e .g . , if we have 16 probe threads and thread-nb = x ,
                # then x*16 security threads will be used .
                # If set to zero this means that the security analysis will be done by the threads of the probe .
    
    exclude-rules = ""  # Range of rules to be excluded from the verification
    
    rules-mask    = ""  # Mapping of rules to the security threads:
            # Format: rules-mask = (thread-index:rule-range);
            # thread-index = a number greater than 0
            # rule-range = number greater than 0, or a range of numbers greater than 0.
            # Example: If we have thread-nb = 3 and "(1:1,2,4-6)(2:3)" ,
            # this means that:
            # thread 1 verifies rules 1 ,2 ,4 ,5 ,6;
            # thread 2 verifies only rule 3; and
            # thread 3 verifies the rest
            # Note: if we have thread-nb = 2 and "(1:1)(2:3)", then only rules 1 and 3 are verified (the others are not)

    output-channel = {} # Reports are sent to the output channels . The default value is a file .
            # More than one output channel is possible . In this case , the value should be a set of comma - separated values ; for example: { redis , kafka , file, mongodb, socket }.
            # Reports are sent to output channels only when the global parameters are enable ( file-output.enable = true ,redis-output.enable = true , kafka-output.enable = true, mongodb-output.enable = true)

    report-rule-description = true # true to include rule's description into the alert reports,
             # otherwise it will be excluded (thus rules's descriptions will be an empty string in the reports)
             # Excluding rules's descriptions will reduce the size of reports.
	
    ignore-remain-flow = false # ignore the security verification on the left of an IP flow when an alert is detected.
	
	
	#The 3 following parameters are specific for libmmt_security2 to override its default configuration
    input.max-message-size = 60000  #size of a message, in bytes, to encapsulate data sending from MMT-Probe to MMT-Security
    security.max-instances = 100000 # Maximum number of instances of a rule
    security.smp.ring-size = 1000   # Maximum number of messages that will be stored in a ring buffer
    
    ip-encapsulation-index = LAST   # If traffic is ip-in-ip, this option selects which IP will be analysed.
                                    # - FIRST: first ip in the protocol hierarchy
                                    # - LAST:  last ip in the protocol hierarchy
                                    # - i: i-th ip in ther protocol hierarchy. 
                                    # For example, given ETH.IP.UDP.GTP.IP.TCP.VPN.IP.SSL,
                                    #   - FIRST, or 1, indicates IP after ETH
                                    #   - LAST, or any number >= 3, indicates IP after VPN
                                    #   - 2 indicates IP after GTP
                                    # NOTE: this option will be ignored in non ip-in-ip traffic
}

# CPU and memory usage monitor of the machin on that MMT-Probe is running
system-report
{   enable = false # set to 1 to perform cpu-mem reporting, false to disable it
    period = 5
    output-channel = {} # see security.output-channel for more detailed information
}

# indicates output for behaviour analysis
behaviour
{
    enable  = false  # set to true to enable , false to disable:
    output-dir = "/opt/mmt/probe/result/behaviour/online/"  # Folder to write the output on
    output-file  = "data.csv"      # file name where the reports will be dumped:

    # retains the last x sample files, if sample-file=true
    # - set to 0 to retain all files
    # - note that the value of retain-files must be greater than the value of thread_nb + 1
    retain-files = 160
}

# Indicates that FTP data reconstruction should be done .
# To enable the reconstruction, also need to enable the options "enable-tcp-reassembly" and "enable-ip-defragmentation"
reconstruct-data ftp
{
    enable = false
    output-dir  = "/tmp/" # indicates the folder where the output file is created:
    output-channel = {} # see security.output-channel for more detailed information
}

# indicate that HTTP response data should be written to files
# To enable the reconstruction, also need to enable the options "enable-tcp-reassembly" and "enable-ip-defragmentation"
reconstruct-data http
{
     enable = false
     output-dir = "/tmp/" # location where the files are reconstructed .
     output-channel = {}
}

data-output
{
    include-user-agent = 32     # Indicates the threshold in terms of data volume for parsing the user agent in Web traffic.
                # The value is in kiloBytes ( kB ) . If the value is zero , this indicates that the parsing of the user agent should be done .
                # To disable the user agent parsing, set the threshold to a negative value (-1).

}

# Specifies the criteria to consider a flow as micro flow.
# - IF a flow has some characters, such as, number of packets/bytes, being less than some litmits,
#   then it will be considered as a micro flow.
# - A micro-flow is identified by protocol ID. This means that a micro-flow represents several flows having
#    + the same protocol ID, and,
#    + number of bytes or packets are less than the thresholds
# - Set value of a limits/threshold to zero to unlimit it
micro-flows
{
    enable = false
    packet-threshold =   2     # packets count threshold to consider a flow as a micro flow:
    byte-threshold   = 100     # data volume threshold to consider a flow as a micro flow:

    # a micro-flow will be reported to its output-channel:
    # - at the end of execution of MMT-Probe
    # - or when one of its stats (packets, bytes, flows) is greater than or equal some litmit as specified below.
    report-packet-count  =  1000    # packets count threshold to report micro flows aggregated statistics:
    report-byte-count    = 10000    # data volume threshold  to report micro flows aggregated statistics:
    report-flow-count    =   500    # limit nb of flows

    output-channel = {}     # see security.output-channel for more detailed information
}

# indicates the strategy for RADIUS reporting
radius-report
{
    enable = false
    message-id = 0     # indicates the kind of message one needs to report .
                       # set to 0 to report all messages .
                       # set to a number (from 1 to 255) to indicate the code of message to report, e.g., 1: Access-Request, 2: Access-Accept, 45: CoA-NAK, etc.
    output-channel = {} # see security.output-channel for more detailed information
}


# Specifies the session timeout time, in seconds, for different type of applications
session-timeout
{
    default-session-timeout = 60  # Default timeout (in seconds) for sessions.
    short-session-timeout =   15
    long-session-timeout =  6000  # This is reasonable for Web and SSL connections especially when long polling is used. 
                                  # Usually applications have a long polling period of about 3~5 minutes.
    live-session-timeout  = 1500  #For persistent connections like messaging applications and so on
}

# Indicates session based reporting
session-report
{
    enable = true
    output-channel = {file} # see security.output-channel for more detailed information

    # enable/disable specific reports for specific protocol applications
    ftp =  false
    http = false
    rtp =  false
    ssl =  false
    gtp =  true

    rtt-base = CAPTOR # Order of timestamp's origin that is used to calculate RTT in QOS reports
                # - SENDER: timestamp being marked in packet by its sender, e.g., tcp option TSVal, TSerc
                #           This timestamp is available only on certain monitoring protocol, e.g., TCP
                # - CAPTOR: timestamp being marked in packet at the captured moment by its captor (e.g., tcpdump, wireshark)
                #            This timestamp is always available.
                # The value of this rtt setting can be one of the followings:
                # - CAPTOR (by default): use only CAPTOR.
                # - SENDER: use only SENDER. Ignore if it is not available
                # - PREFER_SENDER:  use SENDER if it is available, otherwise CAPTOR
}

# Indicates event based reporting: do a report when an event occurs.
# For example: when MMT-Probe sees a IP packet having IP.SRC, it will report ARP.AR_HLN and IP.SRC.
# If a packet does not containt an attribute, e.g., ARP.AR_HLN, the attribute's value is replaced by an empty value (either 0 or "").
# We may have multiple "event-report" but its name, e.g., "ip-event", must be unique.
# For example, event-report http-url{ ...} event-report sample2{ ...}
event-report ip-event
{
   enable = false
   event = "ip.src" # Indicates the event:
   attributes = {"arp.ar_hln", "ip.src", "meta.proto_hierarchy"} # Indicates the list of attributes that are reported when a event is triggered:
}

event-report iot
{
    enable = false
    event = "ieee802154.sequence_number" # Indicates the event:
    attributes = {"ieee802154.src_address_extended", "ieee802154.dst_address_extended", "meta.packet_len"} # Indicates the list of attributes that are reported when a event is triggered:
    output-channel = {} # see security.output-channel for more detailed information
}