Monitoring for Large-Scale Networks

SiLK

rwfilter

NAME

rwfilter - Choose which SiLK Flow records to process

SYNOPSIS

  rwfilter INPUT_ARGS OUTPUT_ARGS PARTITIONING_ARGS [MISC_ARGS]

Selection switches, input switches, or input files are required:

  rwfilter ...
        {{ [--class=CLASS] [--type={all | TYPE[,TYPE ...]}]
           | [--flowtypes=CLASS/TYPE[,CLASS/TYPE ...]] }
         [--sensors=SENSOR[,SENSOR ...]]
         [--start-date=YYYY/MM/DD[:HH] [--end-date=YYYY/MM/DD[:HH]]]
         [--data-rootdir=ROOT_DIRECTORY] [--print-missing-files] }
        | [--input-pipe=INPUT_PATH]
        | [--xargs] | [--xargs=INPUT_PATH]
        | [INPUT_PATH [INPUT_PATH...]]

One or more output switches are required:

  rwfilter ...
        [--all-destination=ALL_PATH [--all-destination=ALL_PATH ...]]
        [--fail-destination=FAIL_PATH [--fail-destination=FAIL_PATH ...]]
        [--pass-destination=PASS_PATH [--pass-destination=PASS_PATH ...]]
        [{ --print-statistics[=STATS_PATH]
           | --print-volume-statistics[=STATS_PATH] }]

One or more partitioning switches are required:

  rwfilter ...
        [--ack-flag=SCALAR] [--active-time=TIME_WINDOW]
        [{--any-address=IP_WILDCARD | --not-any-address=IP_WILDCARD}]
        [--any-cc=COUNTRY_CODE_LIST]
        [{--any-cidr=IP_OR_CIDR_LIST | --not-any-cidr=IP_OR_CIDR_LIST}]
        [--any-index=INTEGER_LIST]
        [{--anyset=IP_SET_FILENAME | --not-anyset=IP_SET_FILENAME}]
        [--aport=INTEGER_LIST] [--application=INTEGER_LIST]
        [--attributes=ATTRIBUTES_LIST]
        [--bytes=INTEGER_RANGE] [--bytes-per-packet=DECIMAL_RANGE]
        [--cwr-flag=SCALAR]
        [{--daddress=IP_WILDCARD | --not-daddress=IP_WILDCARD}]
        [--dcc=COUNTRY_CODE_LIST]
        [{--dcidr=IP_OR_CIDR_LIST | --not-dcidr=IP_OR_CIDR_LIST}]
        [{--dipset=IP_SET_FILENAME | --not-dipset=IP_SET_FILENAME}]
        [--dport=INTEGER_LIST] [--dtype=SCALAR]
        [--duration=DECIMAL_RANGE] [--ece-flag=SCALAR]
        [--etime=TIME_WINDOW] [--fin-flag=SCALAR]
        [--flags-all=HIGH_MASK_FLAGS_LIST]
        [--flags-initial=HIGH_MASK_FLAGS_LIST]
        [--flags-session=HIGH_MASK_FLAGS_LIST]
        [--icmp-code=INTEGER_LIST] [--icmp-type=INTEGER_LIST]
        [--input-index=INTEGER_LIST] [--ip-version=INTEGER_LIST]
        [--ipa-src-expr=IPA_EXPR] [--ipa-dst-expr=IPA_EXPR]
        [--ipa-any-expr=IPA_EXPR]
        [{--next-hop-id=IP_WILDCARD | --not-next-hop-id=IP_WILDCARD}]
        [{--nhcidr=IP_OR_CIDR_LIST | --not-nhcidr=IP_OR_CIDR_LIST}]
        [{--nhipset=IP_SET_FILENAME | --not-nhipset=IP_SET_FILENAME}]
        [--output-index=INTEGER_LIST] [--packets=INTEGER_RANGE]
        [--pmap-file=MAPNAME:PATH [--pmap-file=MAPNAME:PATH ...]
         { [--pmap-src-MAPNAME=LABELS] [--pmap-dst-MAPNAME=LABELS]
           [--pmap-any-MAPNAME=LABELS] } ]
        [--protocol=INTEGER_LIST] [--psh-flag=SCALAR]
        [--python-expr=PYTHON_EXPR]
        [--python-file=FILENAME [--python-file=FILENAME ...]]
        [--rst-flag=SCALAR]
        [{--saddress=IP_WILDCARD | --not-saddress=IP_WILDCARD}]
        [--scc=COUNTRY_CODE_LIST]
        [{--scidr=IP_OR_CIDR_LIST | --not-scidr=IP_OR_CIDR_LIST}]
        [{--sipset=IP_SET_FILENAME | --not-sipset=IP_SET_FILENAME}]
        [--sport=INTEGER_LIST] [--stime=TIME_WINDOW] [--stype=SCALAR]
        [--syn-flag=SCALAR] [--tcp-flags=TCP_FLAGS]
        [--tuple-file=TUPLE_FILENAME { [--tuple-fields=FIELDS]
                                       [--tuple-direction=DIRECTION]
                                       [--tuple-delimiter=CHAR] } ]
        [--urg-flag=SCALAR]

Miscellaneous switches:

  rwfilter ...
        [--compression-method=COMP_METHOD] [--dry-run]
        [--max-fail-records=N] [--max-pass-records=N]
        [--note-add=TEXT] [--note-file-add=FILE]
        [--plugin=PLUGIN [--plugin=PLUGIN ...]]
        [--print-filenames] [--site-config-file=FILENAME]
        [--threads=N]

Help switches:

  rwfilter [--pmap-file=MAPNAME:PATH [--pmap-file=MAPNAME:PATH ...]]
        [--plugin=PLUGIN ...] [--python-file=PATH]
        [--data-rootdir=ROOT_DIRECTORY] [--site-config-file=FILENAME]
        --help

  rwfilter --version

DESCRIPTION

rwfilter serves two purposes: (1) It acts as an interface to the data store to select which SiLK Flow records to process, and (2) it partitions those records into one or more pass and/or fail streams.

The "Selection Switches" let one choose flow records from the SiLK data store by specifying where the flow was collected (its sensor), the date of collection, and/or the flow's direction. The act of selecting records from the data store is sometimes called a "data pull".

The "Partitioning Switches" describe various types of traffic behavior (e.g., TCP traffic, or all traffic going to port 80). When a flow record matches all of the behaviors, it can be written to a pass stream (i.e., file). If a record fails to match any of these behavior predicates, it can be written to a fail stream. (You may also write every record rwfilter reads to an all stream.) These output streams from rwfilter are always binary SiLK Flow records. The output must be either written to a file or piped into another tool in the SiLK Suite, and rwfilter complains if it determines you are attempting to send the stream to a terminal. To view the records, pipe the records into rwcut(1).

In addition to the partitioning switches built in to rwfilter, additional partitioning predicates can be created as C or PySiLK plug-ins, and these can be loaded into rwfilter using the --plugin and/or --python-file switches as described below.

Instead of using the selection switches to choose flow records from the data store, rwfilter can apply the partitioning switches to existing files of SiLK flow records---such as files generated by a previous invocation of rwfilter. To run rwfilter in this mode, you may

specify, on the command line, the files and/or named pipes from which rwfilter should read SiLK Flow records. Specifying stdin or - or the command line causes rwfilter to read flow records from the standard input.
use the --input-pipe switch to specify a named pipe, or specify stdin or - as the argument to this switch to have rwfilter read flow records from the standard input.
use the --xargs switch to specify a file that contains the names of the input files to process. When --xargs is used without an argument, rwfilter attempts to read the names of the file from the standard input. The name of each input file must appear on a single line.

When rwfilter is reading flow records from input files, some of the selection switches act as partitioning switches. The remaining selection switches may not be specified when using the alternate forms of input, and it is an error to specify multiple types of input.

Unlike many other tools in the SiLK tool suite, rwfilter requires that you specify one or more "Output Switches" that tell rwfilter what types of output to produce.

Finally, there are "Miscellaneous Switches" that control other aspects of rwfilter.

Read Selection Argument Values from a File

As of SiLK 3.20, the "Selection Switches" --class, --type, --flowtypes, and --sensors accept a value in the form "@PATH", where @ is the "at" character (ASCII 0x40) and PATH names a file or a path to a file. For example, the following reads the name of types from the file t.txt and uses the sensors S3, S7, and the names and/or IDs read from /tmp/sensor.txt:

 rwfilter --type=@t.txt --sensors=S3,@/tmp/sensor.txt,S7 ...

Multiple @PATH values are allowed within a single argument. If the name of the file is -, the names are read from the standard input.

The file must be a text file. Blank lines are ignored as are comments, which begin with the # character and continue to the end of the line. Whitespace at the beginning and end of a line is ignored as is whitespace that surrounds commas; all other whitespace within a line is significant.

A file may contain a value on each line and/or multiple values on a line separated by commas and optional whitespace. For example:

 # Sensor 4
       S4
 # The first sensors
 S0, S1,S2
 S3     # Sensor 3

An attempt to use an @PATH directive in a file is an error.

When rwfilter is parsing the name of a file, it converts the sequences @, and @@ to , and @, respectively. For example, --class=@cl@@ss.txt@,v reads the class from the file cl@ss.txt,v. It is an error if any other character follows an embedded @ (--flowtypes=@f@il contains @i) or if a single @ occurs at the end of the name (--sensor=@errat@).

OPTIONS

Option names may be abbreviated if the abbreviation is unique or is an exact match for an option. A parameter to an option may be specified as --arg=param or --arg param, though the first form is required for options that take optional parameters.

Selection Switches

To read files from the data store, use the following options to specify which files to process. When rwfilter gets its input from files listed on the command line or from the --xargs or --input-pipe switches, the first four switches (--class, --type, --flowtypes, and --sensors) act as partitioning switches, and specifying any other selection switch produces an error.

--class={CLASS | @PATH}

The --class switch is used to specify a group of data files to process. Only a single class may be selected with the --class switch; for multiple classes, use the --flowtypes switch. The argument may be "@PATH" which causes rwfilter to open the file PATH and read the class name from it; see "Read Selection Argument Values from a File" for details. Classes are defined in the silk.conf(5) site configuration file. If neither the --class nor --flowtypes option is given, the default-class as specified in silk.conf is used. To see the available classes and the default class, either examine the output from rwfilter --help or invoke rwsiteinfo(1) with the switch --fields=class,default-class.

--type={all | TYPE[,TYPE,@PATH ...]}

The --type predicate further specifies data within the selected CLASS by listing the TYPEs of traffic to process. The switch takes either the keyword all to select all types for CLASS or a comma-separated list of types names and "@PATH" directives, where @PATH tells rwfilter to read type names from the file PATH; see "Read Selection Argument Values from a File" for details. Types are defined in silk.conf, they typically refer to the direction of the flow, and they may vary by class. When neither the --type nor --flowtypes switch is given, a list of default types is used: The default-type list is determined by the value of CLASS, and the default types often include only incoming traffic. To see the available types and the default types for each class, examine the --help output of rwfilter or run rwsiteinfo with --fields=class,type,default-type.

--flowtypes=CLASS/TYPE[,CLASS/TYPE,@PATH ...]

The --flowtypes predicate provides an alternate way to specify class/type pairs. The --flowtypes switch allows a single rwfilter invocation to process data from multiple classes. The keyword all may be used for the CLASS and/or TYPE to select all classes and/or types. As of SiLK 3.20.0, the arguments may also include "@PATH" which causes rwfilter to open the file PATH and read the class/type pairs from it; see "Read Selection Argument Values from a File".

--sensors=SENSOR[,SENSOR,SENSOR-GROUP,@PATH ...]

The --sensors switch is used to select data from specific sensors. The parameter is a comma separated list of sensor names, sensor IDs (integers), ranges of sensor IDs, sensor group names, and/or "@PATH" directives. As described in "Read Selection Argument Values from a File", @PATH tells rwfilter to read the names of the sensors from the file PATH. Sensors and sensor groups are defined in the silk.conf(5) site configuration file, and the rwsiteinfo(1) command can be used to print a mapping of sensor names to IDs and classes (--fields=sensor,id-sensor,class:list)). When the --sensors switch is not specified, the default is to use all sensors which are valid for the specified class(es).

--start-date=YYYY/MM/DD[:HH]

--end-date=YYYY/MM/DD[:HH]

The date predicates indicate which days and hours to consider when creating the list of files. The dates may be expressed as seconds since the UNIX epoch or in YYYY/MM/DD[:HH] format, where the hour is optional. A T may be used in place of the : to separate the day and hour. Whether the YYYY/MM/DD[:HH] strings represent times in UTC or the local timezone depend on how SiLK was compiled. To determine how your version of SiLK was compiled, see the Timezone support setting in the output from rwfilter --version.

When times are expressed in YYYY/MM/DD[:HH] format:

When both --start-date and --end-date are specified to hour precision, all hours within that time range are processed.
When --start-date is specified to day precision, the hour specified in --end-date (if any) is ignored, and files for all dates between midnight on start-date and 23:59 on end-date are processed.
When --start-date is specified to hour precision and --end-date is specified to day precision, the hour of the start-date is used as the hour for the end-date.
When --end-date is not specified and --start-date is specified to day precision, files for that complete day are processed.
When --end-date is not specified and --start-date is specified to hour precision, files for that single hour are processed.

When at least one time is expressed as seconds since the UNIX epoch:

When --end-date is specified in epoch seconds, the given --start-date and --end-date are considered to be in hour precision.
When --start-date is specified in epoch seconds and --end-date is specified in YYYY/MM/DD[:HH] format, the start-date is considered to be in day precision if it divisible by 86400, and hour precision otherwise.
When --start-date is specified in epoch seconds and --end-date is not given, the start-date is considered to be in hour-precision.

When neither --start-date nor --end-date is given, rwfilter processes all files for the current day.

It is an error to specify --end-date without specifying --start-date.

It is an error to specify --start-date when rwfilter believes there is some other input specified (see "Non-Selection Input Switches").

--data-rootdir=ROOT_DIRECTORY

Tell rwfilter to use ROOT_DIRECTORY as the root of the data repository, which overrides the location given in the SILK_DATA_ROOTDIR environment variable, which in turn overrides the location that was compiled into rwfilter (/data). It is an error to specify this switch when files are specified on the command line or "Non-Selection Input Switches" are given.

--print-missing-files

This option prints to the standard error the names of the files that rwfilter's file selection switches expected to find but did not. The file names are preceded by the text 'Missing '; each file name appears on a separate line. This switch is useful for debugging, but the list of files it produces can be misleading. For example, suppose there is a decommissioned sensor that still appears in the silk.conf file; rwfilter considers these data files as missing even though their absence is expected. Use the output from this switch judiciously. It is an error to specify this switch when files are specified on the command line or "Non-Selection Input Switches" are given.

Non-Selection Input Switches

Instead of using the "Selection Switches" to read flow records from files in the data store, you can tell rwfilter to process files named on the command line or use one (and only one) of the following switches. To have rwfilter read flow records from the standard input, specify stdin or - as the name of an input file or use the (deprecated) --input-pipe switch.

--xargs
--xargs=INPUT_PATH: Read the names of the input files from INPUT_PATH or from the standard input if INPUT_PATH is not provided. The input is expected to have one filename per line. rwfilter opens each named file in turn and reads records from it as if the filenames had been listed on the command line.
--input-pipe=INPUT_PATH: Specify a source for SiLK Flow records, where INPUT_PATH is a named pipe or the string stdin or - to represent the standard input. You do not need to use this switch, you can simply specify the named pipe or the strings stdin or - on the command line. NOTE: This switch is deprecated, and it will be removed in the SiLK 4.0 release.

Output Switches

At least one of the following output switches must be provided:

--all-destination=ALL_PATH: Write every SiLK Flow record to ALL_PATH, where ALL_PATH refers to a file, a named pipe, the string stderr to refer to the standard error, or the strings stdout or - to refer to the standard output. This switch may be repeated to write all input records to multiple locations.
--fail-destination=FAIL_PATH: Write SiLK Flow records that have failed ANY of the partitioning predicates to FAIL_PATH, where FAIL_PATH refers to a non-existent file, a named pipe, the string stderr to refer to the standard error, or the strings stdout or - to refer to the standard output. This switch may be repeated to write records that fail any predicate to multiple locations.
--pass-destination=PASS_PATH: Write SiLK Flow records that have passed ALL of the partitioning predicates to PASS_PATH, where PASS_PATH refers to a non-existent file, a named pipe, the string stderr to refer to the standard error, or the strings stdout or - to refer to the standard output. This switch may be repeated to write records that pass every predicate to multiple locations.
--print-statistics
--print-statistics=STATS_PATH: Print a one line summary specifying the number of files processed, the total number of records read, the number of records that passed all partitioning predicates, and the number of records that failed. If STATS_PATH is provided, the summary is printed there; otherwise it is printed to the standard error. This switch cannot be mixed with --print-volume-statistics. When running rwfilter with multiple threads and --max-pass-records or --max-fail-records is specified, the statistics may not match the number of records written by rwfilter.
--print-volume-statistics
--print-volume-statistics=STATS_PATH: Print a four line summary of rwfilter's processing. For each of all records, records that pass all the partitioning predicates, and records that fail, print the number of flow records and the number of packets and bytes represented by those flow records. The output also includes the number of files processed. If STATS_PATH is provided, the summary is printed there; otherwise it is printed to the standard error. This switch cannot be mixed with --print-statistics. When running rwfilter with multiple threads and --max-pass-records or --max-fail-records is specified, the statistics may not match the number of records written by rwfilter.

Partitioning Switches

rwfilter supports the following partitioning switches, at least one of which must be specified (unless the only Output Switch is --all-destination). The switches are AND'ed together; i.e., to pass the filter, the record must pass the test implied by each switch. Any record that does not pass is written to the fail-destination(s), if specified.

Each partitioning switch defines a test. These tests can be grouped into several broad categories; within each category, the tests are applied in the order in which the switches appear on the command line. The categories of the partitioning tests are:

tests for IP addresses (including the IPset checks), ports, protocol, times, TCP flags, byte and packet counts, IP version, application, country codes
tests based on the --tuple-file switch
tests that use the address type or prefix map mapping files
tests that use the IP-Association plug-in
tests based on the --python-expr and --python-file switches
tests defined in C-plugins and loaded via --plugin

Partitioning Switches for IP Addresses

There are three families of switches that partition based on an IP address. Each family can partition by the source IP, the destination IP, the next hop IP, or either source or destination IP. Each family includes a --not-* variant to reverse the sense of the test.

The --*cidr-family takes as its argument an IP_OR_CIDR_LIST, which is a one or more of the following separated by commas: an IPv4 address (10.1.2.3), an IPv6 address (2001:db8::10.1.2.3), an unsigned 32-bit integer representing an IPv4 address (167838211), or any of those with a CIDR block designation (192.168.0.0/16, 2001:db8::/32, 167772160/8).

The --*set-family requires that you store the IPs in a binary IPset file and pass the name of the file to the switch. IPset files are created from SiLK Flow records with rwset(1), or from textual input with rwsetbuild(1).

The --*address-family (which includes --next-hop-id) takes as its argument a single IP address, a single CIDR block, or a single SiLK IP Wildcard. A SiLK IP Wildcard may represent multiple, disjointed IPv4 or IPv6 addresses. An IP Wildcard contains an IP in its canonical form, except each part of the IP (where part is an octet for IPv4 or a hexadectet for IPv6) may be a single value, a range, a comma separated list of values and ranges, or the letter x to signify any value for that part of the IP (that is, 0-255 for IPv4). You may not specify a CIDR suffix when using the IP Wildcard notation. The following IP_WILDCARDs all represent the same value:

 ::ffff:0:0/112
 ::ffff:0:x
 ::ffff:0:aaab-ffff,aaaa,0-aaa9
 ::ffff:0.0.0.0/112
 ::ffff:0.0.128-254,0-126,255,127.x

The next hop address often has a value of 0.0.0.0 since the default configuration of SiLK does not store the next hop address in the data repository.

With one restriction, any combination of IP partitioning switches is allowed in a single rwfilter invocation: A positive and negative version of the same switch (e.g., --sipset and --not-sipset) is not allowed. (--sipset and --not-scidr may be used together, as can --sipset and --not-dipset.)

The address-partitioning switches are:

--scidr=IP_OR_CIDR_LIST: Pass the record if its source IP address matches a value in IP_OR_CIDR_LIST, a comma separated list of IPs and/or CIDR blocks. See also --saddress and --sipset.
--dcidr=IP_OR_CIDR_LIST: Pass the record if its destination IP address matches a value in IP_OR_CIDR_LIST. See also --daddress and --dipset.
--any-cidr=IP_OR_CIDR_LIST: Pass the record if either its source or its destination IP address matches a value in IP_OR_CIDR_LIST. This switch does not consider the next hop IP address. See also --any-address and --anyset.
--nhcidr=IP_OR_CIDR_LIST: Pass the record if its next hop IP address matches a value in IP_OR_CIDR_LIST. See also --next-hop-id and --nhipset.
--not-scidr=IP_OR_CIDR_LIST: Pass the record if its source IP address does not match a value in IP_OR_CIDR_LIST, a comma separated list of IPs and/or CIDR blocks. See also --not-saddress and --not-sipset.
--not-dcidr=IP_OR_CIDR_LIST: Pass the record if its destination IP address does not match a value in IP_OR_CIDR_LIST. See also --not-daddress and --not-dipset.
--not-any-cidr=IP_OR_CIDR_LIST: Pass the record if neither its source nor its destination IP address matches a value in IP_OR_CIDR_LIST. See also --not-any-address and --not-anyset.
--not-nhcidr=IP_OR_CIDR_LIST: Pass the record if its next hop IP address does not match a value in IP_OR_CIDR_LIST. See also --not-next-hop-id and --not-nhipset.
--saddress=IP_WILDCARD: Pass the record if its source IP address is matched by the SiLK IP Wildcard IP_WILDCARD. To match on multiple IPs, use --scidr or create an IPset and use --sipset.
--daddress=IP_WILDCARD: Pass the record if its destination IP address is matched by IP_WILDCARD, a SiLK IP Wildcard. See also --dcidr and --dipset.
--any-address=IP_WILDCARD: Pass the record if either its source or its destination IP address is matched by IP_WILDCARD, a SiLK IP Wildcard. This switch does not consider the next hop IP address. See also --any-cidr and --anyset.
--next-hop-id=IP_WILDCARD: Pass the record if its next hop IP address is matched by this IP_WILDCARD, a SiLK IP Wildcard. To match on multiple IPs, use --nhcidr or create an IPset and use --nhipset.
--not-saddress=IP_WILDCARD: Pass the record if its source IP address is not matched by this IP_WILDCARD, a SiLK IP Wildcard. See also --not-scidr and --not-sipset.
--not-daddress=IP_WILDCARD: Pass the record if its destination IP address is not matched by this IP_WILDCARD. See also --not-dcidr and --not-dipset.
--not-any-address=IP_WILDCARD: Pass the record if neither its source nor its destination IP address is matched by this IP_WILDCARD. Does not consider the next hop address. See also --not-any-cidr and --not-anyset.
--not-next-hop-id=IP_WILDCARD: Pass the record if its next hop IP address is not matched by this IP_WILDCARD. See also --not-nhcidr and --not-nhipset.
--sipset=IP_SET_FILENAME: Pass the record if its source IP address is in the list of IPs contained in the binary set file IP_SET_FILENAME. See also --scidr.
--dipset=IP_SET_FILENAME: As --sipset for the destination IP address. See also --dcidr.
--anyset=IP_SET_FILENAME: Pass the record if either its source IP address or its destination IP address is in the list of IPs contained in the binary set file IP_SET_FILENAME. Does not consider the next hop IP. See also --any-cidr.
--nhipset=IP_SET_FILENAME: As --sipset for the next-hop IP address. See also --nhcidr.
--not-sipset=IP_SET_FILENAME: Pass the record if its source IP address is not in the list of IPs contained in the binary set file IP_SET_FILENAME. See also --not-scidr.
--not-dipset=IP_SET_FILENAME: As --not-sipset for the destination IP address. See also --not-dcidr.
--not-anyset=IP_SET_FILENAME: Pass the record if neither its source IP address nor its destination IP address is in the list of IPs contained in the binary set file IP_SET_FILENAME. Does not consider the next hop IP. See also --not-any-cidr.
--not-nhipset=IP_SET_FILENAME: As --not-sipset for the next hop IP address. See also --not-nhcidr.

Partitioning Switches for Remainder of Five-Tuple

The following switches partition based on the protocol and source or destination port. The parameter to each of these switches is an INTEGER_LIST, which is a comma-separated list of individual non-negative integer values and ranges of those values. For example, 1,2,3,5-10,99-103. A range may be specified without an upper limit, such as 1-, in which case the upper limit is set to the maximum value.

--sport=INTEGER_LIST: Pass the record if its source port is in this INTEGER_LIST, possible values are 0-65535.
--dport=INTEGER_LIST: Pass the record if its destination port is in this INTEGER_LIST, possible values are 0-65535
--aport=INTEGER_LIST: Pass the record if its source port and/or its destination port is in this INTEGER_LIST, possible values are 0-65535. For example, use --aport=25 to see all SMTP conversions regardless or where they originated.
--protocol=INTEGER_LIST: Pass the record if its IP Suite Protocol is in this INTEGER_LIST, possible values are 0-255.
--icmp-type=INTEGER_LIST: Pass the record if its ICMP (or ICMPv6) type is in this INTEGER_LIST; possible values 0-255. This switch also verifies that the flow's protocol is 1 (or 58 if the flow is IPv6). It is an error to specify a --protocol that does not include 1 and/or 58.
--icmp-code=INTEGER_LIST: Pass the record if its ICMP (or ICMPv6) code is in this INTEGER_LIST; possible values 0-255. This switch also verifies that the flow's protocol is 1 (or 58 if the flow is IPv6). It is an error to specify a --protocol that does not include 1 and/or 58.

Partitioning Switches for Time

These switches partition based on whether the time stamps on the flow record occur within the specified time window. The form of the argument is range of two dates, start-window and end-window, each in the form YYYY/MM/DD[:HH[:MM[:SS[.ssssss]]]], for example 2003/01/31:23:45:00.000-2003/01/31:23:59:59.999 represents the last fifteen minutes of Jan 31, 2003. (A T may be used in place of : to separate the day and hour.) The start-window and end-window must be set to at least day precision. For the start-window, unspecified hour, minute, second, and millisecond values are set to 0; for the end-window, those values are set to 23, 59, 59, and 999 respectively. Thus 2003/01/31:23-2003/01/31:23 becomes 2003/01/31:23:00:00.000-2003/01/31:23:59:59.999. If an end-window is not given, it is set to the start-window, giving a window of a single millisecond. The date strings are considered to be in the timezone specified when SiLK was compiled, which you can determine from the output of rwfilter --version. You may also specify the times as seconds since the UNIX epoch; when the end-time is in epoch seconds, an unspecified milliseconds value is set to 999 and otherwise the value is unchanged.

--active-time=TIME_WINDOW: Pass the record if the record was active at ANY time during this TIME_WINDOW. If a single time is specified, pass the record if it was active at that instant.
--stime=TIME_WINDOW: Pass the record if its starting time is in this TIME_WINDOW.
--etime=TIME_WINDOW: As --stime for the ending time.
--duration=DECIMAL_RANGE: Pass the record if its duration--that is, the record's end time minus its start time, as measured in seconds--is in this DECIMAL_RANGE. Use floating point numbers to specify millisecond values. The range should be specified as MIN-MAX; for example, 5.0-10.031. If a single value is given, the duration must match that value exactly. The upper limit may be omitted; for example, a range of 1.5- passes records whose duration is at least 1.5 seconds.

Partitioning Switches for Volume

The following switches partition based on the volume of the flow; that is, the number of bytes or packets. For additional volume-related switches, load the flowrate plug-in as described in the flowrate(3) manual page.

These switches accept a range of non-negative integers or decimal values. If the upper limit is omitted, the volume must be at least that size. If the argument is a single value, the volume must match that value exactly.

--bytes=INTEGER_RANGE: Pass the record if its byte count is in this INTEGER_RANGE.
--packets=INTEGER_RANGE: Pass the record if its packet count is in this INTEGER_RANGE.
--bytes-per-packet=DECIMAL_RANGE: Pass the record if its average bytes per packet count (bytes/packet) is in this DECIMAL_RANGE.

Partitioning Switches for TCP Flags

When a flow generator creates a flow record from TCP packets, it creates a field that is the bit-wise OR of the TCP flags from all packets that comprise that flow record. Some flow generators, such as yaf(1), can export two TCP flag fields: one contains the flags on the first packet in the flow, and the second contains the bit-wise OR of the remaining packets.

To partition records based on their TCP flags values, there is a recommended set of switches and legacy-supported switches. The switches accept the following letters to represent the named TCP flag: F=FIN; S=SYN; R=RST; P=PSH; A=ACK; U=URG; E=ECE; C=CWR. As of SiLK 3.20.0, the symbol - is accepted and represents all TCP flags (FSRPAUEC).

The recommended set of switches take a comma separated list of pairs of TCP flags, where the pair is separated by a slash (/). The value to the left of the slash is the HIGH_SET and it must be a subset of the value to the right of the slash, which is the MASK_SET. For a record to pass the filter, the flags in the HIGH_SET must be on and the remaining flags in MASK_SET must be off. Flags not in MASK_SET may have any value. If a list of pairs is given, the record passes if any pair in the list matches. For example, --flags-all=S/S,A/A passes flows that have either the SYN or the ACK flag set, --flags-all=S/SA passes flow records where SYN is high and ACK is low, and --flags-all=/F passes flows where FIN is off. This list of flag pairs is called a HIGH_MASK_FLAGS_LIST.

The recommended switches for TCP flag partitioning are:

--flags-all=HIGH_MASK_FLAGS_LIST: Pass the record if any of the HIGH_SET/MASK_SET pairs is true when looking at the bit-wise OR of the TCP flags across all packets in the flow.
--flags-initial=HIGH_MASK_FLAGS_LIST: As --flags-all, except this switch considers only the initial packet in the flow, for flow generators that can generate that field.
--flags-session=HIGH_MASK_FLAGS_LIST: As --flags-all, except this switch considers the bit-wise OR of the TCP flags across the second through the final packet in the flow; that is, ignoring the flags on the first packet.

The TCP-flag partitioning switches supported for legacy reasons are:

--tcp-flags=TCP_FLAGS: Pass the record if, for any one of its packets, any of the specified TCP_FLAGS was set (high), where TCP_FLAGS contains the letters F,S,R,P,A,U,E,C. For example, --tcp-flags=ASF passes records where ACK is set, or SYN is set, or FIN is set (this is equivalent to --flags-all=A/A,S/S,F/F).
--ack-flag={0|1}: Pass the record when either the argument is 1 and the ACK flag is set (high) or the argument is 0 and the ACK flag is unset (low).
--cwr-flag={0|1}: Pass the record when either the argument is 1 and the CWR flag is set (high) or the argument is 0 and the CWR flag is unset (low).
--ece-flag={0|1}: Pass the record when either the argument is 1 and the ECE flag is set (high) or the argument is 0 and the ECE flag is unset (low).
--fin-flag={0|1}: Pass the record when either the argument is 1 and the FIN flag is set (high) or the argument is 0 and the FIN flag is unset (low).
--psh-flag={0|1}: Pass the record when either the argument is 1 and the PSH flag is set (high) or the argument is 0 and the PSH flag is unset (low).
--rst-flag={0|1}: Pass the record when either the argument is 1 and the RST flag is set (high) or the argument is 0 and the RST flag is unset (low).
--syn-flag={0|1}: Pass the record when either the argument is 1 and the SYN flag is set (high) or the argument is 0 and the SYN flag is unset (low).
--urg-flag={0|1}: Pass the record when either the argument is 1 and the URG flag is set (high) or the argument is 0 and the URG flag is unset (low).

Partitioning Switches for Other Flow Characteristics

Other than the --ip-version switch, the fields queried by the following switches may always be zero. The default configuration of SiLK does not store the fields that contain the SNMP values. The other fields are not present in NetFlow v5, and require use of properly-configured enhanced collection software, such as yaf(1), http://tools.netsa.cert.org/yaf/.

--ip-version={4|6|4,6}

Passes the record if its IP Version is in the specified list. This switch determines how IPv4 and IPv6 flow records are handled when SiLK has been compiled with IPv6 support. When the argument to this switch is 4, rwfilter writes records marked as IPv6 to the fail-destination, regardless of the IP addresses it contains. When the argument to this switch is 6, rwfilter writes records marked as IPv4 to the fail-destination. When SiLK has not been compiled with IPv6 support, the only legal value for this switch is 4, and any IPv6 flows in the input ignored (that is, they are not written to either the pass-destination nor the fail-destination).

--application=INTEGER_LIST

Some flow generation software can inspect the contents of the packets that comprise a flow and use traffic signatures to label the content of the flow. SiLK calls this label the application; yaf refers to it as the appLabel (see the applabel(1) manual page in the yaf distribution). The application value is the port number that is traditionally used for that type of traffic (see the /etc/services file on most UNIX systems). For example, traffic that the flow generator recognizes as FTP has a value of 21, even if that traffic is being routed through the standard HTTP/web port (80). The flow generator uses a value for 0 if the application cannot be determined. The --application switch passes the flow if the flow's application value is in the specified INTEGER_LIST, which is a comma separated list of integers from 0 to 65535 inclusive and ranges of those integers. The list of valid appLabels is determined by your site's yaf installation.

--attributes=ATTRIBUTES_LIST

The attributes field in SiLK Flow records describes characteristics about how the flow record was generated or about the packets that comprise the flow record. The ATTRIBUTES_LIST argument is similar to the HIGH_MASK_FLAGS_LIST argument to the --flags-all switch. ATTRIBUTES_LIST is a comma separated list of up to 8 HIGH_ATTRIBUTES/MASK_ATTRIBUTES pairs, where HIGH_ATTRIBUTES and MASK_ATTRIBUTES are strings of the characters S,T,C,F, and HIGH_ATTRIBUTES is a subset of MASK_ATTRIBUTES. As of SiLK 3.20.0, the symbol - is accepted and represents all attributes (STCF). rwfilter passes the record if, for any pair of attributes in the list, the attributes listed in HIGH_ATTRIBUTES are set and the remaining attributes in MASK_ATTRIBUTES are not-set. The valid attributes are:

S: All the packets in this flow record are exactly the same size.
T: The flow generator prematurely created a record for a long-lived session due to the connection's lifetime reaching the active timeout of the flow generator. (Also, when yaf is run with the --silk switch, it prematurely creates a flow and marks it with T if the byte count of the flow cannot be stored in a 32-bit value.)
C: The flow generator created this flow as a continuation of long-running connection, where the previous flow for this connection met a timeout.
F: The flow generator saw additional packets in this flow following a packet with the FIN flag set (excluding ACK packets).

For a long-lived connection spanning several flow records, the first flow record is marked with a T indicating that it hit the active timeout. The second through next-to-last records are marked with CT indicating that the flow is a continuation of a connection that timed out and that this flow also timed out. The final flow is marked with a C, indicating that it was created as a continuation of an active flow.

--input-index=INTEGER_LIST

Pass the record if its in field is in this INTEGER_LIST, which is a comma separated list of integers from 0 to 65535, inclusive, and ranges of those integers. When present, the in field normally contains the incoming SNMP interface, but it may contain the vlanId if the packing tools were configured to capture it (see sensor.conf(5)).

--output-index=INTEGER_LIST

Pass the record if its out field is in this INTEGER_LIST. When present, the out field normally contains the outgoing SNMP interface, but it may contain the postVlanId if the packing tools were configured to capture it.

--any-index=INTEGER_LIST

Pass the record if its in field or if its out field is in this INTEGER_LIST.

Selection Switches Acting as Partitioning Switches

The following four switches are normally file selection switches that select which files rwfilter reads within the data repository. However, when rwfilter gets input without querying the data repository (that is, from files listed on the command line, from files specified by --xargs, or from the --input-pipe), these switches become partitioning switches and determine whether a record is written to the pass-destination or fail-destination.

--class={CLASS | @PATH}: Pass the record if its class is CLASS or the class named in PATH and its type is listed in the --type switch, or its type is in the default type list for CLASS when --type is not specified. Examine the output of rwfilter --help to see the list of available classes, available types, and their default values, or use rwsiteinfo --fields=class,type,mark-defaults.
--flowtypes=CLASS/TYPE[,CLASS/TYPE,@PATH ...]: Pass the record its if class/type value is one of those listed or read from the file PATH. The keyword all may be used for the CLASS and/or TYPE to select all classes and/or types. This switch cannot be used when either --class or --type is used. Use rwfilter --help or rwsiteinfo --fields=class,type to see the list of available classes and types.
--sensors=SENSOR[,SENSOR,SENSOR-GROUP,@PATH ...]: Pass the record if its sensor is one of those listed. The parameter is a comma separated list of sensor names, sensor IDs (integers), ranges of sensor IDs, sensor group names, and "@PATH" directives. Use the rwsiteinfo(1) command to see the list of sensors. Support for sensor group names was added in SiLK 3.21.0.
--type={all | TYPE[,TYPE,@PATH ...]}: Pass the record if its type is one of those listed or read from the file PATH and its class is specified by --class, or its class is the default class when the --class switch is not specified. Use rwfilter --help to see the list of available classes, available types, and their defaults, or use rwsiteinfo --fields=class,type,mark-defaults.

Partitioning Switches that use Additional Mapping Files

Additional partitioning switches are available that allow one to partition flow records depending on a label, where the label is computed from an IP address or port on the record and an additional mapping file.

--pmap-file=PATH

--pmap-file=MAPNAME:PATH

Load the prefix map file located at PATH and create partitioning switches named --pmap-src-map-name, --pmap-dst-map-name, and --pmap-any-map-name where map-name is either the MAPNAME part of the argument or the map-name specified when the file was created (see rwpmapbuild(1)). If no map-name is available, rwfilter creates switch names as described below (--pmap-saddress, --pmap-sport-proto, etc). Specify PATH as - or stdin to read from the standard input. The switch may be repeated to load multiple prefix map files; each file must have a unique map-name. The --pmap-file switch(es) must precede all other --pmap-* switches. For more information, see pmapfilter(3).

--pmap-src-map-name=LABELS

If the prefix map associated with map-name is an IP prefix map, this matches records with a source IPv4 address that maps to a label contained in the list of labels in LABELS. If the prefix map associated with map-name is a proto-port prefix map, this matches records with a protocol and source port combination that maps to a label contained in the list of labels in LABELS.

--pmap-dst-map-name=LABELS

Similar to --pmap-src-map-name, but uses the destination IP or the protocol and destination port.

--pmap-any-map-name=LABELS

If the prefix map associated with map-name is an IP prefix map, this matches records with a source IP address or a destination IP address that maps to a label contained in the list of labels in LABELS. If the prefix map associated with map-name is a port/protocol prefix map, this matches records with a protocol and source port or destination port combination that maps to a label contained in the list of labels in LABELS.

--pmap-saddress=LABELS

--pmap-daddress=LABELS

--pmap-any-address=LABELS

These are deprecated switches created by pmapfilter that correspond to --pmap-src-map-name, --pmap-dst-map-name, and --pmap-any-map-name, respectively. These switches are available when an IP prefix map is used that is not associated with a map-name.

--pmap-sport-proto=LABELS

--pmap-dport-proto=LABELS

--pmap-any-port-proto=LABELS

These are deprecated switches created by pmapfilter that correspond to --pmap-src-map-name, --pmap-dst-map-name, and --pmap-any-map-name, respectively. These switches are available when a proto-port prefix map is used that is not associated with a map-name.

--scc=COUNTRY_CODE_LIST

--dcc=COUNTRY_CODE_LIST

--any-cc=COUNTRY_CODE_LIST

Pass the record if one its IP addresses maps to a country code that is specified in COUNTRY_CODE_LIST. For --scc, the source IP must match. For --dcc, the destination IP must match. For --any-cc, either the source or the destination must match. COUNTRY_CODE_LIST is a comma separated list of lowercase two-letter country codes---defined by ISO 3166-1 (see for example https://www.iso.org/iso-3166-country-codes.html or https://en.wikipedia.org/wiki/ISO_3166-1_alpha-2)---as well as the following special codes:

--: N/A (e.g. private and experimental reserved addresses)
a1: anonymous proxy
a2: satellite provider
o1: other

For example: cx,uk,kr,jp,--. To use this switch, the country code mapping file must be available in the default location, or in the location specified by the SILK_COUNTRY_CODES environment variable. See ccfilter(3) for details.

--stype={0|1|2|3}

--dtype={0|1|2|3}

Pass a flow record depending on whether the IP address is internal, external, or non-routable. These switches use the mapping file specified by the SILK_ADDRESS_TYPES environment variable, or the address_types.pmap mapping file, as described in addrtype(3). When the parameter is 0, pass the record if its source (--stype) IP address or destination (--dtype) IP address is non-routable. When 1, pass if internal. When 2, pass if external (i.e., routable but not internal). When 3, pass if not internal (non-routable or external).

Partitioning Switches across Multiple Fields

The --tuple-* family of switches allows the user to partition flow records based on multiple values of the five-tuple.

--tuple-file=TUPLE_FILENAME

This switch provides support for partitioning by arbitrary subsets of the basic five-tuple:

 {source-ip,destination-ip,source-port,destination-ip-port,protocol}

A SiLK Flow record passes the test when the record's fields match one of the tuples; if the SiLK record does not match any tuple, the record fails. The tuples are read from the text file TUPLE_FILENAME which must contain lines of delimited fields. The default delimiter is |, but may be specified with the --tuple-delimiter switch. Each field contains one member of the tuple; the fields may appear in any order. The fields may represent any subset of the five-tuple, but each line in the file must define the same subset. A field that is present but has no value generates an error. If you want the field to match any value, it is best that you not include that field in your input.

In addition to the tuple-lines, TUPLE_FILENAME may contain blank lines and comments (which begin with # and continue to the end of the line). The first line of TUPLE_FILENAME may contain a title labeling the fields in the file. This title line is ignored when the --tuple-fields switch is given.

The IP fields may contain an IPv4 address, an integer, or a IP in CIDR block notation. Comma-separated lists (80,443) and ranges (0-1023,8080) are supported for the ports and protocol fields. NOTE: Currently the code is not clever in its support for CIDR notation and ranges in that each occurrence is fully expanded. When this occurs, the memory required to hold the search tree quickly grows.

--tuple-fields=FIELDS

FIELDS contains the list of fields (columns) to parse from the TUPLE_FILENAME in the order in which they appear in the file. When this switch is not provided, rwfilter treats the first line in TUPLE_FILENAME as a title line and attempts to determine the fields (a la rwtuc(1)); rwfilter exits if it cannot determine the fields.

FIELDS is a comma separated list of field-names, field-integers, and ranges of field-integers; a range is specified by separating the start and end of the range with a hyphen (-). Names can be abbreviated to their shortest unique prefix. The field names and their descriptions are:

sIP,sip,1: source IP address
dIP,dip,2: destination IP address
sPort,sport,3: source port
dPort,dport,4: destination port
protocol,5: IP protocol

--tuple-direction=DIRECTION

Allows you to change the comparison between the tuple and the SiLK Flow record. This switch allows one to look for traffic in the reverse direction (or both directions) without having to write all of the rules twice. The available directions are:

forward: The tuple's fields are compared against the corresponding fields on the flow; that is, sIP is compared with sIP, dIP with dIP, sPort with sPort, dPort with dPort, and protocol with protocol. This is the default.
reverse: The tuple's fields are compared against the opposite fields on the flow; that is, sIP is compared with dIP, dIP with sIP, sPort with dPort, dPort with sPort, and protocol with protocol.
both: Both of the above comparisons are performed.

--tuple-delimiter=CHAR

Specifies the character separating the input fields. When the switch is not provided, the default of | is used.

Partitioning Switches that use the PySiLK Plug-in

The SiLK Python plug-in provides support for filtering by expressions or complex functions written in the Python programming language. See the silkpython(3) and pysilk(3) manual pages for information and examples for how to use Python to manipulate SiLK data structures. When multiple Partitioning Switches are given, the Python plug-in is the next-to-last to be invoked. Only the code specified by the --plugin switch is called after the Python code.

--python-file=FILENAME

Pass the record if the result of the processing the flow with the function named rwfilter() in FILENAME is true. The function should take a single silk.RWRec object as an argument. See silkpython(3) for details.

--python-expr=PYTHON_EXPRESSION

Pass the record if the result of the processing the flow with the specified PYTHON_EXPRESSION is true. The expression is evaluated as if it appeared in the following context:

 from silk import *
 def rwfilter(rec):
     return (PYTHON_EXPRESSION)

Partitioning Switches that use the IP-Association Plug-In

The IPA plug-in, ipafilter.so, provides switches that can partition flows using data in an IP Association database. For this plug-in to be available, SiLK must be compiled with IPA support and IPA must be configured. See ipafilter(3) and http://tools.netsa.cert.org/ipa/ for additional information.

--ipa-src-expr=IPA_EXPR: Use IPA_EXPR to partition flows based on the source IP of the flow matching the IPA_EXPR expression.
--ipa-dst-expr=IPA_EXPR: Use IPA_EXPR to partition flows based on the destination IP of the flow matching the IPA_EXPR expression.
--ipa-any-expr=IPA_EXPR: Use IPA_EXPR to partition flows based on either the source or destination IP of the flow matching the IPA_EXPR expression.

Miscellaneous Switches

--compression-method=COMP_METHOD

Specify the compression library to use when writing output files. If this switch is not given, the value in the SILK_COMPRESSION_METHOD environment variable is used if the value names an available compression method. When no compression method is specified, output to the standard output or to named pipes is not compressed, and output to files is compressed using the default chosen when SiLK was compiled. The valid values for COMP_METHOD are determined by which external libraries were found when SiLK was compiled. To see the available compression methods and the default method, use the --help or --version switch. SiLK can support the following COMP_METHOD values when the required libraries are available.

none: Do not compress the output using an external library.
zlib: Use the zlib(3) library for compressing the output, and always compress the output regardless of the destination. Using zlib produces the smallest output files at the cost of speed.
lzo1x: Use the lzo1x algorithm from the LZO real time compression library for compression, and always compress the output regardless of the destination. This compression provides good compression with less memory and CPU overhead.
snappy: Use the snappy library for compression, and always compress the output regardless of the destination. This compression provides good compression with less memory and CPU overhead. Since SiLK 3.13.0.
best: Use lzo1x if available, otherwise use snappy if available, otherwise use zlib if available. Only compress the output when writing to a file.

--dry-run

Perform a sanity check on the input arguments to check that the arguments are acceptable. In addition, prints to the standard output the names of the files that would be accessed (and the names of missing files if --print-missing is specified). rwfglob(1) can also be used to generate the lists of files that rwfilter would access.

--help

Print the available options and exit. Options that add fields (for example, options that load plug-ins, prefix maps, or PySiLK extensions) can be specified before the --help switch so that the new options appear in the output. The available classes and types are included in output; you may specify a different root directory or site configuration file before --help to see the classes and types available for that site.

--max-fail-records=N

Write N records to each --fail-destination. rwfilter stops reading input once it has written these N records unless --pass-destination or --all-destination switch(es) are also specified.

--max-pass-records=N

Write N records to each --pass-destination. rwfilter stops reading input once it has written these N records unless --fail-destination or --all-destination switch(es) are also specified.

--note-add=TEXT

Add the specified TEXT to the header of the output file as an annotation. This switch may be repeated to add multiple annotations to a file. To view the annotations, use the rwfileinfo(1) tool.

--note-file-add=FILENAME

Open FILENAME and add the contents of that file to the header of the output file as an annotation. This switch may be repeated to add multiple annotations. Currently the application makes no effort to ensure that FILENAME contains text; be careful that you do not attempt to add a SiLK data file as an annotation.

--plugin=PLUGIN

Augment the partitioning switches by using run-time loading of the plug-in (shared object) whose path is PLUGIN. The switch may be repeated to load multiple plug-ins. The creation of plug-ins is described in the silk-plugin(3) manual page. When multiple partitioning switches are given, the code specified by the --plugin switch(es) is last to be invoked. When PLUGIN does not contain a slash (/), rwfilter attempts to find a file named PLUGIN in the directories listed in the "FILES" section. If rwfilter finds the file, it uses that path. If PLUGIN contains a slash or if rwfilter does not find the file, rwfilter relies on your operating system's dlopen(3) call to find the file. When the SILK_PLUGIN_DEBUG environment variable is non-empty, rwfilter prints status messages to the standard error as it attempts to find and open each of its plug-ins.

--print-filenames

Print the names of input files as they are read. This can be useful feedback for a long-running rwfilter process.

--site-config-file=FILENAME

Read the SiLK site configuration from the named file FILENAME. When this switch is not provided, rwfilter searches for the site configuration file in the locations specified in the "FILES" section.

--threads=N

Invoke rwfilter with N threads reading the input files. When this switch is not provided, the value in the SILK_RWFILTER_THREADS environment variable is used. If that variable is not set, rwfilter runs with a single thread. Using multiple threads, performance of rwfilter is greatly improved for queries that look at many files but return few records. Preliminary testing has found that performance peaks around four threads per CPU, but performance varies depending on the type of query and the number of records returned.

--version

Print the version number and information about how SiLK was configured, then exit the application.

EXAMPLES

In the following examples, the dollar sign ($) represents the shell prompt. The text after the dollar sign represents the command line. Lines have been wrapped for improved readability, and the back slash (\) is used to indicate a wrapped line.

The most basic filtering involves looking at specific traffic over a specific time. For example:

 $ rwfilter --start-date=2003/02/19:00 --end-date=2003/02/19:23     \
        --proto=6 --pass-destination=tcp-in.rw

creates a file, tcp-in.rw containing all incoming TCP traffic on February 19, 2003. The --start-date and --end-date switches select which files to examine. The --proto switch partitions the flow records into a pass stream (records whose protocol is 6---that is, TCP) and a fail stream (all other records). The --pass-destination switch (often shortened to --pass) tells rwfilter to write the records that pass the --proto test to the file tcp-in.rw.

The tcp-in.rw file contains SiLK Flow data in a binary format. To examine the contents, use the command rwcut(1). This query only selects incoming traffic because the silk.conf(5) configuration file at most sites tells rwfilter to look at incoming traffic unless an explicit --type switch is given.

The following query gets all TCP traffic (for the default class) for February 19, 2003.

 $ rwfilter --type=all --start-date=2003/02/19  \
        --proto=6 --pass-destination=alltcp.rw

Note the addition of --type=all. This query also relies on the default behavior of --start-date to consider a full day's worth of data when no hour is specified.

The above query gets all traffic for the default class. If your silk.conf file has a single class, that query captures all of it. For silk.conf files that specify multiple classes, the following gets all TCP traffic for February 19, 2003:

 $ rwfilter --flowtypes=all/all --start-date=2003/02/19     \
        --proto=6 --pass-destination=alltcp.rw

To get all non-TCP traffic, there are two approaches. rwfilter does not supply a way to choose a negated set of protocols, but you can choose all protocols other than TCP:

 $ rwfilter --start-date=2003/02/19:00 --end-date=2003/02/19:23     \
        --proto=0-5,7-255 --pass-destination=non-tcp.rw

The other approach is to use the --fail-destination switch (often shortened to --fail) that contains the records that failed one or more of the partitioning test(s):

 $ rwfilter --start-date=2003/02/19:00 --end-date=2003/02/19:23     \
        --proto=6 --fail-destination=non-tcp.rw

To print information about the number of flow records that pass a filter, use --print-volume-statistics. This can be combined with other output switches.

 $ rwfilter --start-date=2003/02/19:00 --end-date=2003/02/19:23     \
        --proto=6 --print-volume-stat --pass-destination=tcp-in.rw
      |        Recs|     Packets|         Bytes|  Files|
 Total|      515359|     2722887|    1343819719|    180|
  Pass|      512071|     2706571|    1342851708|       |
  Fail|        3288|       16316|        968011|       |

If you want to see the number of records in a file produced by rwfilter, or to remind yourself how a file was created, use rwfileinfo(1):

 $ rwfileinfo tcp-in.rw
 tcp-in.rw:
   format(id)          FT_RWGENERIC(0x16)
   version             16
   byte-order          littleEndian
   compression(id)     lzo1x(2)
   header-length       208
   record-length       52
   record-version      5
   silk-version        2.4.0
   count-records       512071
   file-size           8576160
   command-lines
       1  rwfilter --start-date=2003/02/19:00 --end-date=2003/02/19:23 \
            --proto=6 --print-volume-stat --pass-destination=tcp-in.rw

Once a file is written, rwfilter can process the file again. Traffic on port 25 is most likely email (SMTP) traffic. To split the email traffic from the other traffic, use:

 $ rwfilter --aport=25 --pass=mail.rw --fail=not-mail.rw tcp-in.rw

This command puts traffic where the source or destination port was 25 into the file mail.rw, and all other traffic into the file not-mail.rw. The --fail-destination is an effective way to reverse the sense of a test. For example, to remove traffic on port 80 from the not-mail.rw file, run the command:

 $ rwfilter --aport=80 --fail=not-mail-web.rw not-mail.rw

To verify that the not-mail-web.rw file does not contain any traffic on ports 25 or 80, you can use the --print-statistics switch and see that 0 records pass:

 $ rwfilter --aport=25,80 --print-stat not-mail-web.rw
 Files     1.  Read    54641.  Pass        0. Fail     54641.

The file maintains a history of the commands that created it:

 $ rwfileinfo not-mail-web.rw
 not-mail-web.rw:
   format(id)          FT_RWGENERIC(0x16)
   version             16
   byte-order          littleEndian
   compression(id)     lzo1x(2)
   header-length       364
   record-length       52
   record-version      5
   silk-version        2.4.0
   count-records       54641
   file-size           762875
   command-lines
       1  rwfilter --start-date=2003/02/19:00 --end-date=2003/02/19:23 \
            --proto=6 --print-volume-stat --pass-destination=tcp-in.rw
       2  rwfilter --aport=25 --pass=mail.rw --fail=not-mail.rw        \
            tcp-in.rw
       3  rwfilter --aport=80 --fail=not-mail-web.rw not-mail.rw

The following finds all outgoing traffic from February 19, 2003, going to an external email server. Traffic going to a server contacts that server on its well-known port, and the flow record's destination port should hold that well-known port:

 $ rwfilter --type=out --start-date=2003/02/19 --print-volume-stat  \
        --dport=25 --proto=6

To limit the result to completed connections, select flow records that contain at least three packets, use the --packets switch with an open-ended range:

 $ rwfilter --type=out --start-date=2003/02/19 --print-volume-stat  \
        --dport=25 --proto=6 --packets=3-

To limit the search to a particular internal CIDR block, 10.1.2.0/24, there are three different IP-partitioning switches you can use. The final approach uses rwsetbuild(1) to create an IPset file from textual input.

 $ rwfilter --type=out --start-date=2003/02/19 --print-volume-stat  \
        --dport=25 --proto=6 --packets=3- --scidr=10.1.2.0/24

 $ rwfilter --type=out --start-date=2003/02/19 --print-volume-stat  \
        --dport=25 --proto=6 --packets=3- --saddress=10.1.2.x

 $ echo "10.1.2.0/24" | rwsetbuild > my-set.set
 $ rwfilter --type=out --start-date=2003/02/19 --print-volume-stat  \
        --dport=25 --proto=6 --packets=3- --sipset=my-set.set

rwfilter does not have to output its records to a file; instead, the output from rwfilter can be piped into a another SiLK tool. You must still use the --pass-destination switch (or --fail-destination or --all-destination switch), but by providing the argument of stdout or - to the switch you tell rwfilter to write its output to the standard output.

For example, to get the IPs of the external email servers that the monitored network contacted, pipe the rwfilter output into rwset(1), and tell rwset to store the destination addresses:

 $ rwfilter --type=out --start-date=2003/02/19 --dport=25           \
        --proto=6 --packets=3- --scidr=10.1.2.0/24 --pass=stdout    \
   | rwset --dip-file=external-mail-servers.set

rwfilter can also pipe its output as input to another rwfilter command, which allows them to be chained together. rwfilter does not read from the standard input by default; you must explicitly give stdin or - as the stream to read:

 $ rwfilter --type=out,outweb --start-date=2003/02/19               \
        --scidr=10.1.2.0/24 --pass=stdout                           \
   | rwfilter --proto=17 --pass=udp.rw --fail=stdout stdin          \
   | rwfilter --proto=6 --pass=stdout --fail=non-tcp-udp.rw stdin   \
   | rwfilter --aport=25 --pass=mail.rw --fail=stdout stdin         \
   | rwfilter --aport=80,443 --pass=web.rw                          \
        --fail=tcp-non-web-mail.rw stdin

This chain of commands looks at outgoing traffic on February 19, 2003, originating from the internal net-block 10.1.2.0/24, creates the following files:

udp.rw: Outgoing UDP traffic
non-tcp-udp.rw: Outgoing traffic that is neither TCP nor UDP
mail.rw: Outgoing TCP traffic on port 25, most of which is probably email (SMTP). Since the query looks at outgoing traffic and the --aport switch was used, this file represents email going from the internal 10.1.2.0/24 to external mail servers, and the responses from any internal mail servers that exist in the 10.1.2.0/24 net-block to external clients.
web.rw: Outgoing TCP traffic on ports 80 and 443, most of which is probably web traffic (HTTP,HTTPS). As with the mail.rw file, this file represents queries to external web servers and responses from internal web servers.
tcp-non-web-mail.rw: Outgoing TCP traffic other than that on ports 25, 80, and 443

Expert users can create even more complicated chains of rwfilter commands using named pipes.

ENVIRONMENT

SILK_RWFILTER_THREADS

The number of threads to use while reading input files or files selected from the data store.

PYTHONPATH

This environment variable is used by Python to locate modules. When --python-file or --python-expr is specified, rwfilter must load the Python files that comprise the PySiLK module, such as silk/__init__.py. If this silk/ directory is located outside Python's normal search path (for example, in the SiLK installation tree), it may be necessary to set or modify the PYTHONPATH environment variable to include the parent directory of silk/ so that Python can find the PySiLK module.

SILK_PYTHON_TRACEBACK

When set, Python plug-ins output traceback information on Python errors to the standard error.

SILK_COUNTRY_CODES

This environment variable allows the user to specify the country code mapping file that the --scc and --dcc switches use. The value may be a complete path or a file relative to the SILK_PATH. See the "FILES" section for standard locations of this file.

SILK_ADDRESS_TYPES

This environment variable allows the user to specify the address type mapping file that the --stype and --dtype switches use. The value may be a complete path or a file relative to the SILK_PATH. See the "FILES" section for standard locations of this file.

SILK_CLOBBER

The SiLK tools normally refuse to overwrite existing files. Setting SILK_CLOBBER to a non-empty value removes this restriction.

SILK_COMPRESSION_METHOD

This environment variable is used as the value for --compression-method when that switch is not provided. Since SiLK 3.13.0.

SILK_CONFIG_FILE

This environment variable is used as the value for the --site-config-file when that switch is not provided.

SILK_DATA_ROOTDIR

This environment variable specifies the root directory of data repository. This value overrides the compiled-in value, and rwfilter uses it unless the --data-rootdir switch is specified. In addition, rwfilter may use this value when searching for the SiLK site configuration files. See the "FILES" section for details.

SILK_PATH

This environment variable gives the root of the install tree. When searching for configuration files and plug-ins, rwfilter may use this environment variable. See the "FILES" section for details.

TZ

When a SiLK installation is built to use the local timezone (to determine if this is the case, check the Timezone support value in the output from rwfilter --version), the value of the TZ environment variable determines the timezone in which rwfilter parses timestamps. If the TZ environment variable is not set, the default timezone is used. Setting TZ to 0 or the empty string causes timestamps to be parsed as UTC. The value of the TZ environment variable is ignored when the SiLK installation uses utc. For system information on the TZ variable, see tzset(3) or environ(7).

SILK_PLUGIN_DEBUG

When set to 1, rwfilter prints status messages to the standard error as it attempts to find and open each of its plug-ins.

SILK_LOGSTATS

When set to a non-empty value, rwfilter treats the value as the path to an external program to execute with information about this rwfilter invocation. If the value in SILK_LOGSTATS does not contain a slash or if it references a file that does not exist, is not a regular file, or is not executable, the SILK_LOGSTATS value is silently ignored. The arguments to the external program are:

The application name, i.e., rwfilter. Note that rwfilter is always used as this argument, regardless of the name of the executable.
The version number of this command line, currently v0001.
The start time of this invocation, as seconds since the UNIX epoch.
The end time of this invocation, as seconds since the UNIX epoch.
The number of data files opened for reading.
The number of records read.
The number of records written.
A variable number of arguments that are the complete command line used to invoke rwfilter, including the name of the executable.

SILK_LOGSTATS_RWFILTER

If set, this environment variable overrides the value specified in SILK_LOGSTATS.

SILK_LOGSTATS_DEBUG

If the environment variable is set to a non-empty value, rwfilter prints messages to the standard error about the SILK_LOGSTATS value being used and either the reason why the value cannot be used or the arguments to the external program being executed.

FILES

${SILK_ADDRESS_TYPES}
${SILK_PATH}/share/silk/address_types.pmap
${SILK_PATH}/share/address_types.pmap
/usr/share/silk/address_types.pmap
/usr/share/address_types.pmap: Possible locations for the address types mapping file required by the --stype and --dtype switches.
${SILK_CONFIG_FILE}
ROOT_DIRECTORY/silk.conf
${SILK_PATH}/share/silk/silk.conf
${SILK_PATH}/share/silk.conf
/usr/share/silk/silk.conf
/usr/share/silk.conf: Possible locations for the SiLK site configuration file which are checked when the --site-config-file switch is not provided, where ROOT_DIRECTORY/ is the directory rwfilter is using as the root of the data repository.
${SILK_COUNTRY_CODES}
${SILK_PATH}/share/silk/country_codes.pmap
${SILK_PATH}/share/country_codes.pmap
/usr/share/silk/country_codes.pmap
/usr/share/country_codes.pmap: Possible locations for the country code mapping file required by the --scc and --dcc switches.
${SILK_DATA_ROOTDIR}/
/data/: Locations for the root directory of the data repository when the --data-rootdir switch is not specified.
${SILK_PATH}/lib64/silk/
${SILK_PATH}/lib64/
${SILK_PATH}/lib/silk/
${SILK_PATH}/lib/
/usr/lib64/silk/
/usr/lib64/
/usr/lib/silk/
/usr/lib/: Directories that rwfilter checks when attempting to load a plug-in.

NOTES

The ability to use @PATH in --class, --type, --flowtypes, and --sensors was added in SiLK 3.20.0.

As of SiLK 3.20.0, --types is an alias for --type.

The --sensors switch also accepts the names of groups defined in the silk.conf(5) file as of SiLK 3.21.0.

rwfilter is the most commonly used application in the suite. It provides access to the data files and performs all the basic queries.

rwfilter supports a variety of I/O options - in addition to reading from the data store, rwfilter results can be chained together with named pipes to output results to multiple files simultaneously. An introduction to named pipes is outside the scope of this document, however.

Two often underused options are --dry-run and --print-statistics. --dry-run performs a sanity check on the arguments and can be used, especially for complicated arguments, to check that the arguments are acceptable. --print-statistics used without --pass-destination or --fail-destination simply prints aggregate statistics to the standard error on a single line, and it can be used to do a quick pass through the data to get aggregate counts before going in deeper into the phenomenon being investigated.

--print-filename can be used as a progress meter; during long jobs, it shows which file is currently being read by rwfilter. --print-filename does not provide meaningful feedback with piped input.

Filters are applied in the order given on the command line. It is best to apply the biggest filters first.

The rwfilter command line is written into the header of the output file(s). You may use the rwfileinfo(1) command to see this information.