Drib is a tool that lets you manage IPv4 and IPv6 address ranges obtained locally or from a remote source. IP ranges can be classified, merged, prioritized and finally rendered according to provided templates.
Possible uses include:
- Firewall rule generation from public IP range lists;
- Policy script generation for Gatekeeper.
Debian packages will be provided with the first Drib release
$ wget https://
$ sudo dpkg -i ...Drib will be added to crates.io on its first release
Use Rust's package manager, cargo, to install Drib from source:
$ cargo install --git https://github.com/andrenth/dribDrib can be run in three modes: aggregate, bootstrap and diff. In aggregate mode, Drib will fetch IP ranges from all configured sources and save them in an aggregate file. Bootstrap will render IP ranges, using the provided template. Diff mode will compare IP ranges with the ones fetched in the previous execution, rendering the resulting difference (i.e. inserted and removed ranges). In both bootstrap and diff modes, IP ranges can either be fetched and agreggated from the sources defined in the configuration file, or obtained from a file generated by a previous execution in aggregate mode.
Run in aggregate mode:
$ drib aggregateRun in bootstrap mode:
# Fetch, aggregate and render a bootstrap template:
$ drib bootstrap
# Render a bootstrap template based on a previously generated aggregate.
$ drib bootstrap --no-downloadRun in diff mode:
# Fetch, aggregate and render diff templates:
$ drib diff
# Render diff templates based on the last two generated aggregates.
$ drib diff --no-downloadThe commands above will read the default configuration file, /etc/drib/drib.yaml.
To specify an alternative configuration file, use the -c or --config command line flag:
$ drib -c /path/to/config/file.yaml bootstrap
$ drib -c /path/to/config/file.yaml diffFor further details, run drib help.
Drib uses YAML for its configuration file. The following directives are supported:
The directory where Drib stores downloaded ranges and calculated range aggregates (defaults to /var/lib/drib).
Drib's log level.
Valid values are error, warn, info, debug or trace (defaults to info).
Sets the number of worker threads in Drib's async runtime's (Tokio) thread pool. Defaults to the number of cores available to the system.
This section defines input and output settings for bootstrap mode. Two subsettings are expected:
inputrefers to a template file (see theTemplatessection below for details) used to render the bootstrap ranges.outputspecifies the path of the rendered file.
The output path is itself a template, so a couple of variables can be used to split the bootstrap output according to protocol (i.e. IPv4 and IPv6), using the {proto} variable, and kind (see the documentation on groups below), using the {kind} variable.
The bootstrap template is rendered multiple times, for each possible protocol and kind combination.
For example, if your groups configuration defines three different kinds, a total of six bootstrap files will be generated (three for IPv4 and three for IPv6).
This means that if the {proto} and {kind} variables are not used in the output setting, a given rendered file may overwrite a previously generated one.
Example:
bootstrap: {
input: "/etc/drib/bootstrap.tpl",
output: "/etc/drib/bootstrap_{proto}_{kind}",
}This section defines input and output settings for diff mode, as described in the bootstrap section.
inputrefers to a template file (see the Templates section below for details) used to render the bootstrap ranges.outputspecifies the path of the rendered file.max_ranges_per_filelimits the number of ranges rendered in a single output file.
The output setting differs from the one in the bootstrap section in that a single variable is available, {i}, which corresponds to the ith output file.
This refers to the fact that diff mode also includes the max_ranges_per_file setting, which allows you to limit the size of the output files.
Once the number of ranges rendered in the output file reaches the max_ranges_per_file value, a new file will be generated, and the i variable will be incremented.
This variable supports an integer modifier that indicates how many digits are used for the index, so, for example, {3i} will represent the index with 3 digits, padding it with zeros if necessary.
Example:
diff: {
input: "/etc/drib/policy_update.lua.tpl",
output: "/etc/drib/policy_update.{2i}",
max_ranges_per_file: 1000,
}The downloads section allows you to specify range sources to be downloaded for later use in the ipv4 and ipv6 sections (see below).
This is useful because a number of public IP lists include both IPv4 and IPv6 addresses, while in Drib you must specify sources separately by protocol.
By using the downloads section, you can avoid downloading the same file twice.
For details on how to refer to a downloaded range, see the groups documentation below.
Each entry in the downloads section defines a download with a given name and two settings:
url: the URL pointing to the file to be downloaded (only HTTP and HTTPS URLs are supported).check_interval: how often to download this file.
The check_interval setting is specified with a suffix that indicates the time unit, i.e. s for seconds, m for minutes or h for hours.
Downloads are also subject to the Last-Modified HTTP header, so a file won't be downloaded after check_interval if it hasn't been modified since its last download.
Example:
downloads: {
amazon: {
url: "https://ip-ranges.amazonaws.com/ip-ranges.json",
check_interval: "1d",
},
fastly: {
url: "https://api.fastly.com/public-ip-list",
check_interval: "1d",
},
}The core of Drib's configuration is in the groups specification.
Two group lists are supported, ipv4 and ipv6.
Each group in a list contains a number of IP range feeds (containing addresses of the approriate protocol version) and two additional settings:
priority: the priority of this group.kind: an arbitrary string to be associated with ranges in this group.
Precedence is given to groups according to their configured priority setting: higher priority groups will always "own" any ranges that intersect with lower priority ones.
In other words, range intersections between groups causes the intersection to be removed from the lower priority group.
The kind parameter can be used to associate a property with the group's ranges.
An example would be specifying if those ranges are to be handled as source or destination addresses.
Feeds are specified by a name, a source and a class. The source indicates how this feed's IP ranges are obtained. A number of sources is supported:
range: a literal IP range or domain specified as a string.file: a local file containing IP ranges or domains.remote: a remote range to be obtained via HTTP or HTTPS.download: a reference to a downloaded file in thedownloadssection of the configuration file.
A range feed takes no extra parameters.
A file feed takes a path parameter, indicating the file that contains the ranges, and a parser parameter that specifies how the file is interpreted (see the Parsers section below).
A remote feed takes an url and a check_interval, in the same way as entries in the downloads section, and also a parser parameter, as in file feeds.
Finally, download feeds takes a name parameter that must match the name of a download entry in the downloads section of the configuration file, along with a parser.
The class of a feed is given by its class parameter.
This is an arbitrary string that can be used to group IP ranges from different feeds.
Within a group, two feeds with different classes cannot have a non-empty intersection.
Group priorities allow the implementation of groups that work as white or blacklists.
To do this, use a group's kind or a feed's class attribute in the templates in a way that allows the tool in charge of processing Drib's output to identify their appropriate roles.
For a concrete example of this, see the use case described below in the "Full example" section.
Example:
ipv4: {
my_group: {
priority: 10,
kind: "source",
# A literal range
my_ip: {
range: "1.2.3.4/32"
class: "1",
},
# Domains loaded from a file
static: {
file: {
path: "/etc/drib/domain_whitelist",
parser: {domains: {one_per_line: {comment: "#"}}},
},
class: "2",
},
# Ranges loaded from a remote range
cloudflare: {
remote: {
url: "https://www.cloudflare.com/ips-v4",
check_interval: "1d",
parser: {ranges: {one_per_line: {comment: "#"}}},
},
class: "3",
},
},
# Reference to a downloaded source
fastly: {
download: {
name: "fastly",
parser: {ranges: {json: {path: "addresses"}}},
},
class: "4",
},
}For feeds of types file, remote and download, Drib requires a parser to be specified so that it can extract IP ranges from each given source.
The first step is to specify if the source contains IP ranges or domain names, by using either ranges or domains keywords.
Then the parser itself must be declared, according to one of the suported parsers described below.
This is a parser with a single element (either IP range or domain) per line.
Its only parameter is the comment string.
Example:
parser: {
ranges: {
one_per_line: {
comment: "#",
},
},
},This is a parser for tabular data. It accepts the following parameters:
comment: the comment string.header: set astrueif the data contains a header,falseotherwise.columns: indices of columns from which data is extracted in each row of the table.join: the string with which the fields extracted from thecolumnssetting are joined to form the final IP range or domain.
Example:
parser: {
ranges: {
csv: {
comment: "#",
header: true,
separator: ",",
columns: [0, 2],
join: "/",
},
},
},This is a parser for JSON data. The parser expects a path to an array of IP ranges or domains. If those are listed in an array of JSON objects, the parser also point to the key that contains the actual data we're interested in.
path: the JSON path pointing to the array containing the ranges or domains.key: if the array pointed to bypathcontains JSON objects, this specifies the key whose value is the range or domain.filter: if the array pointed to by path contains JSON objects, ignore elements that don't match the given filter.
The filter above is a two-element array containing a key name and a value. If the value associated to the given key in an array element doesn't match the given value, the element is ignored.
Example:
Given the JSON document below:
{
"ipv4": {
"prefixes": [
{
"type": "foo",
"prefix": "1.2.3.4/32"
},
{
"type": "foo",
"prefix": "1.2.3.5/32"
},
{
"type": "bar",
"prefix": "1.2.3.6/32"
}
]
}
}The following parser would extract the ranges 1.2.3.4/32 and 1.2.3.5/32 from it:
parser: {
ranges: {
json: {
path: "ipv4.prefixes",
key: "prefix",
filter: ["type", "foo"],
},
},
},Drib uses the Rust crate Tera for its templating, which has a simple and intuitive syntax similar do Django templates.
When running in bootstrap mode, Drib provides a global ranges object which is an array of entry elements.
The entry elements contain information about each range in the following fields:
priority: the priority associated to the range, taken from the definition of the group it comes from.kind: the kind associated to the range, also taken from the group definition.class: the class associated to the range, taken from the definition of the feed it belongs to.protocol: the protocol associated to the range, as a string (either"ipv4"or"ipv6").range: the IP range itself.
The example below creates iptables rules blocking ranges in bootstrap mode:
{% for entry in ranges -%}
{% if entry.protocol == "ipv4" -%}
{% set command = "iptables" -%}
{% else -%}
{% set command = "ip6tables" -%}
{% endif -%}
{{command}} -I INPUT -s {{entry.range}} -j DROP
{% endfor -%}
When running in diff mode, two global objects are provided to the template: ipv4 and ipv6.
Both contain two fields, remove and insert, which are arrays of entry elements as described above.
The example below manages iptables rules in diff mode:
{% for entry in ipv4.remove -%}
{% if entry.kind == "src" -%}
{% set param = "-s" -%}
{% else -%}
{% set param = "-d" -%}
{% endif -%}
iptables -D INPUT {{param}} {{entry.range}} -j DROP
{% endfor -%}
{% for entry in ipv4.insert -%}
{% if entry.kind == "src" -%}
{% set param = "-s" -%}
{% else -%}
{% set param = "-d" -%}
{% endif -%}
iptables -I INPUT {{param}} {{entry.range}} -j DROP
{% endfor -%}
{% for entry in ipv6.remove -%}
{% if entry.kind == "src" -%}
{% set param = "-s" -%}
{% else -%}
{% set param = "-d" -%}
{% endif -%}
ip6tables -D INPUT {{param}} {{entry.range}} -j DROP
{% endfor -%}
{% for entry in ipv6.insert -%}
{% if entry.kind == "src" -%}
{% set param = "-s" -%}
{% else -%}
{% set param = "-d" -%}
{% endif -%}
ip6tables -I INPUT {{param}} {{entry.range}} -j DROP
{% endfor -%}
For clarity, a full usage example is provided. The idea is to manage firewall rules by obtaining IP ranges from public blacklist sources such as Spamhaus' DROP or Team Cymru's Bogons.
Given the large amount of ranges, the example will use Netfilter's ipset feature to avoid a linear search of matching firewall rules.
Before running Drib, we'll setup IP sets where ranges will be inserted, for allowed and blocked IPv4 and IPv6 ranges. Run the following commands:
$ sudo ipset create drib-ipv4-allow hash:net family inet
$ sudo ipset create drib-ipv4-reject hash:net family inet
$ sudo ipset create drib-ipv6-allow hash:net family inet6
$ sudo ipset create drib-ipv6-reject hash:net family inet6Then create the respective iptables rules matching the above sets:
$ sudo iptables -I INPUT -m set --match-set drib-ipv4-allow src -j ACCEPT
$ sudo iptables -I INPUT -m set --match-set drib-ipv4-reject src -j DROP
$ sudo ip6tables -I INPUT -m set --match-set drib-ipv6-allow src -j ACCEPT
$ sudo ip6tables -I INPUT -m set --match-set drib-ipv6-reject src -j DROPWe'll configure Drib to fetch IP ranges from the above mentioned blacklists.
We'll also block access coming from the IP addresses of the www.spammers-r-us.com domain.
Finally, we'll allow connections from our office's IP address, and from major CDN providers (Cloudflare, Cloudfront and Fastly), regardless of any of them being listed in any of the blacklists.
If, for whatever reasons, an IP from one of the CDN providers ends up in one of the blacklists, it will not be blocked, because the whitelist group has a higher value in its priority field, and therefore any intersection with the ranges from the blacklist group will be removed from the latter.
Create /etc/drib/drib.yaml as below (default settings are ommited):
bootstrap: {
input: "/etc/drib/insert_ranges.sh.tpl",
output: "/etc/drib/insert_{proto}_{kind}_ranges.sh",
}
diff: {
input: "/etc/drib/manage_ranges.sh.tpl",
output: "/etc/drib/manage_ranges.{3i}.sh",
max_ranges_per_file: 1000,
}
# Amazon and Fastly provide their IP ranges in the same source, so we
# download them here and reference them later.
downloads: {
amazon: {
url: "https://ip-ranges.amazonaws.com/ip-ranges.json",
check_interval: "1d",
},
fastly: {
url: "https://api.fastly.com/public-ip-list",
check_interval: "1d",
},
}
# Parser definitions: these parsers are used by multiple feeds, so we
# use YAML's anchors and aliases feature to avoid repetition.
one_range_per_line_hash_comments: &one_range_per_line_hash_comments {
ranges: {
one_per_line: {
comment: "#",
},
},
}
one_range_per_line_semicolon_comments: &one_range_per_line_semicolon_comments {
ranges: {
one_per_line: {
comment: ";",
},
},
}
#
# IPv4 groups
#
ipv4: {
whitelist: {
priority: 20,
kind: "src",
office: {
range: "1.2.3.4/32",
class: "allow",
},
cloudflare: {
remote: {
url: "https://www.cloudflare.com/ips-v4",
check_interval: "1d",
parser: *one_range_per_line_hash_comments,
},
class: "allow",
},
fastly: {
download: {
name: "fastly",
parser: {ranges: {json: {path: "addresses"}}},
},
class: "allow",
},
cloudfront: {
download: {
name: "amazon",
parser: {
ranges: {
json: {
path: "prefixes",
key: "ip_prefix",
filter: ["service", "CLOUDFRONT"],
},
},
},
},
class: "allow",
},
},
# The blacklist group has a lower priority than the whitelist group
# above (10 vs 20), so any intersection will be kept in the whitelist
# and removed from the blacklist.
blacklist: {
priority: 10,
kind: "src",
fullbogons: {
remote: {
url: "https://www.team-cymru.org/Services/Bogons/fullbogons-ipv4.txt",
check_interval: "1d",
parser: *one_range_per_line_hash_comments,
},
class: "reject",
},
spamhaus_drop: {
remote: {
url: "https://www.spamhaus.org/drop/drop.txt",
check_interval: "12h",
parser: *one_range_per_line_semicolon_comments,
},
class: "reject",
},
},
}
#
# IPv6 groups
#
ipv6: {
whitelist: {
priority: 20,
kind: "src",
office: {
range: "a:b:c:d::/64",
class: "allow",
},
cloudflare: {
remote: {
url: "https://www.cloudflare.com/ips-v6",
check_interval: "1d",
parser: *one_range_per_line_hash_comments,
},
class: "allow",
},
fastly: {
download: {
name: "fastly",
parser: {ranges: {json: {path: "ipv6_addresses"}}},
},
class: "allow",
},
cloudfront: {
download: {
name: "amazon",
parser: {
ranges: {
json: {
path: "ipv6_prefixes",
key: "ipv6_prefix",
filter: ["service", "CLOUDFRONT"],
},
},
},
},
class: "allow",
},
},
blacklist: {
priority: 10,
kind: "src",
fullbogons: {
remote: {
url: "https://www.team-cymru.org/Services/Bogons/fullbogons-ipv6.txt",
check_interval: "1d",
parser: *one_range_per_line_hash_comments,
},
class: "reject",
},
spamhaus_drop: {
remote: {
url: "https://www.spamhaus.org/drop/dropv6.txt",
check_interval: "12h",
parser: *one_range_per_line_semicolon_comments,
},
class: "reject",
},
},
}Now we proceed to create the bootstrap and diff templates.
They will call the ipset command to insert or remove ranges from the sets created in the section above.
For the bootstrap template, create /etc/drib/insert_ranges.sh.tpl with the following content:
#!/bin/sh
{% for entry in ranges -%}
ipset add drib-{{entry.protocol}}-{{entry.class}} {{entry.range}}
{% endfor -%}
For the diff template, create /etc/drib/manage_ranges.sh.tpl with the content below.
#!/bin/sh
{% for entry in ipv4.remove -%}
ipset del drib-ipv4-{{entry.class}} {{entry.range}}
{% endfor -%}
{% for entry in ipv4.insert -%}
ipset add drib-ipv4-{{entry.class}} {{entry.range}}
{% endfor -%}
{% for entry in ipv6.remove -%}
ipset add drib-ipv6-{{entry.class}} {{entry.range}}
{% endfor -%}
{% for entry in ipv6.insert -%}
ipset del drib-ipv6-{{entry.class}} {{entry.range}}
{% endfor -%}
We can finally run Drib. Run the command below to generate the bootstrap scripts.
$ sudo drib bootstrapThis will generate two scripts, /etc/drib/insert_ipv4_src_ranges.sh and /etc/drib/insert_ipv6_src_ranges.sh, which can be used to populate the appropriate IP sets.
From this point on, the IP sets can be managed by Drib running on diff mode:
$ sudo drib diffThis will create a number of scripts, depending on the number of updates computed by Drib.
In the configuration above, we have limited the output to 1000 ranges per file, so Drib will generate as many files as necessary following the specified output naming pattern, i.e., /etc/drib/manage_ranges.00.sh, /etc/drib/manage_ranges.01.sh and so on.