nbdkit-ip-filter - filter clients by IP address, process ID, user ID, group ID and more
nbdkit --filter=ip PLUGIN [allow=addr[,addr...]]
[deny=addr[,addr...]]nbdkit-ip-filter can allow or deny client connections by their IP address.
Usually it is better to control this outside nbdkit, for example using TCP wrappers or a firewall, but this filter can be used if these are not available.
nbdkit --filter=ip [...] allow=127.0.0.1,::1 deny=allAllow clients to connect on the loopback IPv4 or loopback IPv6 address, deny all other clients.
nbdkit --filter=ip [...] deny=8.0.0.0/8Allow any client except connections from the IPv4 8.0.0.0/8 network.
nbdkit --filter=ip [...] allow=anyipv6 deny=allAllow IPv6 clients to connect from anywhere, deny all other sources.
nbdkit -U $tmpdir/sock --filter=ip [...] allow=uid:`id -u` deny=allOnly allow the current user (id -u) to connect over the socket.
Layer extra security by creating the socket inside a temporary directory only accessible by the user.
nbdkit -U $tmpdir/sock --filter=ip [...] allow=gid:`id -g` deny=allAllow anyone in the same group as the current user to connect to the Unix domain socket.
As in the previous example, layer extra security by creating the socket inside a temporary directory only accessible by the group.
When a client connects, this filter checks its source address against the allow and deny lists as follows:
If the address matches any in the allow list, permission is granted.
If the address matches any in the deny list, permission is denied.
Otherwise permission is granted.
If either the allow or deny parameter is not present then it is assumed to be an empty list. The order in which the parameters appear on the command line does not matter; the allow list is always processed first and the deny list second.
The allow and deny parameters each contain a comma-separated list of any of the following:
These keywords (which both have the same meaning) match any source.
These keywords match any IPv4 address.
These keywords match any IPv6 address.
These keywords match any connection over a Unix domain socket.
These keywords match any connection over an AF_VSOCK socket.
This matches the single IPv4 address A.B.C.D, for example 127.0.0.1.
This matches the range of IPv4 addresses A.B.C.D/NN, for example 192.168.2.0/24 or 10.0.0.0/8.
This matches the single IPv6 address A:B:.... The usual IPv6 address representations can be used (see RFC 5952).
This matches a range of IPv6 addresses A:B:.../NN.
(nbdkit ≥ 1.40, not Windows)
This matches the TLS Distinguished Name (DN) of the client certificate or client certificate's issuer against the WILDCARD. In the example below quotes are used to protect the wildcard from the shell, they are not part of the matching:
nbdkit --filter=ip allow=dn:"*,O=BigCo,*" [...]
nbdkit --filter=ip allow=issuer-dn:"CN=BigCo" [...]See nbdkit_peer_tls_dn(3), nbdkit_peer_tls_issuer_dn(3) and nbdkit-tls(1) for further information about DNs.
Using this rule has important performance implications, see "Late filtering" below.
Comma splitting is not done inside parameters that start with dn: or issuer-dn:. See "Comma splitting" below.
(nbdkit ≥ 1.24, Linux only)
This matches the process ID PID, if the client connects over a Unix domain socket.
Note that process IDs are recycled so this alone is not secure enough to ensure that only a single desired process can connect. However you could use it as an additional check.
(nbdkit ≥ 1.36, not Windows)
This matches the security context (usually the SELinux label, IPSEC label or NetLabel) of the client.
(nbdkit ≥ 1.24)
This matches the numeric user ID UID, if the client connects over a Unix domain socket.
(nbdkit ≥ 1.24)
This matches the numeric group ID GID, if the client connects over a Unix domain socket.
(nbdkit ≥ 1.24)
These match the CID or port number for AF_VSOCK sockets.
Set list of allow rules. This parameter is optional, if omitted the allow list is empty.
Set list of deny rules. This parameter is optional, if omitted the deny list is empty.
If neither the allow nor the deny parameter is given the filter does nothing.
Unix domain sockets and AF_VSOCK sockets were always unfiltered in nbdkit ≤ 1.22. In nbdkit ≥ 1.24 the ability to filter them was added.
In nbdkit ≤ 1.38, connections from non-IP, non-Unix, non-vsock sockets (whatever that would be) were allowed unfiltered. In nbdkit ≥ 1.40 unknown socket families are denied.
Permit known good connections and deny everything else:
nbdkit --filter=ip ... allow=good1,good2,... deny=allBlock troublemakers but allow everything else:
nbdkit --filter=ip ... deny=bad1,bad2,...You can supply each allow and deny parameter multiple times. The rules are concatenated.
Also each allow or deny parameter can contain multiple rules, split at commas (,).
These sets of rules are equivalent:
nbdkit --filter=ip allow=good1 allow=good2,good3 deny=all
nbdkit --filter=ip allow=good1,good2 allow=good3 deny=allBecause TLS Distinguished Names (DNs) contain commas, the filter has a special exception where if the first characters of the allow or deny parameter match "dn:" or "issuer-dn:" then the whole parameter is not split. This lets you match on multiple DNs naturally:
nbdkit --filter=ip \
allow=dn:"*,O=BigCo,*" \
allow=issuer-dn:"CN=BigCo" [...]This filter normally runs the filtering rules very early in the connection, just after nbdkit has received an open socket from the client. Filtering happens in the .preconnect callback.
"Callback lifecycle" in nbdkit-plugin(3) explains the connection lifecycle in more detail.
However if your allow or deny patterns contain any dn: or issuer-dn: rules then all filtering has to be done later in the connection lifecycle, since we must wait until TLS negotiation has been completed. Filtering happens in the .open or .list_exports callback instead.
In practice this makes filtering more expensive and potentially makes it easier to cause Denial of Service (DoS) attacks. You should try to combine dn: and issuer-dn: rules with extra filtering outside nbdkit, such as using a firewall or VPN.
Debug rules and rule matching. If clients are accepted or rejected when they should not be, using -v -D ip.rules=1 can help to debug the problem.
The filter.
Use nbdkit --dump-config to find the location of $filterdir.
nbdkit-ip-filter first appeared in nbdkit 1.18.
nbdkit(1), nbdkit-exitlast-filter(1), nbdkit-exitwhen-filter(1), nbdkit-limit-filter(1), nbdkit-time-limit-filter(1), nbdkit-filter(3), nbdkit_peer_name(3), nbdkit_peer_pid(3), nbdkit_peer_uid(3), nbdkit_peer_gid(3), nbdkit_peer_security_context(3), nbdkit_peer_tls_dn(3), nbdkit_peer_tls_issuer_dn(3), nbdkit-tls(1).
Richard W.M. Jones
Copyright Red Hat
Redistribution and use in source and binary forms, with or without modification, are permitted provided that the following conditions are met:
Redistributions of source code must retain the above copyright notice, this list of conditions and the following disclaimer.
Redistributions in binary form must reproduce the above copyright notice, this list of conditions and the following disclaimer in the documentation and/or other materials provided with the distribution.
Neither the name of Red Hat nor the names of its contributors may be used to endorse or promote products derived from this software without specific prior written permission.
THIS SOFTWARE IS PROVIDED BY RED HAT AND CONTRIBUTORS ''AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL RED HAT OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.