UDP (im_udp)

This module accepts UDP datagrams on the configured address and port. UDP is the transport protocol of the legacy BSD Syslog as described in RFC 3164, so this module can be particularly useful to receive such messages from older devices which do not support other transports.

To examine the supported platforms, see the list of installer packages in the Available Modules chapter.
UDP is an unreliable transport protocol, and does not guarantee delivery. Messages may not be received or may be truncated. It is recommended to use the TCP or SSL transport modules instead, if possible.

To reduce the likelihood of message loss, consider:

  • increasing the socket buffer size with SockBufSize,

  • raising the route priority by setting the Priority directive (to a low number such as 1), and

  • adding additional buffering by increasing the LogqueueSize or adding a pm_buffer instance.

This module does not provide access control. Firewall rules can be used to deny connections from certain hosts.

For parsing syslog messages, see the parse_syslog_bsd() procedure of the xm_syslog module.

Configuration

The im_udp module accepts the following directives in addition to the common module directives.

ListenAddr

The module accepts connections on the IP address or hostname and port defined here. The default address is localhost and the default port is 514. The port number can be defined by appending it at the end of the hostname or IP address using a colon as a separator (host:port). The port section of this directive and the Port directive are mutually exclusive. In case both is defined, the port number defined here takes precedence over a port defined in the Port directive. In case none of them is defined, the default port 514 is used.
To listen on multiple addresses or ports in a single module instance, this directive can be repeated multiple times. Both IPv4 and IPv6 addresses are supported. If a DNS name is used, the number of addresses or cnames should be kept below 16 to avoid potential issues caused by DNS response size limits.

Formerly called Host, this directive is now ListenAddr. Host will become deprecated from NXLog EE 6.0.

When the Host directive is used with a hostname instead of an IP address, the the hostname will be resolved to an IP address for each new connection. If a resolver, e.g. DNS, returns multiple IP addresses, the module will connect to the first IP address. If a single output instance is configured with multiple Host directives or the resolver returns multiple addresses for a name, these hosts are accessed in failover mode. If a Host directive is configured with a hostname, the product performs a name lookup and establishes the connection to the first reachable address in the returned set of addresses. The module will remain connected to that address until it is stopped, or the connection is severed. DNS changes are therefore not picked up by the module without intervention. If the connection fails to the first address of the set, the module will attempt to connect to the next address, until it reaches the end of the set of addresses. Then it performs a lookup on the next Host directive, if so configured. Once all options are exhausted, the module will start over from the first Host directive, cycling through them again until connection can be reestablished.

When the ListenAddr (or similar) directive specifies a hostname, the module binds to the first IP address that is exposed by the system for that hostname. On most systems that support IPv6, this address will be an IPv6 address. This means that client applications and the systems they run on will also need to have IPv6 support enabled, and must be able to connect to the same IPv6 address. NXLog output modules achieve this requirement through failover. For third-party client applications, the configuration details are application-specific but they should have the ability to detect which IP address the server is listening on when using a hostname to connect.

For client applications that don’t support IPv6, to avoid the behavior described above the ListenAddr directive of the listening module may be set to an IPv4 address e.g. 0.0.0.0.

Alternatively, the server-side system may be configured to prioritize IPv4 addresses for the hostname specified by the ListenAddr directive, although this is a more complicated and potentially intrusive approach. On most Linux-based systems, this can be achieved through the /etc/gai.conf configuration file. On BSD-based systems, the ip6addrctl command can be used. Windows is more limited and can only be configured to prioritize IPv4 over IPv6 addresses for ALL hostnames resolved on the system, by setting the following registry value to 0x20:

HKEY_LOCAL_MACHINE\SYSTEM\CurrentControlSet\Services\Tcpip6\Parameters\DisabledComponents

For more information see the Microsoft documentation on Configuring IPv6 in Windows for advanced users.

This limitation will be addressed in a future release by making listening modules bind to all available IPv4/IPv6 addresses that a hostname resolves to.

Port

The module will listen for incoming connections on this port number. The default is port 514.

The Port directive will become deprecated from NXLog EE 6.0. After that, the port can only be defined in the ListenAddr directive.

ReusePort

This optional boolean directive enables synchronous listening on the same port by multiple module instances. Each module instance runs in its own thread, allowing NXLog to process incoming data simultaneously to take better advantage of multiprocessor systems. The default value is FALSE.

To enable synchronous listening, the configuration file should contain multiple im_udp module instances listening on the same port and the ReusePort directive set to TRUE, see the Examples section.

SockBufSize

This optional directive sets the socket buffer size (SO_RCVBUF) to the value specified. If not set, the operating system defaults are used. If UDP packet loss is occurring at the kernel level, setting this to a high value (such as 150000000) may help. On Windows systems the default socket buffer size is extremely low, and using this option is highly recommended.

UseRecvmmsg

This boolean directive specifies that the recvmmsg() system call should be used, if available, to receive multiple messages per call to improve performance. The default is TRUE.

Fields

The following fields are used by im_udp.

$raw_event (type: string)

The received string.

$MessageSourceAddress (type: string)

The IP address of the remote host.

Examples

Pre-v5 syntax examples are included, they will become invalid with NXLog EE 6.0.

Example 1. Using the im_udp module

This configuration accepts log messages via UDP and writes them to a file.

nxlog.conf
<Input udp>
    Module        im_udp
    ListenAddr    192.168.1.1:514
</Input>

# Using the syntax prior to NXLog EE 5,
# where the port is defined in a separate directive.
#<Input udp>
#    Module       im_udp
#    Host         192.168.1.1
#    Port         514
#</Input>

<Output file>
    Module        om_file
    File          "tmp/output"
</Output>

<Route udp_to_file>
    Path          udp => file
</Route>
Example 2. Reusing a single port by multiple module instances

The configuration below provides two im_udp module instances to reuse port 514 via the ReusePort directive. Received messages are written to the /tmp/output file.

nxlog.conf
<Input udp_one>
    Module      im_udp
    ListenAddr 192.168.1.1:514
    ReusePort   TRUE
</Input>

<Input udp_two>
    Module      im_udp
    ListenAddr  192.168.1.1:514
    ReusePort   TRUE
</Input>

# Using the syntax prior to NXLog EE 5,
# where the port is defined in a separate directive.
#<Input udp_one>
#    Module      im_udp
#    Host        192.168.1.1
#    Port        514
#    ReusePort   TRUE
#</Input>
#
#<Input udp_two>
#    Module      im_udp
#    Host        192.168.1.1
#    Port        514
#    ReusePort   TRUE
#</Input>

<Output file>
    Module      om_file
    File        "tmp/output"
</Output>

<Route udp_to_file>
    Path        udp_one, udp_two => file
</Route>