omfwd: syslog Forwarding Output Module

Template

Sets a custom default template for this module.

iobuffer.maxSize

The iobuffer.maxSize parameter sets the maximum size of the I/O buffer used by rsyslog when submitting messages to the TCP send API. This parameter allows limiting the buffer size to a specific value and is primarily intended for testing purposes, such as within an automated testbench. By default, the full size of the I/O buffer is used, which depends on the rsyslog version. If the specified size is too large, an error is emitted, and rsyslog reverts to using the full size.

module(load="omfwd" iobuffer.maxSize="8")

Target

Name or IP address of the system to receive messages. Any resolvable name is fine. Here either a single target or an array of targets can be provided.

If an array is provided, rsyslog forms a “target pool”. Inside the pool, it performs equal load-balancing among them. Targets are changed for each message being sent. If targets become unreachable, they will temporarily not participate in load balancing. If all targets become offline (then and only then) the action itself is suspended. Unreachable targets are automatically retried by omfwd.

NOTE: target pools are ONLY avaiable for TCP transport. If UDP is selected, an error message is emitted and only the first target used.

Single target: Target=”syslog.example.net”

Array of targets: Target=[“syslog1.example.net”, “syslog2.example.net”, “syslog3.example.net”]

Port

Name or numerical value of the port to use when connecting to the target. If multiple targes are defined, different ports can be defined for each target. To do so, use array mode. The first port will be used for the first target, the second for the second target and so on. If fewer ports than targets are defined, the remaining targets will use the first port configured. This also means that you also need to define a single port, if all targets should use the same port.

Note: if more ports than targets are defined, the remaining ports are ignored and an error message is emitted.

pool.resumeinterval

If a targt pool exists, “pool.resumeinterval” configures how often an unavailable target is tried to be activated. A new connection request will be made in roughly “pool.resumeinterval” seconds until connection is reestablished or the action become completely suspenden (in which case the action settings take control).

Please note the word “roughly”: the interval may be some seconds earlier or later on a try-by-try basis because of other ongoing activity inside rsyslog.

Warning: we do NOT recommend to set this interval below 10 seconds, as it can lead DoS-like reconnection behaviour. Actually, the default of 30 seconds is quite short and should be extended if the use case permits.

Protocol

Type of protocol to use for forwarding. Note that tcp includes both legacy plain TCP syslog and RFC5425-based TLS-encrypted syslog. The selection depends on the StreamDriver parameter. If StreamDriver is set to “ossl” or “gtls”, it will use TLS-encrypted syslog.

NetworkNamespace

Name of a network namespace in /var/run/netns/ to use for forwarding.

If the setns() system call is unavailable (e.g., BSD kernel, Linux kernel before v2.6.24), the given namespace will be ignored.

Address

Bind socket to a specific local IP address. This option is supported for UDP only, not TCP.

IpFreeBind

Manages the IP_FREEBIND option on the UDP socket, which allows binding it to an IP address that is not yet associated to any network interface. This option is only relevant if the address option is set.

The parameter accepts the following values:

• 0 - does not enable the IP_FREEBIND option on the UDP socket. If the bind() call fails because of EADDRNOTAVAIL error, socket initialization fails.

• 1 - silently enables the IP_FREEBIND socket option if it is required to successfully bind the socket to a nonlocal address.

• 2 - enables the IP_FREEBIND socket option and warns when it is used to successfully bind the socket to a nonlocal address.

Device

Bind socket to given device (e.g., eth0)

For Linux with VRF support, the Device option can be used to specify the VRF for the Target address.

TCP_Framing

Framing mode used for forwarding: either “traditional” or “octet-counted”. This applies only to TCP-based protocols and is ignored for UDP. In protocol engineering, “framing” refers to how multiple messages over the same connection are separated. Usually, this is transparent to users. However, the early syslog protocol evolved in such a way that users sometimes need to specify the framing.

“Traditional” framing is non-transparent, where messages end when an LF (line feed) is encountered, and the next message starts immediately after the LF. If multi-line messages are received, they are split into multiple messages, with all but the first segment usually incorrectly formatted.

“Octet-counted” framing addresses this issue. Each message is prefixed with its length, so the receiver knows exactly where the message ends. Multi-line messages are handled correctly. This mode is similar to the method described in RFC5425 for TLS-enabled syslog. Unfortunately, few syslog implementations support “octet-counted” framing. As such, “traditional” framing is the default, despite its defects. If the receiver supports “octet-counted” framing, it is recommended to use that mode.

TCP_FrameDelimiter

Sets a custom frame delimiter for TCP transmission when running TCP_Framing in “traditional” mode. The delimiter has to be a number between 0 and 255 (representing the ASCII-code of said character). The default value for this parameter is 10, representing a ‘\n’. When using Graylog, the parameter must be set to 0.

ZipLevel

Compression level for messages.

Up until rsyslog 7.5.1, this was the only compression setting that rsyslog understood. Starting with 7.5.1, we have different compression modes. All of them are affected by the ziplevel. If, however, no mode is explicitly set, setting ziplevel also turns on “single” compression mode, so pre 7.5.1 configuration will continue to work as expected.

The compression level is specified via the usual factor of 0 to 9, with 9 being the strongest compression (taking up most processing time) and 0 being no compression at all (taking up no extra processing time).

compression.Mode

mode is one of “none”, “single”, or “stream:always”. The default is “none”, in which no compression happens at all. In “single” compression mode, Rsyslog implements a proprietary capability to zip transmitted messages. That compression happens on a message-per-message basis. As such, there is a performance gain only for larger messages. Before compressing a message, rsyslog checks if there is some gain by compression. If so, the message is sent compressed. If not, it is sent uncompressed. As such, it is totally valid that compressed and uncompressed messages are intermixed within a conversation.

In “stream:always” compression mode the full stream is being compressed. This also uses non-standard protocol and is compatible only with receives that have the same abilities. This mode offers potentially very high compression ratios. With typical syslog messages, it can be as high as 95+% compression (so only one twentieth of data is actually transmitted!). Note that this mode introduces extra latency, as data is only sent when the compressor emits new compressed data. For typical syslog messages, this can mean that some hundred messages may be held in local buffers before they are actually sent. This mode has been introduced in 7.5.1.

Note: currently only imptcp supports receiving stream-compressed data.

compression.stream.flushOnTXEnd

This setting affects stream compression mode, only. If enabled (the default), the compression buffer will by emptied at the end of a rsyslog batch. If set to “off”, end of batch will not affect compression at all.

While setting it to “off” can potentially greatly improve compression ratio, it will also introduce severe delay between when a message is being processed by rsyslog and actually sent out to the network. We have seen cases where for several thousand message not a single byte was sent. This is good in the sense that it can happen only if we have a great compression ratio. This is most probably a very good mode for busy machines which will process several thousand messages per second and the resulting short delay will not pose any problems. However, the default is more conservative, while it works more “naturally” with even low message traffic. Even in flush mode, notable compression should be achievable (but we do not yet have practice reports on actual compression ratios).

RebindInterval

Permits to specify an interval at which the current connection is broken and re-established. This setting is primarily an aid to load balancers. After the configured number of batches (equals roughly to messages for UDP traffic, dependent on batch size for TCP) has been transmitted, the current connection is terminated and a new one started. Note that this setting applies to both TCP and UDP traffic. For UDP, the new ``connection’’ uses a different source port (ports are cycled and not reused too frequently). This usually is perceived as a ``new connection’’ by load balancers, which in turn forward messages to another physical target system.

KeepAlive

Enable or disable keep-alive packets at the tcp socket layer. The default is to disable them.

KeepAlive.Probes

The number of unacknowledged probes to send before considering the connection dead and notifying the application layer. The default, 0, means that the operating system defaults are used. This has only effect if keep-alive is enabled. The functionality may not be available on all platforms.

KeepAlive.Interval

The interval between subsequential keepalive probes, regardless of what the connection has exchanged in the meantime. The default, 0, means that the operating system defaults are used. This has only effect if keep-alive is enabled. The functionality may not be available on all platforms.

KeepAlive.Time

The interval between the last data packet sent (simple ACKs are not considered data) and the first keepalive probe; after the connection is marked to need keepalive, this counter is not used any further. The default, 0, means that the operating system defaults are used. This has only effect if keep-alive is enabled. The functionality may not be available on all platforms.

ConErrSkip

The ConErrSkip can be used to limit the number of network errors recorded in logs. For example, value 10 means that each 10th error message is logged. Note that this options should be used as the last resort since the necessity of its use indicates network issues. The default behavior is that all network errors are logged.

RateLimit.Interval

Specifies the rate-limiting interval in seconds. Default value is 0, which turns off rate limiting.

RateLimit.Burst

Specifies the rate-limiting burst in number of messages.

StreamDriver

Choose the stream driver to be used. Default is plain tcp, but you can also choose “ossl” or “gtls” for TLS encryption.

StreamDriverMode

Mode to use with the stream driver (driver-specific)

StreamDriverAuthMode

Authentication mode to use with the stream driver. Note that this parameter requires TLS netstream drivers. For all others, it will be ignored. (driver-specific).

StreamDriver.PermitExpiredCerts

Controls how expired certificates will be handled when stream driver is in TLS mode. It can have one of the following values:

• on = Expired certificates are allowed

• off = Expired certificates are not allowed (Default, changed from warn to off since Version 8.2012.0)

• warn = Expired certificates are allowed but warning will be logged

StreamDriverPermittedPeers

Accepted fingerprint (SHA1) or name of remote peer. Note that this parameter requires TLS netstream drivers. For all others, it will be ignored. (driver-specific)

StreamDriver.CheckExtendedKeyPurpose

Whether to check also purpose value in extended fields part of certificate for compatibility with rsyslog operation. (driver-specific)

StreamDriver.PrioritizeSAN

Whether to use stricter SAN/CN matching. (driver-specific)

StreamDriver.TlsVerifyDepth

Specifies the allowed maximum depth for the certificate chain verification. Support added in v8.2001.0, supported by GTLS and OpenSSL driver. If not set, the API default will be used. For OpenSSL, the default is 100 - see the doc for more: https://www.openssl.org/docs/man1.1.1/man3/SSL_set_verify_depth.html For GnuTLS, the default is 5 - see the doc for more: https://www.gnutls.org/manual/gnutls.html

StreamDriver.CAFile

This permits to override the CA file set via global() config object at the per-action basis. This parameter is ignored if the netstream driver and/or its mode does not need or support certificates.

StreamDriver.CRLFile

This permits to override the CRL (Certificate revocation list) file set via global() config object at the per-action basis. This parameter is ignored if the netstream driver and/or its mode does not need or support certificates.

StreamDriver.KeyFile

This permits to override the CA file set via global() config object at the per-action basis. This parameter is ignored if the netstream driver and/or its mode does not need or support certificates.

StreamDriver.CertFile

This permits to override the CA file set via global() config object at the per-action basis. This parameter is ignored if the netstream driver and/or its mode does not need or support certificates.

ResendLastMSGOnReconnect

Permits to resend the last message when a connection is reconnected. This setting affects TCP-based syslog, only. It is most useful for traditional, plain TCP syslog. Using this protocol, it is not always possible to know which messages were successfully transmitted to the receiver when a connection breaks. In many cases, the last message sent is lost. By switching this setting to “yes”, rsyslog will always retransmit the last message when a connection is reestablished. This reduces potential message loss, but comes at the price that some messages may be duplicated (what usually is more acceptable).

Please note that busy systems probably loose more than a single message in such cases. This is caused by an inherant unreliability in plain tcp syslog and there is no way rsyslog could prevent this from happening (if you read the detail description, be sure to follow the link to the follow-up posting). In order to prevent these problems, we recommend the use of omrelp.

udp.SendToAll

When sending UDP messages, there are potentially multiple paths to the target destination. By default, rsyslogd only sends to the first target it can successfully send to. If this option is set to “on”, messages are sent to all targets. This may improve reliability, but may also cause message duplication. This option should be enabled only if it is fully understood.

Note: this option replaces the former -A command line option. In contrast to the -A option, this option must be set once per input() definition.

udp.SendDelay

This is an expert option, do only use it if you know very well why you are using it!

This options permits to introduce a small delay after each send operation. The integer specifies the delay in microseconds. This option can be used in cases where too-quick sending of UDP messages causes message loss (UDP is permitted to drop packets if e.g. a device runs out of buffers). Usually, you do not want this delay. The parameter was introduced in order to support some testbench tests. Be sure to think twice before you use it in production.

gnutlsPriorityString

This strings setting is used to configure driver specific properties. Historically, the setting was only meant for gnutls driver. However with version v8.1905.0 and higher, the setting can also be used to set openssl configuration commands.

For GNUTls, the setting specifies the TLS session’s handshake algorithms and options. These strings are intended as a user-specified override of the library defaults. If this parameter is NULL, the default settings are used. More information about priority Strings here.

For OpenSSL, the setting can be used to pass configuration commands to openssl library. OpenSSL Version 1.0.2 or higher is required for this feature. A list of possible commands and their valid values can be found in the documentation: https://www.openssl.org/docs/man1.0.2/man3/SSL_CONF_cmd.html

The setting can be single or multiline, each configuration command is separated by linefeed (n). Command and value are separated by equal sign (=). Here are a few samples:

extendedConnectionCheck

This setting permits to control if rsyslog should try to detect if the remote syslog server has broken the current TCP connection. It is has no meaning when UDP protocol is used.

Generally, broken connections are not easily detectable. That setting does additional API calls to check for them. This causes some extra overhead, but is traditionally enabled.

Especially in very busy systems it is probably worth to disable it. The extra overhead is unlikely to bring real benefits in such scenarios.

Note: If you need reliable delivery, do NOT use plain TCP syslog transport. Use RELP instead.