Chat now with support
Chat with Support

syslog-ng Premium Edition 6.0.18 - Administration Guide

Preface Chapter 1. Introduction to syslog-ng Chapter 2. The concepts of syslog-ng Chapter 3. Installing syslog-ng Chapter 4. The syslog-ng PE quick-start guide Chapter 5. The syslog-ng PE configuration file Chapter 6. Collecting log messages — sources and source drivers Chapter 7. Sending and storing log messages — destinations and destination drivers Chapter 8. Routing messages: log paths, reliability, and filters Chapter 9. Global options of syslog-ng PE Chapter 10. TLS-encrypted message transfer Chapter 11. FIPS-compliant syslog-ng Chapter 12.  Reliable Log Transfer Protocol™ Chapter 13. Reliability and minimizing the loss of log messages Chapter 14. Manipulating messages Chapter 15. Parsing and segmenting structured messages Chapter 16. Processing message content with a pattern database Chapter 17. Statistics and metrics of syslog-ng Chapter 18. Multithreading and scaling in syslog-ng PE Chapter 19. Troubleshooting syslog-ng Chapter 20. Best practices and examples

Sending messages to UNIX domain sockets

The unix-stream() and unix-dgram() drivers send messages to a UNIX domain socket in either SOCK_STREAM or SOCK_DGRAM mode.

Both drivers have a single required argument specifying the name of the socket to connect to. For the list of available optional parameters, see the section called “unix-stream() and unix-dgram() destination options”.

Declaration: 

unix-stream(filename [options]);
unix-dgram(filename [options]);

Example 7.45. Using the unix-stream() driver

destination d_unix_stream { unix-stream("/var/run/logs"); };

unix-stream() and unix-dgram() destination options

These drivers send messages to a unix socket in either SOCK_STREAM or SOCK_DGRAM mode. The unix-stream() and unix-dgram() destinations have the following options.

disk-buffer()

Description: This option enables putting outgoing messages into the disk buffer of the destination to avoid message loss in case of a system failure on the destination side. It has the following options:

reliable()
Type: yes|no
Default: no

Description: If set to yes, syslog-ng PE cannot lose logs in case of reload/restart, unreachable destination or syslog-ng PE crash. This solution provides a slower, but reliable disk-buffer option. It is created and initialized at startup and gradually grows as new messages arrive. If set to no, the normal disk-buffer will be used. This provides a faster, but less reliable disk-buffer option.

Caution:

Hazard of data loss! If you change the value of reliable() option when there are messages in the disk-buffer, the messages stored in the disk-buffer will be lost.

dir()
Type: string
Default: N/A

Description: Defines the folder where the disk-buffer files are stored. This option has priority over --qdisk-dir=.

Caution:

When creating a new dir() option for a disk buffer, or modifying an existing one, make sure you delete the persist file, or at least remove the relevant persist-entry.

syslog-ng PE creates disk-buffer files based on the path recorded in the persist file. Therefore, if the persist file or the relevant entry is not deleted after modifying the dir() option, then following a restart, syslog-ng PE will look for or create disk-buffer files in their old location. To ensure that syslog-ng PE uses the new dir() setting, the persist file must not contain any information about the destinations which the disk-buffer file in question belongs to.

disk-buf-size()
Type: number (bytes)
Default:
Description: This is a required option. The maximum size of the disk-buffer in bytes. The minimum value is 1048576 bytes. If you set a smaller value, the minimum value will be used automatically. It replaces the old log-disk-fifo-size() option.
mem-buf-length()
Type: number (messages)
Default: 10000
Description: Use this option if the option reliable() is set to no. This option contains the number of messages stored in overflow queue. It replaces the old log-fifo-size() option. It inherits the value of the global log-fifo-size() option if provided. If it is not provided, the default value is 10000 messages. Note that this option will be ignored if the option reliable() is set to yes.
mem-buf-size()
Type: number (bytes)
Default: 163840000
Description: Use this option if the option reliable() is set to yes. This option contains the size of the messages in bytes that is used in the memory part of the disk buffer. It replaces the old log-fifo-size() option. It does not inherit the value of the global log-fifo-size() option, even if it is provided. Note that this option will be ignored if the option reliable() is set to no.
quot-size()
Type: number (messages)
Default: 64
Description: The number of messages stored in the output buffer of the destination.

Options reliable() and disk-buf-size() are required options.

Example 7.46. Examples for using disk-buffer()

In the following case reliable disk-buffer() is used.

destination d_demo {
    network(
            "127.0.0.1"
            port(3333)
            disk-buffer(
                mem-buf-size(10000)
                disk-buf-size(2000000)
                reliable(yes)
                dir("/tmp/disk-buffer")
            )
        );
};

In the following case normal disk-buffer() is used.

destination d_demo {
    network(
            "127.0.0.1"
            port(3333)
            disk-buffer(
                mem-buf-length(10000)
                disk-buf-size(2000000)
                reliable(no)
                dir("/tmp/disk-buffer")
            )
        );
};

flags()
Type: no_multi_line, syslog-protocol
Default: empty set

Description: Flags influence the behavior of the destination driver.

  • no-multi-line: The no-multi-line flag disables line-breaking in the messages: the entire message is converted to a single line.

  • syslog-protocol: The syslog-protocol flag instructs the driver to format the messages according to the new IETF syslog protocol standard (RFC5424), but without the frame header. If this flag is enabled, macros used for the message have effect only for the text of the message, the message header is formatted to the new standard. Note that this flag is not needed for the syslog driver, and that the syslog driver automatically adds the frame header to the messages.

flush-lines()
Type: number (messages)
Default: Use global setting.

Description: Specifies how many lines are sent to a destination at a time. The syslog-ng PE application waits for this number of lines to accumulate and sends them off in a single batch. Setting this number high increases throughput as fully filled frames are sent to the destination, but also increases message latency.

For optimal performance when sending messages to an syslog-ng PE server, make sure that the flush-lines() is smaller than the window size set using the log-iw-size() option in the source of your server.

flush-timeout() (OBSOLETE)
Type: time in milliseconds
Default: Use global setting.

Description: This is an obsolete option. Specifies the time syslog-ng waits for lines to accumulate in its output buffer. For details, see the flush-lines() option.

NOTE:

This option will be removed from the list of acceptable options. After that, your configuration will become invalid if it still contains the flush-timeout() option. To avoid future problems, remove this option from your configuration.

frac-digits()
Type: number (digits of fractions of a second)
Default: Value of the global option (which defaults to 0)

Description: The syslog-ng application can store fractions of a second in the timestamps according to the ISO8601 format. The frac-digits() parameter specifies the number of digits stored. The digits storing the fractions are padded by zeros if the original timestamp of the message specifies only seconds. Fractions can always be stored for the time the message was received. Note that syslog-ng can add the fractions to non-ISO8601 timestamps as well.

log-fifo-size()
Type: number (messages)
Default: Use global setting.

Description: The number of messages that the output queue can store.

keep-alive()
Type: yes or no
Default: yes

Description: Specifies whether connections to destinations should be closed when syslog-ng is reloaded. Note that this applies to the client (destination) side of the syslog-ng connections, server-side (source) connections are always reopened after receiving a HUP signal unless the keep-alive option is enabled for the source.

so-broadcast()
Type: yes or no
Default: no

Description: This option controls the SO_BROADCAST socket option required to make syslog-ng send messages to a broadcast address. For details, see the socket(7) manual page.

so-keepalive()
Type: yes or no
Default: no

Description: Enables keep-alive messages, keeping the socket open. This only effects TCP and UNIX-stream sockets. For details, see the socket(7) manual page.

mark-freq()
Accepted values: number (seconds)
Default: 1200

Description: An alias for the obsolete mark() option, retained for compatibility with syslog-ng version 1.6.x. The number of seconds between two MARK messages. MARK messages are generated when there was no message traffic to inform the receiver that the connection is still alive. If set to zero (0), no MARK messages are sent. The mark-freq() can be set for global option and/or every MARK capable destination driver if mark-mode() is periodical or dst-idle or host-idle. If mark-freq() is not defined in the destination, then the mark-freq() will be inherited from the global options. If the destination uses internal mark-mode(), then the global mark-freq() will be valid (does not matter what mark-freq() set in the destination side).

mark-mode()
Accepted values: internal | dst-idle | host-idle | periodical | none | global
Default:

internal for pipe, program drivers

none for file, unix-dgram, unix-stream drivers

global for syslog, tcp, udp destinations

host-idle for global option

Description: The mark-mode() option can be set for the following destination drivers: file(), program(), unix-dgram(), unix-stream(), network(), pipe(), syslog() and in global option.

  • internal: When internal mark mode is selected, internal source should be placed in the log path as this mode does not generate mark by itself at the destination. This mode only yields the mark messages from internal source. This is the mode as syslog-ng PE 3.x worked. MARK will be generated by internal source if there was NO traffic on local sources:

    file(), pipe(), unix-stream(), unix-dgram(), program()

  • dst-idle: Sends MARK signal if there was NO traffic on destination drivers. MARK signal from internal source will be dropped.

    MARK signal can be sent by the following destination drivers: network(), syslog(), program(), file(), pipe(), unix-stream(), unix-dgram().

  • host-idle: Sends MARK signal if there was NO local message on destination drivers. For example MARK is generated even if messages were received from tcp. MARK signal from internal source will be dropped.

    MARK signal can be sent by the following destination drivers: network(), syslog(), program(), file(), pipe(), unix-stream(), unix-dgram().

  • periodical: Sends MARK signal perodically, regardless of traffic on destination driver. MARK signal from internal source will be dropped.

    MARK signal can be sent by the following destination drivers: network(), syslog(), program(), file(), pipe(), unix-stream(), unix-dgram().

  • none: Destination driver drops all MARK messages. If an explicit mark-mode() is not given to the drivers where none is the default value, then none will be used.

  • global: Destination driver uses the global mark-mode() setting. The syslog-ng interprets syntax error if the global mark-mode() is global.

NOTE:

In case of dst-idle, host-idle and periodical, the MARK message will not be written in the destination, if it is not open yet.

Available in syslog-ng PE 4 LTS and later.

so-rcvbuf()
Type: number (bytes)
Default: 0

Description: Specifies the size of the socket receive buffer in bytes. For details, see the socket(7) manual page.

so-sndbuf()
Type: number (bytes)
Default: 0

Description: Specifies the size of the socket send buffer in bytes. For details, see the socket(7) manual page.

suppress()
Type: seconds
Default: 0 (disabled)

Description: If several identical log messages would be sent to the destination without any other messages between the identical messages (for example, an application repeated an error message ten times), syslog-ng can suppress the repeated messages and send the message only once, followed by the Last message repeated n times. message. The parameter of this option specifies the number of seconds syslog-ng waits for identical messages.

template()
Type: string
Default: A format conforming to the default logfile format.

Description: Specifies a template defining the logformat to be used in the destination. Macros are described in the section called “Macros of syslog-ng PE”. Please note that for network destinations it might not be appropriate to change the template as it changes the on-wire format of the syslog protocol which might not be tolerated by stock syslog receivers (like syslogd or syslog-ng itself). For network destinations make sure the receiver can cope with the custom format defined.

template-escape()
Type: yes or no
Default: no

Description: Turns on escaping for the ', ", and backspace characters in templated output files. This is useful for generating SQL statements and quoting string contents so that parts of the log message are not interpreted as commands to the SQL server.

throttle()
Type: number (messages per second)
Default: 0

Description: Sets the maximum number of messages sent to the destination per second. Use this output-rate-limiting functionality only when using disk-buffer as well to avoid the risk of losing messages. Specifying 0 or a lower value sets the output limit to unlimited.

time-zone()
Type: name of the timezone, or the timezone offset
Default: unspecified

Description: Convert timestamps to the timezone specified by this option. If this option is not set, then the original timezone information in the message is used. Converting the timezone changes the values of all date-related macros derived from the timestamp, for example, HOUR. For the complete list of such macros, see the section called “Date-related macros”.

The timezone can be specified as using the name of the (for example time-zone("Europe/Budapest")), or as the timezone offset in +/-HH:MM format (for example +01:00). On Linux and UNIX platforms, the valid timezone names are listed under the /usr/share/zoneinfo directory.

ts-format()
Type: rfc3164, bsd, rfc3339, iso
Default: Use the global option (which defaults to rfc3164)

Description: Override the global timestamp format (set in the global ts-format() parameter) for the specific destination. For details, see the section called “A note on timezones and timestamps”.

Sending messages to a user terminal — usertty() destination

This driver writes messages to the terminal of a logged-in user.

The usertty() driver has a single required argument, specifying a username who should receive a copy of matching messages. Use the asterisk * to specify every user currently logged in to the system.

Declaration: 

usertty(username);

The usertty() does not have any further options nor does it support templates.

Example 7.47. Using the usertty() driver

destination d_usertty { usertty("root"); };

Chapter 8. Routing messages: log paths, reliability, and filters

Log paths

Log paths determine what happens with the incoming log messages. Messages coming from the sources listed in the log statement and matching all the filters are sent to the listed destinations.

To define a log path, add a log statement to the syslog-ng configuration file using the following syntax:

log {
    source(s1); source(s2); ...
    optional_element(filter1|parser1|rewrite1);
    optional_element(filter2|parser2|rewrite2);
    ...
    destination(d1); destination(d2); ...
    flags(flag1[, flag2...]);
};

Caution:

Log statements are processed in the order they appear in the configuration file, thus the order of log paths may influence what happens to a message, especially when using filters and log flags.

NOTE:

The order of filters, rewriting rules, and parsers in the log statement is important, as they are processed sequentially.

Example 8.1. A simple log statement

The following log statement sends all messages arriving to the localhost to a remote server.

source s_localhost { network(ip(127.0.0.1) port(1999)); };
destination d_tcp { network("10.1.2.3" port(1999) localport(999)); };
log { source(s_localhost); destination(d_tcp); };

All matching log statements are processed by default, and the messages are sent to every matching destination by default. So a single log message might be sent to the same destination several times, provided the destination is listed in several log statements, and it can be also sent to several different destinations.

This default behavior can be changed using the flags() parameter. Flags apply to individual log paths, they are not global options. For details and examples on the available flags, see the section called “Log path flags”. The effect and use of the flow-control flag is detailed in the section called “Managing incoming and outgoing messages with flow-control”.

Embedded log statements

Starting from version 3.0, syslog-ng can handle embedded log statements (also called log pipes). Embedded log statements are useful for creating complex, multi-level log paths with several destinations and use filters, parsers, and rewrite rules.

For example, if you want to filter your incoming messages based on the facility parameter, and then use further filters to send messages arriving from different hosts to different destinations, you would use embedded log statements.

Figure 8.1. Embedded log statement

Embedded log statement

Embedded log statements include sources — and usually filters, parsers, rewrite rules, or destinations — and other log statements that can include filters, parsers, rewrite rules, and destinations. The following rules apply to embedded log statements:

  • Only the beginning (also called top-level) log statement can include sources.

  • Embedded log statements can include multiple log statements on the same level (that is, a top-level log statement can include two or more log statements).

  • Embedded log statements can include several levels of log statements (that is, a top-level log statement can include a log statement that includes another log statement, and so on).

  • After an embedded log statement, you can write either another log statement, or the flags() option of the original log statement. You cannot use filters or other configuration objects. This also means that flags (except for the flow-control flag) apply to the entire log statement, you cannot use them only for the embedded log statement.

  • Embedded log statements that are on the same level receive the same messages from the higher-level log statement. For example, if the top-level log statement includes a filter, the lower-level log statements receive only the messages that pass the filter.

Figure 8.2. Embedded log statements

Embedded log statements

Embedded log filters can be used to optimize the processing of log messages, for example, to re-use the results of filtering and rewriting operations.

Using embedded log statements

Embedded log statements (for details, see the section called “Embedded log statements”) re-use the results of processing messages (for example the results of filtering or rewriting) to create complex log paths. Embedded log statements use the same syntax as regular log statements, but they cannot contain additional sources. To define embedded log statements, use the following syntax:

log {
    source(s1); source(s2);...

    optional_element(filter1|parser1|rewrite1);
    optional_element(filter2|parser2|rewrite2);
    ...

    destination(d1); destination(d2); ...

    #embedded log statement
    log {
        optional_element(filter1|parser1|rewrite1);
        optional_element(filter2|parser2|rewrite2);
        ...
        destination(d1); destination(d2); ...
        #another embedded log statement
        log {
            optional_element(filter1|parser1|rewrite1);
            optional_element(filter2|parser2|rewrite2);
            ...
            destination(d1); destination(d2);
            ...
        };
    };
    #set flags after the embedded log statements
    flags(flag1[, flag2...]);
};

Example 8.2. Using embedded log paths

The following log path sends every message to the configured destinations: both the d_file1 and the d_file2 destinations receive every message of the source.

log { source(s_localhost); destination(d_file1); destination(d_file2); };

The next example is equivalent with the one above, but uses an embedded log statement.

log { source(s_localhost); destination(d_file1);
    log {destination(d_file2); };
};

The following example uses two filters:

  • messages coming from the host 192.168.1.1 are sent to the d_file1 destination, and

  • messages coming from the host 192.168.1.1 and containing the string example are sent to the d_file2 destination.

filter filter_host { host(192.168.1.1) };
filter filter_example { message("example") };
log { source(s_localhost); filter(filter_host); destination(d_file1);
    log {filter(filter_example); destination(d_file2); };
};

The following example collects logs from multiple source groups and uses the source() filter in the embedded log statement to select messages of the s_network source group.

log { source(s_localhost); source(s_network); destination(d_file1);
    log {source(s_network); destination(d_file2); };
};

Log path flags

Flags influence the behavior of syslog-ng, and the way it processes messages. The following flags may be used in the log paths, as described in the section called “Log paths”.

Table 8.1. Log statement flags

Flag Description
catchall This flag means that the source of the message is ignored, only the filters of the log path are taken into account when matching messages. A log statement using the catchall flag processes every message that arrives to any of the defined sources.
fallback

This flag makes a log statement 'fallback'. Fallback log statements process messages that were not processed by other, 'non-fallback' log statements.

'Processed' means that every filter of a log path matched the message. Note that in case of embedded log paths, the message is considered to be processed if it matches the filters of the outer log path, even if it does not match the filters of the embedded log path. For details, see Example 8.3, “Using log path flags”.

final

This flag means that the processing of log messages processed by the log statement ends here, other log statements appearing later in the configuration file will not process the messages processed by the log statement labeled as 'final'. Note that this does not necessarily mean that matching messages will be stored only once, as there can be matching log statements processed before the current one (syslog-ng PE evaluates log statements in the order they appear in the configuration file).

'Processed' means that every filter of a log path matched the message. Note that in case of embedded log paths, the message is considered to be processed if it matches the filters of the outer log path, even if it does not match the filters of the embedded log path. For details, see Example 8.3, “Using log path flags”.

flow-control Enables flow-control to the log path, meaning that syslog-ng will stop reading messages from the sources of this log statement if the destinations are not able to process the messages at the required speed. If disabled, syslog-ng will drop messages if the destination queues are full. If enabled, syslog-ng will only drop messages if the destination queues/window sizes are improperly sized. For details, see the section called “Managing incoming and outgoing messages with flow-control”.

Caution:

The final, fallback, and catchall flags apply only for the top-level log paths, they have no effect on embedded log paths.

Example 8.3. Using log path flags

Let's suppose that you have two hosts (myhost_A and myhost_B) that run two applications each (application_A and application_B), and you collect the log messages to a central syslog-ng server. On the server, you create two log paths:

  • one that processes only the messages sent by myhost_A, and

  • one that processes only the messages sent by application_A.

This means that messages sent by application_A running on myhost_A will be processed by both log paths, and the messages of application_B running on myhost_B will not be processed at all.

  • If you add the final flag to the first log path, then only this log path will process the messages of myhost_A, so the second log path will receive only the messages of application_A running on myhost_B.

  • If you create a third log path that includes the fallback flag, it will process the messages not processed by the first two log paths, in this case, the messages of application_B running on myhost_B.

  • Adding a fourth log path with the catchall flag would process every message received by the syslog-ng server.

    log { source(s_localhost); destination(d_file); flags(catchall); };

The following example shows a scenario that can result in message loss. Do NOT use such a configuration, unless you know exactly what you are doing. The problem is if a message matches the filters in the first part of the first log path, syslog-ng PE treats the message as 'processed'. Since the first log path includes the final flag, syslog-ng PE will not pass the message to the second log path (the one with the fallback flag). As a result, syslog-ng PE drops messages that do not match the filter of the embedded log path.

# Do not use such a configuration, unless you know exactly what you are doing.
log {
    source(s_network);
    # Filters in the external log path.
    # If a message matches this filter, it is treated as 'processed'
    filter(f_program);
    filter(f_message);
    log {
        # Filter in the embedded log path.
        # If a message does not match this filter, it is lost, it will not be processed by the 'fallback' log path
        filter(f_host);
        destination(d_file1);
    };
    flags(final);
};

log {
    source(s_network);
    destination(d_file2);
    flags(fallback);
};

Managing incoming and outgoing messages with flow-control

This section describes the internal message-processing model of syslog-ng, as well as the flow-control feature that can prevent message losses. To use flow-control, the flow-control flag must be enabled for the particular log path.

The syslog-ng application monitors (polls) the sources defined in its configuration file, periodically checking each source for messages. When a log message is found in one of the sources, syslog-ng polls every source and reads the available messages. These messages are processed and put into the output buffer of syslog-ng (also called fifo). From the output buffer, the operating system sends the messages to the appropriate destinations.

In large-traffic environments many messages can arrive during a single poll loop, therefore syslog-ng reads only a fixed number of messages from each source. The log-fetch-limit() option specifies the number of messages read during a poll loop from a single source.

Figure 8.3. Managing log messages in syslog-ng

Managing log messages in syslog-ng

NOTE:

The log-fetch-limit() parameter can be set as a global option, or for every source individually.

Every destination has its own output buffer. The output buffer is needed because the destination might not be able to accept all messages immediately. The log-fifo-size() parameter sets the size of the output buffer. The output buffer must be larger than the log-fetch-limit() of the sources, to ensure that every message read during the poll loop fits into the output buffer. If the log path sends messages to a destination from multiple sources, the output buffer must be large enough to store the incoming messages of every source.

TCP and unix-stream sources can receive the logs from several incoming connections (for example many different clients or applications). For such sources, syslog-ng reads messages from every connection, thus the log-fetch-limit() parameter applies individually to every connection of the source.

Figure 8.4. Managing log messages of TCP sources in syslog-ng

Managing log messages of TCP sources in syslog-ng

The flow-control of syslog-ng introduces a control window to the source that tracks how many messages can syslog-ng accept from the source. Every message that syslog-ng reads from the source lowers the window size by one, every message that syslog-ng successfully sends from the output buffer increases the window size by one. If the window is full (that is, its size decreases to zero), syslog-ng stops reading messages from the source. The initial size of the control window is by default 1000: the log-fifo-size() must be larger than this value in order for flow-control to have any effect. If a source accepts messages from multiple connections, all messages use the same control window.

NOTE:

If the source can handle multiple connections (for example, network()), the size of the control window is divided by the value of the max-connections() parameter and this smaller control window is applied to each connection of the source.

When flow-control is used, every source has its own control window. As a worst-case situation, the output buffer of the destination must be set to accommodate all messages of every control window, that is, the log-fifo-size() of the destination must be greater than number_of_sources*log-iw-size(). This applies to every source that sends logs to the particular destination. Thus if two sources having several connections and heavy traffic send logs to the same destination, the control window of both sources must fit into the output buffer of the destination. Otherwise, syslog-ng does not activate the flow-control, and messages may be lost.

The syslog-ng application handles outgoing messages the following way:

Figure 8.5. Handling outgoing messages in syslog-ng PE

Handling outgoing messages in syslog-ng PE

  • Output queue: Messages from the output queue are sent to the target syslog-ng server. The syslog-ng application puts the outgoing messages directly into the output queue, unless the output queue is full. The output queue can hold 64 messages, this is a fixed value and cannot be modified.

  • Disk buffer: If the output queue is full and disk-buffering is enabled, syslog-ng Premium Edition puts the outgoing messages into the disk buffer of the destination.

  • Overflow queue: If the output queue is full and the disk buffer is disabled or full, syslog-ng puts the outgoing messages into the overflow queue of the destination. (The overflow queue is identical to the output buffer used by other destinations.) The log-fifo-size() parameter specifies the number of messages stored in the overflow queue. For details on sizing the log-fifo-size() parameter, see the section called “Managing incoming and outgoing messages with flow-control”.

There are two types of flow-control: Hard flow-control and soft flow-control.

  • Soft flow-control: In case of soft flow-control there is no message lost if the destination can accept messages, but it is possible to lose messages if it cannot accept messages (for example non-writeable file destination, or the disk becomes full), and all buffers are full. Soft flow-control cannot be configured, it is automatically available for file and logstore destinations.

    Example 8.4. Soft flow-control

    source s_file {file("/tmp/input_file.log");};
    destination d_file {file("/tmp/output_file.log");};
    destination d_tcp { network("127.0.0.1" port(2222) log-fifo-size(1000)); };
    log{ source(s_file); destination(d_file); destination(d_tcp);};
    

    Caution:

    Hazard of data loss! For destinations other than file and logstore, soft flow-control is not available. Thus, it is possible to lose log messages on those destinations. To avoid data loss on those destinations, use hard flow-control.

  • Hard flow-control: In case of hard flow-control there is no message lost. To use hard flow-control, enable the flow-control flag in the log path. Hard flow-control is available for all destinations.

    Example 8.5. Hard flow-control

    source s_file {file("/tmp/input_file.log");};
    destination d_file {file("/tmp/output_file.log");};
    destination d_tcp { network("127.0.0.1" port(2222) log-fifo-size(1000)); };
    log{ source(s_file); destination(d_file); destination(d_tcp) flags(flow-control);};
    

Flow-control and multiple destinations

Using flow-control on a source has an important side-effect if the messages of the source are sent to multiple destinations. If flow-control is in use and one of the destinations cannot accept the messages, the other destinations do not receive any messages either, because syslog-ng stops reading the source. For example, if messages from a source are sent to a remote server and also stored locally in a file, and the network connection to the server becomes unavailable, neither the remote server nor the local file will receive any messages.

NOTE:

Creating separate log paths for the destinations that use the same flow-controlled source does not avoid the problem.

If you use flow-control and reliable disk-based buffering together with multiple destinations, the flow-control starts slowing down the source only when:

  • one destination is down, and

  • the number of messages stored in the disk buffer of the destination reaches (disk-buf-size() minus mem-buf-size()).

Configuring flow-control

For details on how flow-control works, see the section called “Managing incoming and outgoing messages with flow-control”. The summary of the main points is as follows:

  • The syslog-ng application normally reads a maximum of log-fetch-limit() number of messages from a source.

  • From TCP and unix-stream sources, syslog-ng reads a maximum of log-fetch-limit() from every connection of the source. The number of connections to the source is set using the max-connections() parameter.

  • Every destination has an output buffer (log-fifo-size()).

  • Flow-control uses a control window to determine if there is free space in the output buffer for new messages. Every source has its own control window, the log-iw-size() parameter sets the size of the control window.

  • When a source accepts multiple connections, the size of the control window is divided by the value of the max-connections() parameter and this smaller control window is applied to each connection of the source.

  • The output buffer must be larger than the control window of every source that logs to the destination.

  • If the control window is full, syslog-ng stops reading messages from the source until some messages are successfully sent to the destination.

  • If the output buffer becomes full, and neither disk-buffering nor flow-control is used, messages may be lost.

Caution:

If you modify the max-connections() or the log-fetch-limit() parameter, do not forget to adjust the log-iw-size() and log-fifo-size() parameters accordingly.

Example 8.6. Sizing parameters for flow-control

Suppose that syslog-ng has a source that must accept up to 300 parallel connections. Such situation can arise when a network source receives connections from many clients, or if many applications log to the same socket. Therefore, set the max-connections() parameter of the source to 300. However, the log-fetch-limit() (default value: 10) parameter applies to every connection of the source individually, while the log-iw-size() (default value: 1000) parameter applies to the source. In a worst-case scenario, the destination does not accept any messages, while all 300 connections send at least log-fetch-limit() number of messages to the source during every poll loop. Therefore, the control window must accommodate at least max-connections()*log-fetch-limit() messages to be able to read every incoming message of a poll loop. In the current example this means that (log-iw-size() should be greater than 300*10=3000. If the control window is smaller than this value, the control window might fill up with messages from the first connections — causing syslog-ng to read only one message of the last connections in every poll loop.

The output buffer of the destination must accommodate at least log-iw-size() messages, but use a greater value: in the current example 3000*10=30000 messages. That way all incoming messages of ten poll loops fit in the output buffer. If the output buffer is full, syslog-ng does not read any messages from the source until some messages are successfully sent to the destination.

source s_localhost {
            network(ip(127.0.0.1) port(1999) max-connections(300)); };
destination d_tcp {
            network("10.1.2.3" port(1999); localport(999)); log-fifo-size(30000); };
log { source(s_localhost); destination(d_tcp); flags(flow-control); };

If other sources send messages to this destination, than the output buffer must be further increased. For example, if a network host with maximum 100 connections also logs into the destination, than increase the log-fifo-size() by 10000.

source s_localhost {
            network(ip(127.0.0.1) port(1999) max-connections(300)); };
source s_tcp {
            network(ip(192.168.1.5) port(1999) max-connections(100)); };
destination d_tcp {
            network("10.1.2.3" port(1999); localport(999)); log-fifo-size(40000); };
log { source(s_localhost); destination(d_tcp); flags(flow-control); };

Related Documents