Documentation Home
MySQL 8.0 Reference Manual
Related Documentation Download this Manual
PDF (US Ltr) - 43.3Mb
PDF (A4) - 43.4Mb
Man Pages (TGZ) - 297.2Kb
Man Pages (Zip) - 402.4Kb
Info (Gzip) - 4.3Mb
Info (Zip) - 4.3Mb
Excerpts from this Manual

25.4.3.10 NDB Cluster TCP/IP Connections

TCP/IP is the default transport mechanism for all connections between nodes in an NDB Cluster. Normally it is not necessary to define TCP/IP connections; NDB Cluster automatically sets up such connections for all data nodes, management nodes, and SQL or API nodes.

To override the default connection parameters, it is necessary to define a connection using one or more[tcp] sections in theconfig.ini file. Each[tcp] section explicitly defines a TCP/IP connection between two NDB Cluster nodes, and must contain at a minimum the parametersNodeId1 andNodeId2, as well as any connection parameters to override.

It is also possible to change the default values for these parameters by setting them in the[tcp default] section.

Important

Any[tcp] sections in theconfig.ini file should be listedlast, following all other sections in the file. However, this is not required for a[tcp default] section. This requirement is a known issue with the way in which theconfig.ini file is read by the NDB Cluster management server.

Connection parameters which can be set in[tcp] and[tcp default] sections of theconfig.ini file are listed here:

  • AllowUnresolvedHostNames

    Version (or later)NDB 8.0.22
    Type or unitsboolean
    Defaultfalse
    Rangetrue, false
    AddedNDB 8.0.22
    Restart Type

    Node Restart:Requires arolling restart of the cluster. (NDB 8.0.13)

    By default, when a management node fails to resolve a host name while trying to connect, this results in a fatal error. This behavior can be overridden by settingAllowUnresolvedHostNames totrue in the[tcp default] section of the global configuration file (usually namedconfig.ini), in which case failure to resolve a host name is treated as a warning andndb_mgmd startup continues uninterrupted.

  • Checksum

    Version (or later)NDB 8.0.13
    Type or unitsboolean
    Defaultfalse
    Rangetrue, false
    Restart Type

    Node Restart:Requires arolling restart of the cluster. (NDB 8.0.13)

    This parameter is disabled by default. When it is enabled (set toY or1), checksums for all messages are calculated before they placed in the send buffer. This feature ensures that messages are not corrupted while waiting in the send buffer, or by the transport mechanism.

  • Group

    Whenndb_optimized_node_selection is enabled, node proximity is used in some cases to select which node to connect to. This parameter can be used to influence proximity by setting it to a lower value, which is interpreted as“closer”. See the description of the system variable for more information.

  • HostName1

    Version (or later)NDB 8.0.13
    Type or unitsname or IP address
    Default[...]
    Range...
    Restart Type

    Node Restart:Requires arolling restart of the cluster. (NDB 8.0.13)

    TheHostName1 andHostName2 parameters can be used to specify specific network interfaces to be used for a given TCP connection between two nodes. The values used for these parameters can be host names or IP addresses.

  • HostName2

    Version (or later)NDB 8.0.13
    Type or unitsname or IP address
    Default[...]
    Range...
    Restart Type

    Node Restart:Requires arolling restart of the cluster. (NDB 8.0.13)

    TheHostName1 andHostName2 parameters can be used to specify specific network interfaces to be used for a given TCP connection between two nodes. The values used for these parameters can be host names or IP addresses.

  • NodeId1

    Version (or later)NDB 8.0.13
    Type or unitsnumeric
    Default[none]
    Range1 - 255
    Restart Type

    Node Restart:Requires arolling restart of the cluster. (NDB 8.0.13)

    To identify a connection between two nodes it is necessary to provide their node IDs in the[tcp] section of the configuration file as the values ofNodeId1 andNodeId2. These are the same uniqueId values for each of these nodes as described inSection 25.4.3.7, “Defining SQL and Other API Nodes in an NDB Cluster”.

  • NodeId2

    Version (or later)NDB 8.0.13
    Type or unitsnumeric
    Default[none]
    Range1 - 255
    Restart Type

    Node Restart:Requires arolling restart of the cluster. (NDB 8.0.13)

    To identify a connection between two nodes it is necessary to provide their node IDs in the[tcp] section of the configuration file as the values ofNodeId1 andNodeId2. These are the same uniqueId values for each of these nodes as described inSection 25.4.3.7, “Defining SQL and Other API Nodes in an NDB Cluster”.

  • NodeIdServer

    Version (or later)NDB 8.0.13
    Type or unitsnumeric
    Default[none]
    Range1 - 63
    Restart Type

    Node Restart:Requires arolling restart of the cluster. (NDB 8.0.13)

    Set the server side of a TCP connection.

  • OverloadLimit

    Version (or later)NDB 8.0.13
    Type or unitsbytes
    Default0
    Range0 - 4294967039 (0xFFFFFEFF)
    Restart Type

    Node Restart:Requires arolling restart of the cluster. (NDB 8.0.13)

    When more than this many unsent bytes are in the send buffer, the connection is considered overloaded.

    This parameter can be used to determine the amount of unsent data that must be present in the send buffer before the connection is considered overloaded. SeeSection 25.4.3.14, “Configuring NDB Cluster Send Buffer Parameters”, for more information.

  • PreferIPVersion

    Version (or later)NDB 8.0.26
    Type or unitsenumeration
    Default4
    Range4, 6
    AddedNDB 8.0.26
    Restart Type

    Initial System Restart:Requires a complete shutdown of the cluster, wiping and restoring the cluster file system from abackup, and then restarting the cluster. (NDB 8.0.13)

    Determines the preference of DNS resolution for IP version 4 or version 6. Because the configuration retrieval mechanism employed by NDB Cluster requires that all connections use the same preference, this parameter should be set in the[tcp default] of theconfig.ini global configuration file.

  • PreSendChecksum

    Version (or later)NDB 8.0.13
    Type or unitsboolean
    Defaultfalse
    Rangetrue, false
    Restart Type

    Node Restart:Requires arolling restart of the cluster. (NDB 8.0.13)

    If this parameter andChecksum are both enabled, perform pre-send checksum checks, and check all TCP signals between nodes for errors. Has no effect ifChecksum is not also enabled.

  • Proxy

    Version (or later)NDB 8.0.13
    Type or unitsstring
    Default[...]
    Range...
    Restart Type

    Node Restart:Requires arolling restart of the cluster. (NDB 8.0.13)

    Set a proxy for the TCP connection.

  • ReceiveBufferMemory

    Version (or later)NDB 8.0.13
    Type or unitsbytes
    Default2M
    Range16K - 4294967039 (0xFFFFFEFF)
    Restart Type

    Node Restart:Requires arolling restart of the cluster. (NDB 8.0.13)

    Specifies the size of the buffer used when receiving data from the TCP/IP socket.

    The default value of this parameter is 2MB. The minimum possible value is 16KB; the theoretical maximum is 4GB.

  • SendBufferMemory

    Version (or later)NDB 8.0.13
    Type or unitsunsigned
    Default2M
    Range256K - 4294967039 (0xFFFFFEFF)
    Restart Type

    Node Restart:Requires arolling restart of the cluster. (NDB 8.0.13)

    TCP transporters use a buffer to store all messages before performing the send call to the operating system. When this buffer reaches 64KB its contents are sent; these are also sent when a round of messages have been executed. To handle temporary overload situations it is also possible to define a bigger send buffer.

    If this parameter is set explicitly, then the memory is not dedicated to each transporter; instead, the value used denotes the hard limit for how much memory (out of the total available memory—that is,TotalSendBufferMemory) that may be used by a single transporter. For more information about configuring dynamic transporter send buffer memory allocation in NDB Cluster, seeSection 25.4.3.14, “Configuring NDB Cluster Send Buffer Parameters”.

    The default size of the send buffer is 2MB, which is the size recommended in most situations. The minimum size is 64 KB; the theoretical maximum is 4 GB.

  • SendSignalId

    Version (or later)NDB 8.0.13
    Type or unitsboolean
    Defaultfalse (debug builds: true)
    Rangetrue, false
    Restart Type

    Node Restart:Requires arolling restart of the cluster. (NDB 8.0.13)

    To be able to retrace a distributed message datagram, it is necessary to identify each message. When this parameter is set toY, message IDs are transported over the network. This feature is disabled by default in production builds, and enabled in-debug builds.

  • TcpBind_INADDR_ANY

    Setting this parameter toTRUE or1 bindsIP_ADDR_ANY so that connections can be made from anywhere (for autogenerated connections). The default isFALSE (0).

  • TcpSpinTime

    Version (or later)NDB 8.0.20
    Type or unitsµsec
    Default0
    Range0 - 2000
    AddedNDB 8.0.20
    Restart Type

    Node Restart:Requires arolling restart of the cluster. (NDB 8.0.13)

    Controls spin for a TCP transporter; no enable, set to a nonzero value. This works for both the data node and management or SQL node side of the connection.

  • TCP_MAXSEG_SIZE

    Version (or later)NDB 8.0.13
    Type or unitsunsigned
    Default0
    Range0 - 2G
    Restart Type

    Node Restart:Requires arolling restart of the cluster. (NDB 8.0.13)

    Determines the size of the memory set during TCP transporter initialization. The default is recommended for most common usage cases.

  • TCP_RCV_BUF_SIZE

    Version (or later)NDB 8.0.13
    Type or unitsunsigned
    Default0
    Range0 - 2G
    Restart Type

    Node Restart:Requires arolling restart of the cluster. (NDB 8.0.13)

    Determines the size of the receive buffer set during TCP transporter initialization. The default and minimum value is 0, which allows the operating system or platform to set this value. The default is recommended for most common usage cases.

  • TCP_SND_BUF_SIZE

    Version (or later)NDB 8.0.13
    Type or unitsunsigned
    Default0
    Range0 - 2G
    Restart Type

    Node Restart:Requires arolling restart of the cluster. (NDB 8.0.13)

    Determines the size of the send buffer set during TCP transporter initialization. The default and minimum value is 0, which allows the operating system or platform to set this value. The default is recommended for most common usage cases.

Restart types.  Information about the restart types used by the parameter descriptions in this section is shown in the following table:

Table 25.21 NDB Cluster restart types

SymbolRestart TypeDescription
NNodeThe parameter can be updated using a rolling restart (seeSection 25.6.5, “Performing a Rolling Restart of an NDB Cluster”)
SSystemAll cluster nodes must be shut down completely, then restarted, to effect a change in this parameter
IInitialData nodes must be restarted using the--initial option