The Path Manager is in charge of subflows, from creation to deletion, and also address announcements. Typically, it is the client side that initiates subflows, and the server side that announces additional addresses via the ADD_ADDR and REMOVE_ADDR options.

graph LR;
    C_1(<div style="display: inline-block; min-width: 35px"><font size="7">fa:fa-mobile</font></div>)
    S_1((<div style="display: inline-block; min-width: 60px"><font size="7">fa:fa-cloud</font></div>))

    C_1 -. "Potential subflow" -.- S_1
    C_1 <== "Initial subflow" ==> S_1
    C_1 ~~~|"Subflows creation"| C_1
    S_1 ~~~|"Address announcement"| S_1

    linkStyle 0 stroke:orange;
    linkStyle 1 stroke:green;

As of Linux v5.19, there are two path managers controlled by the netns-aware net.mptcp.pm_type sysctl knob: the in-kernel one (type 0), and the userspace one (type 1).

In-kernel Path-Manager

With the In-kernel Path-Manager, the same rules are applied to all connections. Address endpoints and limits can be set to control its behavior.


This configuration can be automated with tools like Network Manager – in command lines, look for mptcp-flags in the settings – and mptcpd. Here, the focus is on the manual configuration, using the ip mptcp command.


MPTCP endpoints can be configured with this command:

ip mptcp endpoint add <IP address> dev <interface> [ signal | subflow ] [ backup ] [ fullmesh ]

It is important to specify the network interface linked to the address by adding dev <interface>. If not, the routing will probably not be done properly, and will require manual configuration, see below: Manual Routing Configuration.

One of the following flags needs to be set:

  • signal: The endpoint will be announced to each peer via an MPTCP ADD_ADDR sub-option. Typically, a server would be responsible for this.
  • subflow: The endpoint will be used to create an additional subflow using the given source IP address. A client would typically do this.

Optionally, the following flags can be added next to one of the previous ones:

  • backup: Subflows created from this endpoint instruct the peers to only send data on it when all non-backup subflows are unavailable.
  • fullmesh: The MPTCP path manager will try to create an additional subflow for each known peer address, using this endpoint as the source IP address. It requires the subflow flag, and it is not compatible with the signal one.

The IP address is an IPv4 or IPv6 address. The endpoints are netns-aware.


  • Servers can announce extra IP addresses:
    ip mptcp endpoint add dev eth0 signal
  • Clients can create additional subflows from a cellular interface, and flag this subflow as “backup”, to be used to carry data only if the main path is unavailable:
    ip mptcp endpoint add dev usb0 subflow backup


It is also important to make sure the limits are high enough:

ip mptcp limits set [ subflows NR ] [ add_addr_accepted NR ]
  • subflows is the limit of additional created and accepted subflows (paths), for both the client and server sides (default is 2).
  • add_addr_accepted is the limit of accepted ADD_ADDR – IP address notification from the other peer – that will result in the creation of subflows, typically only for the client side (default is 0).

The limits are per MPTCP connection, and netns-aware.

It is possible to reach the limits with fewer established subflows than expected, e.g. when new subflow requests cannot reach the other peer. In case of problem, please increase the limits, use ss -Mai to check the counters, and modify the routing or firewall rules to avoid using certain paths between specific IP addresses. For example, in a lab setup with dedicated links, use specific routes rather than letting the kernel select the default route.


By default, the Path-Manager tries not to create too many subflows using the encoded endpoints, and following the imposed limits.

Technically, the Path-Manager can only attach new subflows, and announce addresses on MPTCP connections that are considered as fully established: typically when the first data has been sent.

Announcing new addresses

All endpoints flagged as signal will be announced via an ADD_ADDR notification.

This will be done, one at a time: once the MPTCP connection is fully established, and each time a new subflow is established. This behaviour might change in the future, if someone implements the ticket #334.

Creating new subflows

There are two cases that involve the creation of new subflows, if allowed by the subflows limit:

  • Endpoints flagged as subflow will be used to create new subflows, one at a time, to the address of the server used in the initial subflow. Each endpoint should only be used once, except if the endpoint has the fullmesh flag.

  • Upon the reception of an ADD_ADDR, and if the add_addr_accepted limits has not been reached yet, a new subflow will be created using the local address the routing configuration will pick, except if there are endpoints with the fullmesh flag. In this case, each endpoints with the fullmesh flag will be used to create a new subflow to the announced address.

Note that when subflows are closed before the end of a connection – e.g. due to an error on the network, or if the other peer closed subflows – the Path-Manager will not try to re-establish them. This behaviour might change in the future, if someone implements the ticket #440.

Accepting new subflows

If the request is valid, the peer will accept the creation of new subflows, as long as the subflows limit has not been reached yet.

Userspace Path-Manager

With the userspace MPTCP path-manager – sysctl net.mptcp.pm_type=1 – different rules can be applied for each connection. The path-manager will then need to be controlled by a userspace daemon, i.e. mptcpd. In this case, the configuration has to be done on the userspace daemon side.

mptcpd can help to create custom userspace Path-Managers: please check this Plugins page for more details about that.