Runtime Configuration for CAPIO-CL

CAPIO-CL supports runtime configuration through a user-provided TOML configuration file.
This allows you to customize various aspects of the library without recompiling it.

A configuration file can be loaded into a running CAPIO-CL instance by calling:

capiocl::Engine engine;
engine.load("path/to/config.toml");

If no configuration file is provided, CAPIO-CL falls back to its built-in defaults.

---

TOML Configuration Structure

The configuration file uses a structured namespace under the top-level table monitor.mcast.

Available configuration parameters:

Key	Type	Default	Description
monitor.mcast.commit.ip	string	224.224.224.1	Multicast IP address used for commit messages
monitor.mcast.commit.port	integer	12345	UDP port for commit messages
monitor.mcast.delay_ms	integer	300	Artificial delay (in milliseconds) inserted before sending multicast messages. Useful for debugging or simulating slower networks.
monitor.mcast.homenode.ip	string	224.224.224.2	IP address of the home node for monitoring operations
monitor.mcast.homenode.port	integer	12345	Port associated with the home node monitoring endpoint

---

Example configuration file

Below is a complete example of a config.toml file:

# Example CAPIO-CL TOML configuration

[monitor.mcast]

# Multicast settings for commit messages
commit.ip   = "224.224.224.1"
commit.port = 12345

# Delay (in milliseconds)
delay_ms = 300

# Home node information
homenode.ip   = "224.224.224.2"
homenode.port = 12345

---

How CAPIO-CL Uses These Settings

commit.ip and commit.port

These fields define where CAPIO-CL sends commit broadcast messages.
Commit messages are used for consistency coordination across distributed nodes.

delay_ms

A small configurable delay may help with:

• debugging message-ordering issues,
• testing high-latency environments,
• reproducing specific timing-related behavior.

A value of 0 means no delay.

homenode.ip and homenode.port

These define the central monitoring endpoint (the “home node”).
CAPIO-CL uses this endpoint to coordinate monitoring metadata and cluster-wide communication.