Struct twitch_irc::ClientConfig
source · pub struct ClientConfig<L: LoginCredentials> {
pub login_credentials: L,
pub max_channels_per_connection: usize,
pub max_waiting_messages_per_connection: usize,
pub time_per_message: Duration,
pub connection_rate_limiter: Arc<Semaphore>,
pub new_connection_every: Duration,
pub connect_timeout: Duration,
pub metrics_config: MetricsConfig,
pub tracing_identifier: Option<Cow<'static, str>>,
}
Expand description
Configures settings for a TwitchIRCClient
.
Fields§
§login_credentials: L
Gets a set of credentials every time the client needs to log in on a new connection.
See LoginCredentials
for details.
max_channels_per_connection: usize
A new connection will automatically be created if a channel is joined and all currently established connections have joined at least this many channels.
max_waiting_messages_per_connection: usize
A new connection will automatically be created if any message is to be sent
and all currently established connections have recently sent more than this many
messages (time interval is defined by max_waiting_messages_duration_window
)
time_per_message: Duration
We assume messages to be “waiting” for this amount of time after sending them out, e.g. typically 100 or 150 milliseconds (purely a value that has been measured/observed, not documented or fixed in any way)
connection_rate_limiter: Arc<Semaphore>
rate-limits the opening of new connections. By default this is constructed with 1 permit only, which means connections cannot be opened in parallel. If this is set to more than 1 permit, then that many connections can be opened in parallel.
This is designed to be wrapped in an Arc to allow it to be shared between multiple TwitchIRCClient instances.
new_connection_every: Duration
Allow a new connection to be made after this period has elapsed. By default this is set to 2 seconds, and combined with the permits=1 of the semaphore, allows one connection to be made every 2 seconds.
More specifically, after taking the permit from the semaphore, the permit will be put back after this period has elapsed.
connect_timeout: Duration
Imposes a general timeout for new connections. This is in place in addition to possible operating system timeouts (E.g. for new TCP connections), since additional “connect” work takes place after the TCP connection is opened, e.g. to set up TLS or perform a WebSocket handshake. Default value: 20 seconds.
metrics_config: MetricsConfig
Disable or enable and configure the collection of metrics on this TwitchIRCClient
using the prometheus
crate. See more information about the possible options on the
MetricsConfig
enum.
This crate is currently capable of exporting the following prometheus metrics:
-
twitchirc_messages_received
with labelcommand
counts all incoming messages. (Counter) -
twitchirc_messages_sent
counts messages sent out, with acommand
label. (Counter) -
twitchirc_channels
withtype=allocated/confirmed
counts how many channels you are joined to (Gauge). Allocated channels are joins that passed through theTwitchIRCClient
but may be waiting e.g. for the connection to finish connecting. Once a confirmation response is received by Twitch that the channel was joined successfully, that channel is additionallyconfirmed
. -
twitchirc_connections
counts how many connections this client has in use (Gauge). The labelstate=initializing/open
identifies how many connections are in the process of connecting (initializing
) vs how many connections are already established (open
). -
twitchirc_connections_failed
counts every time a connection fails (Counter). Note however, depending on conditions e.g. how many channels were joined on that channel, there can be cases where a connection failing would not mandate the creation of a new connection (e.g. if you have parted channels on other connections, making it so all the channels the failed connection was joined to can be re-joined on those already existing connections). -
twitchirc_connections_created
on the other hand tracks how many times, since the creation of the client, a new connection has been made.
tracing_identifier: Option<Cow<'static, str>>
Allows you to differentiate between multiple clients with
the tracing
crate.
This library logs a variety of trace, debug, info, warning and error messages using the
tracing
crate. An example log line using the default tracing_subscriber
output format
might look like this:
2022-02-07T10:44:23.297571Z INFO client_loop: twitch_irc::client::event_loop: Making a new pool connection, new ID is 0
You may optionally set this configuration variable to some string, which will then
modify all log messages by giving the client_loop
span the name
attribute:
use std::borrow::Cow;
use twitch_irc::ClientConfig;
let mut config = ClientConfig::default();
config.tracing_identifier = Some(Cow::Borrowed("bot_one"));
All log output will then look like this (note the additional {name=bot_one}
:
2022-02-07T10:48:34.769272Z INFO client_loop{name=bot_one}: twitch_irc::client::event_loop: Making a new pool connection, new ID is 0
Essentially, this library makes use of tracing
Span
s to differentiate between
different async tasks and to also differentiate log messages coming from different
connections.
Specifying this option will further allow you to differentiate between multiple
clients if your application is running multiple of them. It will add the name=your_value
attribute to the client_loop
span, which is root for all further deeper spans in the
client. This means that all log output from a single client will all be under that span,
with that name.
Implementations§
source§impl<L: LoginCredentials> ClientConfig<L>
impl<L: LoginCredentials> ClientConfig<L>
sourcepub fn new_simple(login_credentials: L) -> ClientConfig<L>
pub fn new_simple(login_credentials: L) -> ClientConfig<L>
Create a new configuration from the given login credentials, with all other configuration options being default.