对频繁事件的速率限制日志

我们经常想要在事件发生时告知用户，但这些事件可能非常频繁。在这种情况下，我们不希望每小时记录一千次“出现了问题！”，相反，我们希望日志看起来更像是“连接到哨兵X：错误（在过去一小时中发生1310/2000次）”。

该crate是Arti的一部分，不适用于外部使用：它假设您的日志系统是tracing，并且您正在使用tor_rtcompat进行异步运行时。

设置

在您可以使用此crate之前，您需要调用install_runtime，否则消息将不会被收集。

示例

use tor_log_ratelim::log_ratelim;
# use std::num::ParseIntError;
pub fn parse_u8(source: &str, s: &str) -> u8 {
    let r: Result<u8, ParseIntError> = s.parse();
    log_ratelim!(
        // The activity we were performing
        "parsing an integer from {}", source;
        // A Result to decide whether it succeeded 
        r; 
        // An error message to report on failure, with rate limiting,
        // after some time has elapsed.
        // The error itself is always reported.
        Err(_) => WARN, "Had to use default";
        // A success message to report (without rate limiting)
        // on every success.
        Ok(v) => TRACE, "Got {}", v;
    );
    r.unwrap_or(0)
}

上面的示例可能会产生类似以下这些的WARN输出

WARN: Parsing an integer from cache: error (Problem occurred 7/10 times in the last minute): Had to use default: invalid digit found in string"
WARN: Parsing an integer from cache: error (Problem occurred 81/92 times in the last 5 minutes): Had to use default: number too large to fit in target type"
WARN: Parsing an integer from cache: now working (Problem occurred 0/106 times in the last 5 minutes)

以及类似的TRACE输出

TRACE: Parsing an integer from cache: Got 7
TRACE: Parsing an integer from cache: Got 14
TRACE: Parsing an integer from the network: Got 2

有关log_ratelim!以及更简单的调用方式的信息，请参阅其文档。