Chyun

大神, 模型训练策略能详细说一下吗?

I can use "search source=*210324*|fields event_duration, inbuilt_meta.log_timestamp|eval day_str=date_format(inbuilt_meta.log_timestamp, 'yyyy-MM-dd HH')" to get the right answer, but the method do not aggregate all of data, it just aggregate the head part...

创建producer时, 加上刷新时间, `local prod = producer:new(broker_list, { producer_type = "async", refresh_interval=10000})` 默认是不刷新的, 当broker故障时, 会刷新一次元数据, 但是当broker恢复以后, 客户端是不会自动刷新元数据的。

* Organization: shulidata * Location: Hangzhou, China * Contact: [email protected] * Scenario: for big data platform ETL processing --- * 组织：shulidata * 地点：中国杭州 * 联系方式：[email protected] * 场景：用于大数据平台ETL处理