了解流分析作业监视以及如何监视查询Understand Stream Analytics job monitoring and how to monitor queries

简介:监视页Introduction: The monitor page

Azure 门户提供了可用于监视和排查查询和作业性能问题的关键性能指标。The Azure portal surfaces key performance metrics that can be used to monitor and troubleshoot your query and job performance. 若要查看这些指标,请浏览到想要查看其指标的流分析作业并查看“概览”页上的“监视” 部分。To see these metrics, browse to the Stream Analytics job you are interested in seeing metrics for and view the Monitoring section on the Overview page.


此窗口如下所示:The window will appear as shown:


可用于流分析的指标Metrics available for Stream Analytics

指标Metric 定义Definition
积压的输入事件数Backlogged Input Events 积压的输入事件的数量。Number of input events that are backlogged. 此指标的非零值意味着作业无法跟上传入事件的数量。A non-zero value for this metric implies that your job isn't able to keep up with the number of incoming events. 如果此值缓慢增长或始终为非零,则应横向扩展作业。If this value is slowly increasing or consistently non-zero, you should scale out your job. 可以访问了解和调整流单元了解详细信息。You can learn more by visiting Understand and adjust Streaming Units.
数据转换错误数Data Conversion Errors 无法转换为预期输出架构的输出事件的数量。Number of output events that could not be converted to the expected output schema. 可以将错误策略更改为“删除”,以删除遇到此情况的事件。Error policy can be changed to 'Drop' to drop events that encounter this scenario.
早期输入事件数Early Input Events 应用程序时间戳早于其到达时间超过 5 分钟的事件。Events whose application timestamp is earlier than their arrival time by more than 5 minutes.
失败的函数请求数Failed Function Requests 失败的 Azure 机器学习函数(如果存在)调用数。Number of failed Azure Machine Learning function calls (if present).
函数事件数Function Events 发送到 Azure 机器学习函数(如果存在)的事件数。Number of events sent to the Azure Machine Learning function (if present).
函数请求数Function Requests Azure 机器学习函数(如果存在)的调用数。Number of calls to the Azure Machine Learning function (if present).
输入反序列化错误Input Deserialization Errors 不可反序列化的输入事件数。Number of input events that could not be deserialized.
输入事件字节数Input Event Bytes 流分析作业收到的数据量(以字节为单位)。Amount of data received by the Stream Analytics job, in bytes. 这可以用于验证正在发送到输入源的事件。This can be used to validate that events are being sent to the input source.
输入事件数Input Events 从输入事件反序列化的记录数。Number of records deserialized from the input events. 此计数不包括导致反序列化错误的传入事件。This count does not include incoming events that result in deserialization errors. 在内部恢复和自联接等方案中,流分析可以多次引入相同的事件。The same events can be ingested by Stream Analytics multiple times in scenarios such as internal recoveries and self joins. 因此,如果作业包含简单的“传递”查询,则建议不要预期输入事件与输出事件的指标相匹配。Therefore it is recommended not to expect Input Events and Output Events metrics to match if your job has a simple 'pass through' query.
收到的输入源数Input Sources Received 作业收到的消息数。Number of messages received by the job. 对于事件中心,消息是单个 EventData。For Event Hub, a message is a single EventData. 对于 Blob,消息是单个 Blob。For Blob, a message is a single blob. 请注意,输入源在反序列化之前不计数。Please note that Input Sources are counted before deserialization. 如果存在反序列化错误,则输入源数可能大于输入事件数。If there are deserialization errors, input sources can be greater than input events. 否则,它可能小于或等于输入事件数,因为每条消息可能包含多个事件。Otherwise, it can be less than or equal to input events since each message can contain multiple events.
延迟输入事件数Late Input Events 到达时间晚于已配置的延迟到达容错时段的事件。Events that arrived later than the configured late arrival tolerance window. 详细了解 Azure 流分析事件顺序注意事项Learn more about Azure Stream Analytics event order considerations .
无序事件数Out-of-Order Events 收到的无序事件的数目,系统根据事件排序策略来删除这些事件,或者为其提供一个经过调整的时间戳。Number of events received out of order that were either dropped or given an adjusted timestamp, based on the Event Ordering Policy. 这可能会受“无序容错时段”设置的影响。This can be impacted by the configuration of the Out of Order Tolerance Window setting.
输出事件数Output Events 流分析作业发送到输出目标的数据量,以事件计数来衡量。Amount of data sent by the Stream Analytics job to the output target, in number of events.
运行时错误Runtime Errors 与查询处理相关的错误总数(不包括引入事件或输出结果时发现的错误)Total number of errors related to query processing (excluding errors found while ingesting events or outputting results)
流单元利用率 %SU % Utilization 如果资源利用率持续超过 80%,则水印延迟增加,积压的事件数增加,请考虑增加流单元。If resource utilization is consistently over 80%, the watermark delay is rising, and the number of backlogged events is rising, consider increasing streaming units. 高利用率指示作业使用的资源数接近分配的最大资源数。High utilization indicates that the job is using close to the maximum allocated resources.
水印延迟Watermark Delay 作业中所有输出的所有分区之间的最大水印延迟。The maximum watermark delay across all partitions of all outputs in the job.

可以使用这些指标来监视流分析作业的性能You can use these metrics to monitor the performance of your Stream Analytics job.

在 Azure 门户中自定义监视Customizing Monitoring in the Azure portal

可以在“编辑图表”设置中调整图表类型、显示的指标和时间范围。You can adjust the type of chart, metrics shown, and time range in the Edit Chart settings. 有关详细信息,请参阅如何自定义监视For details, see How to Customize Monitoring.


最新输出Latest output

对作业进行监视时需要关注的另一个数据点是最后的输出的时间(显示在“概述”页面中)。Another interesting data point to monitor your job is the time of the last output, shown in the Overview page. 此时间是作业的最新输出的应用程序时间(即,使用来自事件数据的时间戳的时间)。This time is the application time (i.e. the time using the timestamp from the event data) of the latest output of your job.

获取帮助Get help

如需进一步的帮助,请参阅有关 Azure 流分析的 Microsoft 问答页For further assistance, try our Microsoft Q&A question page for Azure Stream Analytics

后续步骤Next steps