我有一个Spark（2.1）流式传输作业，每个批次将处理后的数据写入azure blob存储（批处理间隔为1分钟）。每隔一段时间（每隔几个小时一次）我就会得到＆＃39; java.net.ConnectException＆＃39;连接超时消息。这确实会重试并最终成功。但是这个问题导致1分钟流批次的完成延迟，并且在发生此错误时导致它在2到3分钟内完成。
以下是带有错误消息的执行程序日志片段。我有spark.executor.cores = 5。
是否存在可能导致此问题的某种连接限制？
17/10/11 16:09:02 INFO root: {89f867cc-cbd3-4fa9-a549-4e07be3f69b0}: {Starting operation.}
17/10/11 16:09:02 INFO root: {89f867cc-cbd3-4fa9-a549-4e07be3f69b0}: {Starting operation with location 'PRIMARY' per location mode 'PRIMARY_ONLY'.}
17/10/11 16:09:02 INFO root: {89f867cc-cbd3-4fa9-a549-4e07be3f69b0}: {Starting request to 'http://<name>.blob.core.windows.net/rawData/2017/10/11/16/data1.json' at 'Wed, 11 Oct 2017 16:09:02 GMT'.}
17/10/11 16:09:02 INFO root: {89f867cc-cbd3-4fa9-a549-4e07be3f69b0}: {Waiting for response.}
..
17/10/11 16:11:09 WARN root: {89f867cc-cbd3-4fa9-a549-4e07be3f69b0}: {Retryable exception thrown. Class = 'java.net.ConnectException', Message = 'Connection timed out (Connection timed out)'.}
17/10/11 16:11:09 INFO root: {89f867cc-cbd3-4fa9-a549-4e07be3f69b0}: {Checking if the operation should be retried. Retry count = '0', HTTP status code = '-1', Error Message = 'An unknown failure occurred : Connection timed out (Connection timed out)'.}
17/10/11 16:11:09 INFO root: {89f867cc-cbd3-4fa9-a549-4e07be3f69b0}: {The next location has been set to 'PRIMARY', per location mode 'PRIMARY_ONLY'.}
17/10/11 16:11:09 INFO root: {89f867cc-cbd3-4fa9-a549-4e07be3f69b0}: {The retry policy set the next location to 'PRIMARY' and updated the location mode to 'PRIMARY_ONLY'.}
17/10/11 16:11:09 INFO root: {89f867cc-cbd3-4fa9-a549-4e07be3f69b0}: {Operation will be retried after '0'ms.}
17/10/11 16:11:09 INFO root: {89f867cc-cbd3-4fa9-a549-4e07be3f69b0}: {Retrying failed operation.}