ubject: Re:Re: 使用hdfs保存checkpoint一段时间后报错
Hi Yun Tang
谢谢唐老师...
HDFS异常问题找到了，是我对hadoop集群重新格式化的时候，没有把SecondaryNameNode的tmp目录的内容删掉，导致格式化后，NameNode、DataNode中VERSION里面ClusterID与SecondaryNameNode的不一致，所以会不定时报出异常，问题已经解决啦。感谢~~
另外有个小问题请教一下：
Flink提供的默认EventTimeTrigger.java里面，在onElement(...)和onEventTime(
d-to-0-nodes-instead-of-minreplication-1-there-are-1/36310025<https://stackoverflow.com/questions/36015864/hadoop-be-replicated-to-0-nodes-instead-of-minreplication-1-there-are-1/36310025>
>
>祝好
>唐云
>
>
>
>From: 邵志鹏
>Sent: Tuesday, Apri
com/questions/36015864/hadoop-be-replicated-to-0-nodes-instead-of-minreplication-1-there-are-1/36310025>
祝好
唐云
From: 邵志鹏
Sent: Tuesday, April 30, 2019 15:26
To: user-zh@flink.apache.org
Subject: 使用hdfs保存checkpoint一段时间后报错
使用hdfs保存checkpoint一段时间后报错，自动重启后正常运行一段时间后继
使用hdfs保存checkpoint一段时间后报错，自动重启后正常运行一段时间后继续报同样的错
Caused by: java.io.IOException: Could not flush and close the file system
output stream to
hdfs://master:9000/flink/checkpoints/PaycoreContextHopJob/cbb3a580d0323fbace80e71a25c966d0/chk-11352/fc4b8b08-2c32-467c-a1f4-f384eba246ff
in order to
