Hi Kai, Ok, I will have a look and try, thanks for your help on this.
Best Regards, Evan From: Zheng, Kai [mailto:[email protected]] Sent: 2016年1月8日 18:06 To: yaoxiaohua Subject: RE: dfs.datanode.failed.volumes.tolerated change Ref. http://blog.cloudera.com/blog/2015/05/new-in-cdh-5-4-how-swapping-of-hdfs-da tanode-drives/ Please discuss this in the public email thread you initiated, so others may also help you, in case I don’t or can’t answer your questions. Regards, Kai From: yaoxiaohua [mailto:[email protected]] Sent: Friday, January 08, 2016 4:57 PM To: Zheng, Kai <[email protected]> Subject: RE: dfs.datanode.failed.volumes.tolerated change I don’t change anything in dfs.datanode.data.dir property, I just change dfs.datanode.failed.volumes.tolerated property. Thanks for your help. From: Zheng, Kai [mailto:[email protected]] Sent: 2016年1月8日 16:40 To: yaoxiaohua; [email protected] Subject: RE: dfs.datanode.failed.volumes.tolerated change As far as I know, Hadoop 2.6 supports disk hot-swapping on a DataNode without restarting the DataNode. Roughly you need to do two operations: 1) change dfs.datanode.data.dir in the DataNode configuration to update according to your removed/added disks; 2) let the DataNode reload its configuration. Please google and checkout related docs for this feature. Hope this helps. Regards, Kai From: yaoxiaohua [mailto:[email protected]] Sent: Friday, January 08, 2016 11:24 AM To: [email protected] Subject: dfs.datanode.failed.volumes.tolerated change Hi, The datanode process shutdown abnormally, I set dfs.datanode.failed.volumes.tolerated =2 in hdfs-site. xml. Because I found one disk is fail to access. Then restart the datanode process, it works. One day later, we replace a good harddisk , and create folder and chown to hadoop. Then I want to know if I don’t restart datanode process, when the datanode can know that The disk is good now? May I have to restart datanode process to update this? Env:Hadoop 2.6 Best Regards, Evan
