Re: Cleaning up hdfs working directory
Hi, Ketan AFAIK, now kylin working directory just only supports HDFS or some other file systems which are compatible with HDFS API. You can have a try to cleanup garbage file by using Kylin storage cleanup tool. It should reduce your data size. Thanks Zhengshuai > On Jan 24, 2019, at 4:03 PM, kdcool6932 wrote: > > Hi Kylin,We are having a a lot of data in our hdfs working directory, around > 10tb , for last one year or so, this is acutally more than the hbase usage of > kylin(around 9TB) on one of our kylin cluster. We are using kylin 2.3.1 on > this cluster.1. Are all these files required for Kylin functionality ??2. Is > there a way to clean them up(kylin clean up job is not helping here), and > keep only required data on hdfs??3. Also does this storage needs to be on > hdfs only, or can we point it to some non dfs storage, like local FS or s3 > bucket ? > This might help us in reducing our hdfs storage and using it more judiciously. > Thanks,ke...@exponential.com > > > Sent from my Samsung Galaxy smartphone.
Re: Cleaning up hdfs working directory
Hi Take a look at this: http://kylin.apache.org/docs/howto/howto_cleanup_storage.html kdcool6932 于2019年1月24日周四 上午8:04写道: > Hi Kylin,We are having a a lot of data in our hdfs working directory, > around 10tb , for last one year or so, this is acutally more than the hbase > usage of kylin(around 9TB) on one of our kylin cluster. We are using kylin > 2.3.1 on this cluster.1. Are all these files required for Kylin > functionality ??2. Is there a way to clean them up(kylin clean up job is > not helping here), and keep only required data on hdfs??3. Also does this > storage needs to be on hdfs only, or can we point it to some non dfs > storage, like local FS or s3 bucket ? > This might help us in reducing our hdfs storage and using it more > judiciously. > Thanks,ke...@exponential.com > > > Sent from my Samsung Galaxy smartphone. -- Regards! Aron Tao
Cleaning up hdfs working directory
Hi Kylin,We are having a a lot of data in our hdfs working directory, around 10tb , for last one year or so, this is acutally more than the hbase usage of kylin(around 9TB) on one of our kylin cluster. We are using kylin 2.3.1 on this cluster.1. Are all these files required for Kylin functionality ??2. Is there a way to clean them up(kylin clean up job is not helping here), and keep only required data on hdfs??3. Also does this storage needs to be on hdfs only, or can we point it to some non dfs storage, like local FS or s3 bucket ? This might help us in reducing our hdfs storage and using it more judiciously. Thanks,ke...@exponential.com Sent from my Samsung Galaxy smartphone.