|
|
-
bulkload on fully distributed mode - permissions
Christopher Dorner 2011-12-13, 10:29
Hi,
i stumbled upon an error which was not present in pseudo distributed mode. When i try to run a bulkload, it fails after creating the hfiles with following error:
org.apache.hadoop.security.AccessControlException: org.apache.hadoop.security.AccessControlException: org.apache.hadoop.security.AccessControlException: Permission denied: user=hbase, access=WRITE, inode="/user/dorner/bulkload/output/Tsp/v":dorner:hadoop:drwxr-xr-x After some reading, I know that HBase needs to own the files. I am not in the sudo group on the cluster to change the owner (and probably can't get in). And i can't ask the administrator every time to change the owner of the file. Is there a way to change this behavior and allow hbase to access those files without changing permissions and owners?
On my local pseudo-distributed mode, it runs perfect without any changes. I am using cloudera CDH3u2.
Regards, Christopher
-
RE: bulkload on fully distributed mode - permissions
Paul Mackles 2011-12-13, 11:56
If you can chmod a+w the directory /user/dorner/bulkload/output/Tsp, hbase should be able to do what it needs to do (I am assuming the error is coming from completebulkload). It is trying to rename the files.
-----Original Message----- From: Christopher Dorner [mailto:[EMAIL PROTECTED]] Sent: Tuesday, December 13, 2011 3:29 AM To: [EMAIL PROTECTED] Subject: bulkload on fully distributed mode - permissions
Hi,
i stumbled upon an error which was not present in pseudo distributed mode. When i try to run a bulkload, it fails after creating the hfiles with following error:
org.apache.hadoop.security.AccessControlException: org.apache.hadoop.security.AccessControlException: org.apache.hadoop.security.AccessControlException: Permission denied: user=hbase, access=WRITE, inode="/user/dorner/bulkload/output/Tsp/v":dorner:hadoop:drwxr-xr-x After some reading, I know that HBase needs to own the files. I am not in the sudo group on the cluster to change the owner (and probably can't get in). And i can't ask the administrator every time to change the owner of the file. Is there a way to change this behavior and allow hbase to access those files without changing permissions and owners?
On my local pseudo-distributed mode, it runs perfect without any changes. I am using cloudera CDH3u2.
Regards, Christopher
-
Re: bulkload on fully distributed mode - permissions
Christopher Dorner 2011-12-14, 14:59
Thank you, that helped. So hbase actually does not need to be the owner but have write access to the files.
Am 13.12.2011 12:56, schrieb Paul Mackles: > If you can chmod a+w the directory /user/dorner/bulkload/output/Tsp, hbase should be able to do what it needs to do (I am assuming the error is coming from completebulkload). It is trying to rename the files. > > -----Original Message----- > From: Christopher Dorner [mailto:[EMAIL PROTECTED]] > Sent: Tuesday, December 13, 2011 3:29 AM > To: [EMAIL PROTECTED] > Subject: bulkload on fully distributed mode - permissions > > Hi, > > i stumbled upon an error which was not present in pseudo distributed mode. > When i try to run a bulkload, it fails after creating the hfiles with > following error: > > org.apache.hadoop.security.AccessControlException: > org.apache.hadoop.security.AccessControlException: > org.apache.hadoop.security.AccessControlException: Permission denied: > user=hbase, access=WRITE, > inode="/user/dorner/bulkload/output/Tsp/v":dorner:hadoop:drwxr-xr-x > > > After some reading, I know that HBase needs to own the files. I am not > in the sudo group on the cluster to change the owner (and probably can't > get in). And i can't ask the administrator every time to change the > owner of the file. Is there a way to change this behavior and allow > hbase to access those files without changing permissions and owners? > > On my local pseudo-distributed mode, it runs perfect without any changes. > I am using cloudera CDH3u2. > > Regards, > Christopher
|
|