Remote mount at fileset level
This topic lists the steps to configure HDFS Transparency to mount a remote fileset.
To configure HDFS Transparency to mount a remote fileset, perform the following:
- From the owning cluster, add the remote fileset into the allowed fileset list. For more
information, see Fileset access control for remote
clusters.
[root@owningcluster ~]# mmauth grant accessingcluster -f gpfs --fileset fset1,root mmauth: [I] Collecting fileset information ... mmauth: Granting cluster accessingcluster access to file system gpfs: access type rw; root credentials will not be remapped. mmauth: Propagating the cluster configuration data to all affected nodes. mmauth: Command successfully completed
Note: For the command to run successfully, ensure that the root fileset is included in the allow list when you are enabling the remote fileset access control. - Ensure that only the granted filesets are visible from the accessing
cluster.
[root@accessingcluster ~]# mmlsfileset remotefs Filesets in file system 'remotefs': Name Status Path root Linked /remotefs indepfset1 Linked /remotefs/indepfset1 depfset1 Linked /remotefs/depfset1
- Stop the HDFS Transparency cluster.
mmhdfs hdfs stop
- For remote mount, set the granted fileset as the gpfs.data.dir parameter in
gpfs-site.xml. Both independent and dependent filesets are supported. If you
are operating with 2fs mode, no configuration changes are needed.
<property> <name>gpfs.mnt.dir</name> <value>/remotefs</value> <description>gpfs mount point</description> </property> <property> <name>gpfs.data.dir</name> <value>indepfset1</value> <description>Set this to a sub directory in gpfs mount point will make this sub directory as the root directory from hadoop client point of view. Leave it empty to make whole gpfs file system visible to hadoop client. When specify the sub directory, gpfs mount point should not be included in the string.</description> </Property>
- Upload and sync the changes across all the transparency
nodes.
mmhdfs config upload
- Start all the transparency nodes and check the status of
namenode.
mmhdfs hdfs start; hdfs haadmin -getAllServiceState
- Change the fileset permissions from 700 to
755.
hdfs dfs -chmod 755 /