You are here: 24 Data Staging > How-to > Data Staging to or from a Shared File System in a Grid > Configuring the $CLUSTERHOST Variable
24.5 Configuring the $CLUSTERHOST variable
Context
When you submit a data staging job in a grid environment, you can use the $CLUSTERHOST variable instead of specifying a destination if you do not know the cluster where the job will run but want the data staged to the same location. Before the variable will work correctly, you must first configure it by customizing the ds_config.py script to match your unique system.
Use case
In a grid with three clusters, you have a partition named master where you want all data staged to a host named gridheadNAS; a partition named csdept where you want all data staged to a host named fs001.cs.example.edu; and a partition named lab where you want all data staged to a host named bigfilesystem.
To configure the $CLUSTERHOST variable
- Open the ds_config.py file for modification. It is located in
[moab]$ vi /opt/moab/tools/data-staging/ds_config.py
- Locate the PARTITION_TO_HOST parameter.
...
PARTITION_TO_HOST = {"partition_1_name":"cluster_1_staging_hostname","partition_2_name":"cluster_2_staging_hostname",
"partition_3_name":"cluster_3_staging_hostname"}
...
- Replace the partition names and associated cluster hostnames with those that match your system. For the use case provided above, you would customize it the following way:
...
PARTITION_TO_HOST = {"master":"gridheadNAS","csdept":"fs001.cs.example.edu",
"lab":"bigfilesystem"}
...
Related Topics