One minute mvs: data set file system (/dsfs) on z/OS

• One minute mvs: data set file system (/dsfs) on z/OS
• Setting up data set file system /dsfs on z/OS
• Creating data sets and members from /dsfs
• Problems I experienced when using /dsfs

There is a good overview of dsfs here.

My understanding of how dsfs works is that you can access z/OS datasets, members and spool, through a Unix file system interface, and so use commands like “ls”. When you read a member, it is read into the dsfs file system, and the Unix commands work on this data. When you write to the Unix file, it is written to the dsfs file system, and when the file is closed, the data is written to the dataset.

Some information is cached in the dsfs file system, and some definitions, such as DCB information on a per userid basis is also stored.

DSFS is defined to OMVS as a Physical File System. Part of the definition of the PFS is the started task name. When the PFS is defined to OMVS, it starts the DSFS started task.

To stop it, you issue a command to OMVS to stop the PFS.

Permissions

The dsfs started task has threads which do work. When you issue a request, one of the threads becomes your userid and accesses the data sets, so there is no change to the standard RACF data set protection.

User defaults

You can define defaults for when data sets are created. These are user dependant. To issue these for another user, you have to become a super user to become that id, and then issue the dfsadm command.

Changing configuration

You can change the configuration file and restart DSFS (which will be disruptive).

You can make changes to the active system (which are not saved) using the omvs command

• dsadm configquery
• dsadm config

Diagnosis

Messages are produced, and are documented in the M&C manual. if you get a reason code, you can use bpxmtext code to display the meaning of the code.