Components of the DSFS tree
The DSFS tree has four conceptual levels: root, path, HLQ
directory, and data sets.
Root directory
Mount the utility file system at /dsfs in the z/OS® UNIX file system tree, which results in the root of the DSFS tree at path /dsfs. Because this directory is read-only, users cannot store files or directories in it. DSFS places the /txt, /bin, and /rec path directories in this directory at mount time if they are not already present.
All DSFS users have access to the root directory.
Path directory
- Binary (/bin)
- Data sets that are accessed by this path are treated by DSFS as binary data.
When a data set is opened by DSFS on behalf of users who are accessing the data set through this
path, DSFS reads the records from the data set. Next, it stores the records sequentially in a file
in the utility file system as a POSIX byte stream. Updates made by applications to this file are
stored in the utility file system file. When the application closes the file, DSFS writes the data
to the data set, storing the bytes into sequential records in the data set.
- For variable-length records, the maximum record size is used for each record except for the last record, which can be partial.
- For fixed records, the last record is padded with binary zeros if the last record was short.
- Record (/rec)
- Data sets that are accessed by this path are treated by DSFS as z/OS record format files. When it accesses a data set, DSFS
prepends a 4-byte header to each record it reads from the data set and stores the updated records
into the POSIX file that is used to represent the data set.
- For fixed record data sets, each header must specify a length equal to the LRECL of the data set.
- For variable-length data sets, each header is used to determine the size of the record that is written to the data set.
- Text (/txt)
- Data sets that are accessed by this path are treated by DSFS as text files. Upon access to a data set, DSFS reads each record from the data set. If it is in fixed format, it strips the trailing blanks. For both fixed and variable record formats, it appends a newline character to the end of the record in the POSIX byte stream. Updates made by applications to this file are stored in the corresponding utility file system file. When applications close the file, DSFS scans the file’s byte stream for newline characters, where each newline character delineates records to be written to the data set. The newline character is not written to the data set. For fixed-length record data sets, DSFS pads with blanks each record to ensure it matches the LRECL.
High-level qualifier directories
DSFS
places a high-level qualifier directory name in a path directory if the user
application changes the working directory or uses a path name that specifies the high-level qualifier directory name in that path. A HLQ directory name can
consist of one or more qualifiers of a data set name. Once a HLQ directory has been created, it can
only be deleted if the HLQ directory name is excluded by the HLQ list.
New HLQ directories are created only when there is at least one
existing PDS, PDSE, or sequential data set whose name begins with the HLQ directory
name.Users might want to use more than one qualifier to
reduce the amount of catalog processing needed to populate or refresh the HLQ
directories.To control the data sets that DSFS can add to its tree, the
administrator can create a list of excluded HLQ directory names by using both the HLQ_LIST parameter
in the IDFFSPRM file and the dsadm config
-hlq_list_add or -hlq_list_remove command. The user must have
authority to the DFSMS catalog that contains the high-level qualifiers to access the HLQ directory.
Upon
first access, DSFS will populate the HLQ directory in its utility file system with names of
supported data sets whose names begin with the HLQ directory name and have at least one other
qualifier. The supported data sets are fixed or variable record
PS/PDS/PDSE.
HLQ directory name, the DSFS utility file system must be
encrypted.The
names that appear in the data set level of the DSFS directory tree do not have the HLQ directory name portion of the data set name. In Figure 1, physical sequential data set scott.file1.out
is represented as a file called file1.out in the HLQ directory named
scott. The sequential data set
scott.file1.out could also be accessed as scott.file1/out
if scott.file1 were used as the HLQ directory name.
The same HLQ directory names can be accessed from multiple paths to allow different data sets of
that HLQ directory name to be processed in different formats. Dynamic allocation (DYNALLOC) and
ENQUEUE (ENQ) serialization is used to prevent the same data set from being processed by multiple
paths at the same time. Similarly, this serialization is also used to prevent the same data set from
being processed by multiple HLQ directories at the same time.
Creation parameters
With
the dsadm createparm command, users can specify the data set attributes to use
when PS or PDS/PDSE data sets are created. Creation parameters are assigned to an HLQ directory and
apply to all data sets created with the same HLQ directory name.
These creation parameters are permanently stored in the utility file system and
associated with the HLQ directory. They are permanently stored in the utility file system for the
directory that represents the high-level qualifier directory name. The user ID of the issuer of the
dsadm createparm command must match the HLQ in order to set or replace the
creation parameters.
Data sets
HLQ directory that is accessed by a DSFS user has a corresponding file (PS) or directory (PDS
or PDSE) created inside the HLQ directory with the high-level qualifier directory
name removed. Names appear in lowercase, but users can access the names by using uppercase,
lowercase, or mixed-case because DSFS is
case-insensitive.# ls my*
my.file
# ls -l my*
-rwxrwxrwx 1 BPXROOT SYS1 1 Jul 23 00:00 my.file
# ls -l My*
ls: FSUM6785 File or directory "My*" is not found
# ls -l My.file
-rwxrwxrwx 1 BPXROOT SYS1 1 Jul 23 00:00 My.file Users can create, rename, and remove data sets with the standard UNIX commands such as rm or mv.
Partitioned data sets (PDS and PDSE) are represented as directories in DSFS, with the members represented as files in DSFS. Users can read, write, rename, create, and remove members.