- bufhigh(xx, yy)
- Specifies
the below-the-2GB bar virtual storage limit
for data buffers on z/OS NFS Server. When this maximum
limit of allocated buffer storage is reached buffer reclamation is
initiated (see the percentsteal attribute in this table).
- xx
- the high water mark data buffer storage limit (in
bytes, KB, or MB). The valid range is 1 to 2047
MB.
The default storage limit (xx) is 32 MB.
A higher storage limit (xx) means more caching, and
potentially better read performance.
- yy
- the watermark in percent of the storage limit (xx)
for printing a data buffer utilization alert message. The valid values
are 0 (which turns off the data buffer utilization reporting mechanism)
and from 50 to 90. If no percentage (yy )
value is specified, a default value of 80 percent is used. If an
invalid value is specified for the percentage (yy),
the default value is used.
At z/OS Server startup, the actual
value (xx) specified with bufhigh may be
adjusted by the z/OS NFS Server internally depending on the available
region size and other z/OS NFS Server memory requirements to enable
the z/OS NFS Server to execute properly.
Within
limits, the bufhigh values can be changed while the z/OS NFS
Server is running with the MODIFY operator command (see the bufhigh
operand in Entering operands of the modify command for the z/OS NFS server).
The bufhigh attribute
does not apply to z/OS UNIX files.
|
- cachewindow(n)
- Specifies the window size
used in logical I/O to buffer NFS Clients' RPC WRITES received
out of order. The value of n is a number
from 1 to 2048 (the default is 112). The cachewindow attribute
does not apply to z/OS UNIX files. The cachewindow
attribute is ignored if the server-adjusted logicalcache is greater
than 2GB The suggested value is some small multiple of the number
of BIODs running on a client. The general rule in setting the n value
of cachewindow(n) is
n = (( num of BIOD + 1 ) * (client_max_IO_buffer_size/transfer_size))
- num of BIOD is the number of blocked I/O daemons set by
the client workstation. This value is usually set to defaults at the
installation of the operating system or by your system administrator.
- client_max_IO_buffer_size is the amount of I/O data requested
by the client (for example, client writes 8192 bytes of data to the
remote file system). This value is determined by your application
programs.
- transfer_size is the actual size of data being sent across
the network (for example, the 8192 bytes of data can be broken down
to 16 smaller packets of 512 bytes (16x512=8192)). This value is determined
dynamically by your client workstation.
|
- checklist
- When specified, the server bypasses saf checking
(even when saf or safexp is specified) for the list
of files and directories underneath mount points which either matches
a mount point entry or is a child of a mount point entry specified
on the dirsuf parameter in the exports data set. CHECKLIST
is only valid if SAF checking is the security option for the particular
data access; otherwise, it is ignored even if it is specified. See
GFSAPEXP in NFSSAMP library for a sample exports data set.
- nochecklist
- When specified,
the server operates as before and ignores the information that is
specified on the dirsuf parameter in the exports data set.
|
- delegation
- When specified,
the server temporarily delegates management of a file's resources
to an NFS client for NFS Version 4. When a file's management is delegated
to an NFS client, all file access requests can be managed locally
by the NFS client while the file is delegated.
- nodelegation
- When specified,
the server does not delegate management of a file's resources to an
NFS client for NFS Version 4.
|
- denyrw
- When
specified, the server honors deny requests for file share reservations
(the Windows Share_Deny value) from the NFS client. The deny requests
may be specified on an NFS V4 Open operation or an NLM_share RPC.
- nodenyrw
- When specified,
the server ignores deny requests from NFS clients (the Windows Share_Deny
value), and treats the requests as if deny_none were specified.
|
- dhcp
- When specified, the
server accepts dynamic IP addresses for the NFS client, using the
dynamic host configuration protocol (dhcp). The client system must
have a static host name and must dynamically update the DNS server
with their IP address changes.
- nodhcp
- When
specified, the server supports only NFS clients that use a static
IP address.
|
- DlyDTimeout(n)
- Specifies the minimum delay detection
time value in sec before the delay detection mechanism observes a
delay in an external call/function and prints message GFSA1030E on
the console.
Valid values are 0 and a range of 5 to 60 seconds.
Any value of DlyDtimeout from 1 to 4 seconds is converted to 5 seconds.
If DlyDTimeout is set to 0 the delay detection mechanism is turned
off. The default value is 10 seconds.
|
- fileidsize(n)
-
Specifies
how to control the handling of fileid sizes by the NFS server in NFS
objects. Fileids may be recognized either as 32-bit or 64-bit addresses.
Valid
values are 32 and 64.
The default value is fileidsize(64).
|
- fn_delimiter
- Specifies a character 'c' to
be used as a delimiter between the file name and the attributes that
follow it. This capability allows those sites that have UNIX data
sets containing commas to copy and store their data on the NFS server.
The following example specifies the default delimiter as a semicolon:
fn_delimiter(;)
So
a user can process a file called 'comma,in-name' by entering: vi "comma,in-name;text,lf"
Note: - If the comment symbol was set as ";" with the altsym keyword
and the fn_delimiter(;) attribute uses semicolon then the fn_delimiter
semicolon will be treated as a delimiter between the file name and
the attributes that follow, not as a comment symbol.
- It is admissible to use the semicolon as the comment symbol after
the right parenthesis if the altseq keyword is used.
- The following example shows allowable multi-line syntax:
fn_delimiter \
(;) ; the second semicolon is the comment symbol if altseq kwd is used
(;) must
be located on one line.
A user can also include a default file name delimiter
as a comma as follows: fn_delimiter(,)
- fn_delimiter(,)
- The default file name delimiter is a comma.
|
- hfs(prefix) or hfsprefix(prefix)
- Specifies a z/OS UNIX file system prefix to
be imbedded in the mount directory path name. The default value of
the z/OS UNIX file system prefix is /hfs. Mount requests
received by the z/OS NFS server beginning with the z/OS UNIX file
system prefix value are identified as mount requests for z/OS UNIX.
The z/OS UNIX file system prefix value is not part of the path name.
Note: - hfsprefix is preferred and should be used in future updates, but
hfs is still accepted.
- The z/OS UNIX file system must already be mounted locally by z/OS
UNIX. Otherwise, the client mount request will fail.
- The prefix value can only be 7 characters or less including the
beginning "/"
- The prefix value is case insensitive. It is always folded to upper
case.
|
- hfsfbtimeout(n)
- Specifies
how to control the timeout of the z/OS UNIX vnode token used by the
NFS server. The timeout value controls how long before vnode tokens saved
in file blocks are released.
The valid range is 1 to 32,767 seconds. - The value of n can go as low as 1 second but to avoid the
possibility of the client hanging (because of network delays). The
value of n is not recommended to be lower than 5 second.
- The value of n may need to be increased if the network
is slow and the accessed directory has a lot of entries.
The hfsfbtimeout attribute default value is 60 seconds.
|
- hfssec(krb5,krb5i,krb5p,sys)
- Specifies the acceptable
network transmission levels of security which can be used as the authentication
flavor on NFS version 4 requests for accesses to z/OS UNIX files.
This attribute is only used when not overridden by authentication
specifications in the exports file. Multiple values for this attribute
can be specified using the comma as delimiter. The following are
the supported values:
- krb5
- Provides Kerberos V5 based integrity on the RPC credentials (but
not data), when the RPC authentication flavor is RPCSEC_GSS. It uses
the DES_MAC_MD5 integrity algorithm and the RPCSEC_GSS service of
rpc_gss_svc_none.
- krb5i
- Provides Kerberos V5 based integrity on both the RPC credentials
and data, when the RPC authentication flavor is RPCSEC_GSS. It uses
the DES_MAC_MD5 integrity algorithm and the RPCSEC_GSS service of
rpc_gss_svc_integrity.
- krb5p
- Provides Kerberos V5 based integrity and privacy on both the RPC
credentials and data, when the RPC authentication flavor is RPCSEC_GSS.
It uses the DES_MAC_MD5 algorithm for integrity and 56 bit DES for
privacy. The RPCSEC_GSS service used here is rpc_gss_svc_privacy.
- sys
- Specifies that the AUTH_SYS authentication flavor can also be
used to access this file system. Note that the AUTH_SYS authentication
flavor does not provide any integrity or privacy protection.
The hfssec attribute default
is hfssec(sys,krb5,krb5i,krb5p). Note: File systems
that require integrity or privacy protection over network transmissions
of data should explicitly specify the desired settings. Do not rely
on the default settings, because the default settings allow for RPC
accesses using the AUTH_SYS authentication flavor, which does not
provide any integrity or privacy protection.
|
- impprefix(impprefix)
- Specifies how to interpret a mount path name that
does not have a path type prefix, where immprefix is one of
the following:
- NONE
- An explicit prefix must always be specified for an absolute path.
Implicit prefix resolution is not valid in this case.
- HFS
- If no explicit prefix is present, assume the path is a z/OS UNIX
file system.
- MVS
- If no explicit prefix is present, assume the path is an MVS high-level qualifier. This is the default.
- HFS, MVS
- If no explicit prefix is present, first assume the path is a z/OS
UNIX file system. If no matching z/OS UNIX file system can be found,
assume that it is an MVS high-level qualifier.
- MVS, HFS
- If no explicit prefix is present, first assume the path is an MVS high-level qualifier. If no matching high
level qualifier can be found, assume that it is a z/OS UNIX file system.
Note: - For the "MVS,HFS" setting, MVS selection requires that at least
one MVS data set exists in the system catalog with the specified High
Level Qualifier (HLQ).
- The"MVS" setting, results in equivalent implicit prefix processing
to releases prior to V1R11; that is, z/OS UNIX requires a prefix and
MVS does not. However, an MVS prefix can still be specified on path
names.
- Since NFS v4 mount processing is performed one qualifier at a
time, when 2 options are specified, the object existence test for
determining whether to move on to the second option is based on the
first path name qualifier only. If it exists, then the first option
is selected and that cannot change if a later qualifier is not found.
|
- leadswitch
- Specifies that the server returns ‘/’ as
the first character in each export entry.
- noleadswitch
- Specifies
that the server will not return ‘/’ as the first character in each
export entry.
The leadswitch attribute is ignored for z/OS
UNIX file objects.
|
- leasetime(n)
- Specifies the length of time (the lease interval)
in seconds that the z/OS NFS server allows clients to:
- Reclaim locks and share reservations following an NFS server restart.
During this grace period, clients can reclaim locks on behalf of their
users.
- Remain active without communicating with the NFS server. If an
NFS V4 client does not communicate with the z/OS NFS server for the
length of the lease interval, its client id will expire.
The value of n can range from 5
to 3600. The specified value must be smaller than the value of the logout attribute,
if logout is not set to zero. The default value is 120.
Note: When using the NFS version 4 protocol, the leasetime value
should be set to a value larger than or equal to the attrtimeout,
writetimeout and readtimeout attributes. Otherwise, it is possible
for performance problems to occur when attempting to access MVS data
sets.
|
- logicalcache(n)
- Specifies the above-the-bar virtual storage for allocated
logical cache buffers in the logical I/O processing. If n
is greater than the available storage above-the-bar (implied by the
MEMLIMIT parameter in the startup procedure) at startup, the z/OS
NFS Server shuts down immediately.
The value
of n is an integer from 1 to 4096GB. The
default value is 16M .
The logicalcache attribute
does not apply to z/OS UNIX files.
At z/OS Server
startup, the actual value ( n) specified with logicalcache may
be adjusted by the z/OS NFS Server internally depending on the available
MEMLIMIT and other z/OS NFS Server memory requirements to enable the
z/OS NFS Server to execute properly. - If n is greater than the available storage above-the-bar
after z/OS NFS Server starts its threads (which uses some memory above-the-bar
due to the LE runtime THREADSTACK64 options), the Server shuts down
immediately.
- If n is less than the available storage above-the-bar after
z/OS NFS Server starts its threads, then the server may increase the
initial nn specification up to the smaller of 4096GB or one-half of
the available storage if n is smaller than one-half of the
available storage; or the server honors the specified nn if n is
greater than one-half of the available storage (no expansion).
- If the total number of threads in the nfsstasks attributes
is X, then LE use X * 0.25MB for thread stacks (due to THREADSTACK64(256K)
runtime option). If MEMLIMIT is 1024GB and there are 100 threads that
initially use 25MB (100 * 0.25MB) and the logicalcache is 100MB,
then the server may expand the logicalcache to 512GB.
|
- logout(n)
- Specifies
the time limit for inactivity in seconds for a given user on a client.
The default value is 1800. When the limit is reached,
the user is automatically logged out. The client user must enter the mvslogin command
again to reestablish the client's z/OS session. This value should
normally be the same as the value defined for TSO/E logout at your
site. The value of n can range from 61 seconds
to 20 megaseconds (approximately 243 days).
|
- maxrdforszleft(n)
- Specifies the number
of physical block buffers left after determining a file's size. This
operation is done for later server read requests to the same file.
The buffers left are subject to trimming during a "buffer steal" operation.
The value of n is an integer from 1 to 1024.
The
default value is 32.
|
- maxtimeout(n)
- Specifies the maximum
timeout allowed. This attribute and the mintimeout attribute
define the range of values that client users can specify for attrtimeout, readtimeout,
and writetimeout. The value of n is
the number of seconds from 1 to 32,767 (9 hours, 6 minutes, and 7
seconds). This attribute does not affect the logout attribute.
- nomaxtimeout
- Allows client
users to specify noattrtimeout, noreadtimeout, and nowritetimeout.
|
- mintasks(n,m,o)
- Defines the minimum number of NFS tasks or threads allowed to
run. Tasks may be terminated for reasons such as 80A or 878 ABENDs.
- n
- Specifies the minimum number of subtasks which handle the asynchronous
I/O operations or short blocking operations. If the number of active
‘short’ tasks becomes less than n the shutdown
process of the NFS server starts.
- m
- Specifies the minimum number of subtasks which handle z/OS UNIX
file requests. If the number of active z/OS UNIX tasks becomes less
than m the shutdown process of the NFS server starts.
- o
- Specifies the minimum number of subtasks which handle long blocking
operations. If the number of active legacy long service tasks becomes
less than o the shutdown process of the NFS server starts.
If n, m, or o are
greater than the corresponding values in nfstasks, they are
assigned to half the nfstasks values. If n, m,
or o are not specified, they are assigned default values of
4, 4 and 1, respectively. - Valid range for n is from 4 to 99
- Valid range for m is from 4 to 100
- Valid range for o is less than or equal to 99
- Valid range for n + o is less than or equal to 100
|
- mintimeout(n)
- Specifies
the minimum timeout. This attribute and maxtimeout define
the range of values that can be specified for attrtimeout, readtimeout,
and writetimeout. The value of n is
the number of seconds from 1 to 32,767.
The default value is 1.
|
- mixcase/upcase
- Specifies
messages display in mixed or upper case.
Note: Starting
with z/OS NFS V1R10, this attribute is ignored.
|
- mvslogindelay()
- Specifies
the delay time value in seconds since z/OS NFS Server startup.
To
avoid an NFS client's cache invalidation due to access errors after
z/OS NFS server startup, in SECURITY(SAF/SAFEXP) mode, the server
maps the reply error NFS3ERR_ACCES/NFS4ERR_ACCES to NFS3ERR_JUKEBOX/NFS4ERR_DELAY
on NFSv3 and NFSv4 RPC requests, until the mvslogin is received or
the mvslogindelay expires. This allows an NFS client time to reissue
an MVSlogin.
- The maximum value for mvslogindelay is 300 seconds.
- The minimum value for mvslogindelay is 0 seconds.
The default value is 0 seconds (off). If mvslogindelay is
set to 0 the error mapping is turned OFF.
|
- mvsprefix(prefix)
- Specifies an MVS data set prefix to be appended
to the front of MVS data set name for a mount path directory. The
default value of prefix is /mvs.
The
mvs prefix enables you to explicitly specify a prefix for identifying
MVS data sets, simliar to the way in which the hfs prefix does for
z/OS Unix files.
|
- mvssec(krb5,krb5i,krb5p,sys)
- Specifies
the acceptable network transmission levels of security which can be
used as the authentication flavor on NFS version 4 requests for accesses
to MVS data sets. This attribute is only used when not overridden
by authentication specifications in the exports file. Multiple values
for this attribute can be specified using the comma as delimiter.
The following are the supported values:.
- krb5
- Provides Kerberos V5 based integrity on the RPC credentials (but
not data), when the RPC authentication flavor is RPCSEC_GSS. It uses
the DES_MAC_MD5 integrity algorithm and the RPCSEC_GSS service of
rpc_gss_svc_none.
- krb5i
- Provides Kerberos V5 based integrity on both the RPC credentials
and data, when the RPC authentication flavor is RPCSEC_GSS. It uses
the DES_MAC_MD5 integrity algorithm and the RPCSEC_GSS service of
rpc_gss_svc_integrity.
- krb5p
- Provides Kerberos V5 based integrity and privacy on both the RPC
credentials and data, when the RPC authentication flavor is RPCSEC_GSS.
It uses the DES_MAC_MD5 algorithm for integrity and 56 bit DES for
privacy. The RPCSEC_GSS service used here is rpc_gss_svc_privacy.
- sys
- Specifies that the AUTH_SYS authentication flavor can also be
used to access this data set. Note that the AUTH_SYS authentication
flavor does not provide any integrity or privacy protection.
The mvssec attribute default
is mvssec(sys,krb5,krb5i,krb5p). Note: File systems
that require integrity or privacy protection over network transmissions
of data should explicitly specify the desired settings. Do not rely
on the default settings, because the default settings allow for RPC
accesses using the AUTH_SYS authentication flavor, which does not
provide any integrity or privacy protection.
|
- nfstasks(n,m,o,t,u)
- Specifies
the number of server processes to initiate on startup.
If nfstasks( n, m)
is specified, then the following is true. - The value of n is the number of subtasks
that handle the asynchronous input/output (I/O) operations or short
blocking operations (the maximum number of concurrent NFS server requests) in
the z/OS MVS data path.
- The value of m is the number of subtasks
that handle the long blocking operations (the maximum number of concurrent
NFS server recall and z/OS UNIX requests). Increase this value if
your server supports lots of active recall or z/OS UNIX clients.
Based on system resources available below the 16 Mb
line, the maximum n value may not be achievable. The precise
maximum value will be system configuration dependent. If an 80A or
878 Abend is experienced during NFS server startup, use a smaller
value for n.
If nfstasks( n, m, o)
or nfstasks( n, m, o, t, u)
is specified, then the following is true. - The value of n is the number of subtasks
that handle the asynchronous input/output (I/O) operations or short
blocking operations (the maximum number of concurrent NFS server requests) in
the z/OS MVS data path.
- The value of m is the number of subtasks
that handle z/OS UNIX requests. Increase this value if your server
supports lots of active z/OS UNIX requests.
- The value of o is the
number of subtasks that handle the long blocking operations (the maximum
number of concurrent NFS server recall requests). Increase this value
if your server supports lots of active recall operations.
- The value of t is the number
of transport subtasks that handle TCP network requests.
- The value of u is the number
of transport subtasks that handle UDP network requests.
Based on system resources available below the 16 Mb line,
the maximum n + o value may not be achievable. The
precise maximum value will be system configuration dependent. If
an 80A or 878 Abend is experienced during NFS server startup, use
a smaller value for n + o.
The following
are valid value ranges for n, m, o, t,
and u. - Valid range for n is from 4 to 99.
- Valid range for m is from 4 to 100.
- Valid range for o is from 1 to 99.
- Valid range for n + o is from 5 to
100
- Valid range for t is from 4 to 32.
- Valid range for u is from 4 to 32.
The nfstasks attribute default is nfstasks(8,16,8,4,4).
|
- nfsv4domain(NFSv4_default_domain)
- specifies
the default domain for the NFS v4 protocol (NFSv4) name mapping.
The
NFSV4DOMAIN attribute serves for redefinition of a name of this unique
domain. In accordance with RFC3530 NFSv4 attributes "owner" and "owner_group"are
transferred between the client and server in the form of "user_name@domain"
and "group_name@domain". The server provides the mapping of names
to ids and vice versa. NFSv4_default_domain identifies the
user/group name space with one to one correspondence between the names
and their numeric identifiers (uids and gids).
z/OS
NFS Server will accept as valid a set of users and groups for default
domain. The server will treat other domains as having no valid translations.
If the NFSV4DOMAIN attribute is not used, the server uses the system-defined
domain. The NFSv4_default_domain will be converted internally
to lower case.
For further details on NFSv4
name mapping, see NFS v4 protocol name mapping.
|
- nlm
- Specifies that the initialization of the z/OS NFS server should
include starting the NLM and NSM daemons.
- nonlm
- Specifies that the
initialization of the z/OS NFS server should not include starting
the NLM and NSM daemons. The system will run without lockd and statd.
Specifying nonlm does not affect the availability of byte range
locking and share reservation support for NFS version 4 protocol access.
If nonlm is
specified, the NLM may not be started after NFS has initialized.
If NLM is desired, you must stop and restart NFS after specifying
the nlm site attribute. The only way to stop NLM is to shut
down the NFS server. It is no longer necessary to define the NLM and
NSM startup procedures to a z/OS UNIX segment as UID(0) to RACF because
the NLM and NSM startup procedures are no longer supported. Note: - The lock data sets must always be allocated, even if nonlm is
specified in the site attributes.
- The old startup procedures for NLM and NSM are no longer shipped
with z/OS; these procedures are obsolete and old copies from previous
releases should not be used on z/OS V1R7 or later releases.
|
- pcnfsd
- Specifies that z/OS NFS server is to start
the PCNFSD server.
- nopcnfsd
- Specifies that z/OS NFS server is not to start the PCNFSD server.
|
- percentsteal(n)
- Specifies
the percent of data buffers that can be reclaimed for use when the bufhigh(xx,yy) limit has been reached. A
higher value means a reclaim operation is frequently
performed, and the cached data is significantly trimmed on each
reclaim. This can result in poor read performance, because readahead
buffers might be stolen. Lower values result in less frequent
reclaim operations, and the cached data normal
water mark is higher, meaning possibly better performance by reading
from cached data.
The value of n is
an integer from 1 to 99.
The percentsteal attribute
default value is 20.
The percentsteal attribute
does not apply to z/OS UNIX files.
|
- public(legacy_path,hfs_path)
- Specifies
the legacy path (MVS z/OS conventional data) and HFS path (z/OS UNIX
data) that is associated with the public file handle for WebNFS access.
The first path, if specified, is the legacy path. The second path
is the HFS path.
If the first path is not present, a comma must
precede the second path. If the public keyword is specified,
then one of the paths must be specified. The public keyword
must be specified after the hfsprefix(), mvsprefix(), and impprefix() keywords
in the site attribute table. A lookup request with the public
file handle determines which of the two paths is being referenced
by the pathname that is specified. An absolute pathname will tell
the server which of the paths is referenced by matching one of the
paths specified. A lookup request with a relative pathname
will be interpreted as a z/OS UNIX request if HFS is active (hfs_path
has been provided); otherwise, it is treated as a MVS request.
The public attribute
default value is no public path.
|
- pubsec(krb5,krb5i,krb5p,sys)
- Specifies
the acceptable network transmission levels of security for accesses
to public file systems which can be specified as the authentication
flavor of the RPC request. This attribute is only used when not overridden
by authentication specifications in the exports file. Multiple values
for this attribute can be specified using the comma as delimiter.
The following are the supported values:.
- krb5
- Provides Kerberos V5 based integrity on the RPC credentials (but
not data), when the RPC authentication flavor is RPCSEC_GSS. It uses
the DES_MAC_MD5 integrity algorithm and the RPCSEC_GSS service of
rpc_gss_svc_none.
- krb5i
- Provides Kerberos V5 based integrity on both the RPC credentials
and data, when the RPC authentication flavor is RPCSEC_GSS. It uses
the DES_MAC_MD5 integrity algorithm and the RPCSEC_GSS service of
rpc_gss_svc_integrity.
- krb5p
- Provides Kerberos V5 based integrity and privacy on both the RPC
credentials and data, when the RPC authentication flavor is RPCSEC_GSS.
It uses the DES_MAC_MD5 algorithm for integrity and 56 bit DES for
privacy. The RPCSEC_GSS service used here is rpc_gss_svc_privacy.
- sys
- Specifies that the AUTH_SYS authentication flavor can also be
used to access file systems. Note that the AUTH_SYS authentication
flavor does not provide any integrity or privacy protection.
The pubsec attribute default is pubsec(sys,krb5,krb5i,krb5p). Note: File
systems that require integrity or privacy protection over network
transmissions of data should explicitly specify the desired settings.
Do not rely on the default settings, because the default settings
allow for RPC accesses using the AUTH_SYS authentication flavor, which
does not provide any integrity or privacy protection.
|
- readaheadmax(n)
- Specifies
the number of bytes to be read to fill internal buffers during read
processing to enhance satisfying read requests directly from cache.
This reduces the amount of synchronous physical I/O performed for
NFS read requests for sequential read file access. It also reduces
context switching overhead on NFS read requests by allowing more read
requests to be satisfied directly from cache.
The value of n is
an integer from 1 KB to 128 KB (normally 2 to 4 times the common block
size used for file access, which is recommended at 8 KB for AIX® file
activity).
The readaheadmax attribute default value
is 16K. Specifying zero (0) will deactivate readahead.
The readaheadmax attribute does not apply to
z/OS UNIX files.
|
- readdirtimeout(n)
- Specifies the amount of time, in seconds, before the internal
readdir cache that is used for MVS z/OS conventional data sets is
timed out or discarded. The valid range is from 1 to 32,767 (9 hours,
6 minutes, and 7 seconds). The value of n can
go as low as 1 second, but to avoid the possibility of client hanging (due
to network delays and staled cache), n is
not recommended to be lower than 5 seconds. The value of n may
need to be increased if the network is slow and the accessed directory
has a lot of entries.
The readdirtimeout attribute default
value is 30 seconds.
|
- rec878
- Specifies
that the recovery processing of 878 and 80A ABENDs will be turned
on, and affected tasks will attempt to recover.
- norec878
- Specifies that
the recovery processing of 878 and 80A ABENDs will be turned off.
That is, if this type of ABEND occurs, the server will shutdown without
recovery. It should only be used for debug.
|
- remount
- When
specified, the server processes NFS requests after the NFS Server
is restarted even though the HFS file system was remounted with a
new HFS file system number (USS device number) after its last usage.
Use of the remount attribute causes the NFS Server to automatically
access a remounted HFS file system even though it may have been changed
prior to remounting. Any active client mounts are re-established.
- noremount
- When specified,
the server fails NFS requests (with return value NFSERR_STALE) if
the HFS file system was remounted with a new HFS file system number
(USS device number) after its last usage.
The remount/noremount attributes apply only to
HFS file systems.
|
- restimeout(n,m)
- Specifies
a retention period and a clock time for the removal of mount points
and control blocks that have been inactive longer than the specified
retention period.
- n
- Specifies the resource retention period for mounts and associated
resources. If they have been inactive for more than n hours,
they are removed.
The valid range for n is
0 to 720 hours (30 days). The default is 48 hours. If n is
set to 0, the z/OS NFS server does not remove any mount points or
associated resources.
- m
- Specifies the time of day to do the cleanup for mounts and associated
resources that have been inactive more than n hours.
The time of day is specified as a 24 hour local time value.
The
valid range for m is 0 to 23. The default
is 0 (that is, midnight). Because cleanup work slows
down the server, set m so that cleanup work
occurs when the server is lightly loaded. If a mount handle is removed
by the cleanup activity, the user must do the umount and mount operations
to access the mount point again. The resource cleanup is also done
when the server is shutting down.
|
- security(mvs[,hfs,public])
- Specifies security options for MVS data sets, z/OS
UNIX files, and data that is accessed using the public file handle.
- mvs
- Specifies the security option for MVS z/OS conventional data sets.
The mvs parameter is a required parameter.
- hfs
- Specifies the security option for z/OS UNIX files. The hfs parameter
is an optional parameter.
- public
- Specifies the security option for data that is accessed with the
public file handle. The public parameter
is an optional parameter.
Note: When the optional parameters (hfs and public)
are not specified, they are assigned the same security option as the
first parameter.
You can specify the following security
options:
- exports
- Exports list checking. For z/OS UNIX files, checks UNIX permission bits. The UID is obtained from
the client RPC request. No SAF checking.
- none
- Neither SAF checking nor exports list checking. For z/OS UNIX
files, checks UNIX permission bits. The UID is
obtained from the client RPC request.
- saf
- SAF checking. No exports checking. For z/OS UNIX files,
checks UNIX permission bits. The UID is
obtained from the z/OS UNIX segment using mvslogin. There is
no transparent access across z/OS NFS server restart. An mvslogin
is required
- safexp
- SAF checking and exports list checking. For z/OS UNIX files, checks
UNIX permission bits. The UID is obtained from the z/OS UNIX segment
using mvslogin. There is no transparent access across z/OS
NFS server restart. An mvslogin is required
The security attribute
default is security(safexp,safexp, safexp).
|
- setgid(POSIX | ZOSUNIX)
- z/OS NFS Server
uses POSIX rules in GID inheritance for new z/OS UNIX objects. If
the S_ISGID bit of the parent directory is on, the new GID is set
to the GID of the parent directory. Otherwise, it is set from the
GID of the process. A new directory inherits the S_ISGID bit value
of the parent directory.
- POSIX
- z/OS NFS Server uses POSIX rules in GID inheritance for new z/OS
UNIX objects. If the S_ISGID bit of the parent directory is on, the
new GID is set to the GID of the parent directory. Otherwise, it is
set from the GID of the process. A new directory inherits the S_ISGID
bit value of the parent directory.
This is the default value.
- ZOSUNIX
- z/OS NFS Server provides compatibility with z/OS UNIX. When the
RACF profile FILE.GROUPOWNER.SETGID in the UNIXPRIV class is set,
z/OS NFS Server uses POSIX rules, as stated previously. Otherwise,
a new GID is always set to the GID of the parent directory, and for
a new directory, the S_ISGID bit is always set off.
Note: Some NFS
clients (such as, SUN and AIX) force GID setting after object creation
and prevent compatibility with z/OS UNIX even though the setgid(ZOSUNIX)
attribute is set.
|
- sfmax(n)
- Specifies the maximum
size (in kilobytes) of allocated storage for all of the side files.
The value of n is an integer from 0 to 2000.
The default value is 0 and it signifies that no mapping
is allowed on the NFS server. If sfmax is set to 0,
specifying the sidefile keyword in the attributes data set
will cause the server to shut down and specifying the sidefile keyword
in any subsequent mount commands causes the mount to fail because
mapping is not allowed on the NFS server. If the amount of storage
specified cannot be obtained during server initialization then the
server will shut down immediately.
|
- smf(level[,switch])
- Specifies
the level of SMF support and defines whether or not to start SMF record
collection at the NFS server startup.
The following level options
can be specified: - none
- No SMF records are to be produced.
- all
- All SMF NFS type 42 records are to be produced.
- userfile
- Both user session and file usage SMF records are to be produced.
Alternately a list of levels
( subtype_list) delimited by commas, can be specified. In this
mode of specification, at least one of the subtype levels (user,
file, audit) must be specified, and the remaining levels are optional.
- file
- Produces file usages SMF records (subtype 7).
- user
- Produces user session SMF records (subtype 8).
- audit
- Produces file creation, removal, and rename records (subtype 26).
The following switch options
can be specified: - off
- Activation of SMF records collection can be done manually by issuing
the modify command. The switch parameter
is optional.
- on
- Activates SMF records collection at the NFS server startup.
The full syntax of the smf attribute
follows: smf(none|all||userfile | subtype_list[,on|off])
An
example of the smf attribute follows: smf(user,on)
An
example follows that shows the smf attribute when the value
of switch is off: smf(user)
|
- stringprep
- Specifies that
z/OS NFS server is to enable stringprep normalization. Stringprep
normalization is the NFS version 4 globalization function for converting
inbound strings to UTF-8 format.
- nostringprep
- Specifies that z/OS NFS server is to not enable stringprep normalization.
|
- symresolve
- Specifies
that the z/OS NFS server is to resolve a symbolic link (symlink) found
in an EXPORT or CHECKLIST entry and add it to the in-memory EXPORT
or CHECKLIST list. The new entry is created in memory only. This
option only applies to NFSv4 LOOKUP in z/OS UNIX space when a symlink
is found within an EXPORT entry.
Note: - Only absolute paths are supported; symlinks pointing to relative
paths are not supported.
- If the path of a symlink is changed, an EXPORTFS command must
be run to allow z/OS NFS Server to reinterpret the new symlink path
at the next mount.
- For effects of using the showmount command,
see Using commands on the z/OS NFS client.
- nosymresolve
- Specifies that z/OS NFS server is not to resolve a symlink found
in an EXPORT or CHECKLIST entry.
|