cvaffinity(1)cvaffinity(1)NAME
cvaffinity - Set an affinity type
SYNOPSIS
cvaffinity -s <key> filename
cvaffinity -l filename
cvaffinity -d filename
DESCRIPTION
cvaffinity can be used to set an affinity for a specific storage pool
on a file or directory, or list the current affinity. An affinity is
created in a storage pool through the volume configuration (see
cvfs_config(4).) It is a name, up to eight (8) characters, describing a
special media type. Use cvadmin (1) to see what affinity sets are
assigned to the configured storage pools.
If the affinity does not exist for any of the storage pools, then the
allocation will occur on the non-exclusive storage pool. If there is
no non-exclusive storage pool, an ENOSPC is returned.
USAGE -s <key> is the Affinity Key to associate with the file or direc-
tory and is defined as a 'MediaType' name in the file system
configuration. Use the program 'cvadmin' to see the Affinity
Keys active in this file system. For files with an Affinity,
new blocks allocated to that file are placed on a storage pool
with the specified Affinity. For directories with an Affinity
new files created in that directory inherit the Affinity from
the directory.
-l This option says to just list the affinity for the specified
file and exit.
-d This option says to delete the affinity from the specified file
or directory, if one exists.
<filename>
Specifies any file or directory on the targeted volume. When
the -k option (set session affinity) is specified, or no other
option is specified, the filename option is used as a read-only
reference handle in order to access the volume. When the -s,
-d, or -l options are used, the filename option specifies the
file or directory operated on.
EXAMPLES
List the affinity on the file /usr/clips/foo.
rock # cvaffinity -l /usr/clips/foo
Set this file or directory to use the storage pool that has the jmfn8
affinity type.
rock # cvaffinity -s jmfn8 /usr/clips/filename
Remove the affinity from the /usr/clips/mydir, if one is currently
assigned.
rock # cvaffinity -d /usr/clips/mydir
SEE ALSO cvfs_config(4), cvadmin(1)Xsan File System December 2005 cvaffinity(1)
Check Out this Related Man Page
lgrp_affinity_get(3LGRP) Locality Group Library Functions lgrp_affinity_get(3LGRP)NAME
lgrp_affinity_get, lgrp_affinity_set - get of set lgroup affinity
SYNOPSIS
cc [ flag ... ] file... -llgrp [ library ... ]
#include <sys/lgrp_user.h>
lgrp_affinity_t lgrp_affinity_get(idtype_t idtype, id_t id, lgrp_id_t lgrp);
int lgrp_affinity_set(idtype_t idtype, id_t id, lgrp_id_t lgrp, lgrp_affinity_t affinity);
DESCRIPTION
The lgrp_affinity_get() function returns the affinity that the LWP or set of LWPs specified by the idtype and id arguments have for the
given lgroup.
The lgrp_affinity_set() function sets the affinity that the LWP or set of LWPs specified by idtype and id have for the given lgroup. The
lgroup affinity can be set to LGRP_AFF_STRONG, LGRP_AFF_WEAK, or LGRP_AFF_NONE.
If the idtype is P_PID, the affinity is retrieved for one of the LWPs in the process or set for all the LWPs of the process with process ID
(PID) id. The affinity is retrieved or set for the LWP of the current process with LWP ID id if idtype is P_LWPID. If id is P_MYID, then
the current LWP or process is specified.
The operating system uses the lgroup affinities as advice on where to run a thread and allocate its memory and factors this advice in with
other constraints. Processor binding and processor sets can restrict which lgroups a thread can run on, but do not change the lgroup
affinities.
Each thread can have an affinity for an lgroup in the system such that the thread will tend to be scheduled to run on that lgroup and allo-
cate memory from there whenever possible. If the thread has affinity for more than one lgroup, the operating system will try to run the
thread and allocate its memory on the lgroup for which it has the strongest affinity, then the next strongest, and so on up through some
small, system-dependent number of these lgroup affinities. When multiple lgroups have the same affinity, the order of preference among
them is unspecified and up to the operating system to choose. The lgroup with the strongest affinity that the thread can run on is known
as its "home lgroup" (see lgrp_home(3LGRP)) and is usually the operating system's first choice of where to run the thread and allocate its
memory.
There are different levels of affinity that can be specified by a thread for a particuliar lgroup. The levels of affinity are the follow-
ing from strongest to weakest:
LGRP_AFF_STRONG /* strong affinity */
LGRP_AFF_WEAK /* weak affinity */
LGRP_AFF_NONE /* no affinity */
The LGRP_AFF_STRONG affinity serves as a hint to the operating system that the calling thread has a strong affinity for the given lgroup.
If this is the thread's home lgroup, the operating system will avoid rehoming it to another lgroup if possible. However, dynamic reconfig-
uration, processor offlining, processor binding, and processor set binding and manipulation are examples of events that can cause the oper-
ating system to change the thread's home lgroup for which it has a strong affinity.
The LGRP_AFF_WEAK affinity is a hint to the operating system that the calling thread has a weak affinity for the given lgroup. If a thread
has a weak affinity for its home lgroup, the operating system interpets this to mean that thread does not mind whether it is rehomed,
unlike LGRP_AFF_STRONG. Load balancing, dynamic reconfiguration, processor binding, or processor set binding and manipulation are examples
of events that can cause the operating system to change a thread's home lgroup for which it has a weak affinity.
The LGRP_AFF_NONE affinity signifies no affinity and can be used to remove a thread's affinity for a particuliar lgroup. Initially, each
thread has no affinity to any lgroup. If a thread has no lgroup affinities set, the operating system chooses a home lgroup for the thread
with no affinity set.
RETURN VALUES
Upon successful completion, lgrp_affinity_get() returns the affinity for the given lgroup.
Upon successful completion, lgrp_affinity_set() return 0.
Otherwise, both functions return -1 and set errno to indicate the error.
ERRORS
The lgrp_affinity_get() and lgrp_affinity_set() functions will fail if:
EINVAL The specified lgroup, affinity, or ID type is not valid.
EPERM The effective user of the calling process does not have appropriate privileges, and its real or effective user ID does not
match the real or effective user ID of one of the LWPs.
ESRCH The specified lgroup or LWP(s) was not found.
ATTRIBUTES
See attributes(5) for descriptions of the following attributes:
+-----------------------------+-----------------------------+
| ATTRIBUTE TYPE | ATTRIBUTE VALUE |
+-----------------------------+-----------------------------+
|Interface Stability |Evolving |
+-----------------------------+-----------------------------+
|MT-Level |MT-Safe |
+-----------------------------+-----------------------------+
SEE ALSO lgrp_home(3LGRP), liblgrp(3LIB), attributes(5)SunOS 5.10 16 Apr 2003 lgrp_affinity_get(3LGRP)