NAME | SYNOPSIS | DESCRIPTION | RETURN VALUES | ENVIRONMENT VARIABLES | ATTRIBUTES | SUMMARY OF TRUSTED SOLARIS CHANGES | SEE ALSO | NOTES
#include <rpcsvc/nis.h>nis_result * nis_lookup(nis_name name, uint_t flags);
These functions are used to locate and manipulate all NIS+ objects (see nis_objects(3N) ) except the NIS+ entry objects. To look up the NIS+ entry objects within a NIS+ table, refer to nis_subr(3N) .
nis_lookup() resolves a NIS+ name and returns a copy of that object from a NIS+ server. nis_add() and nis_remove() add and remove objects to the NIS+ namespace, respectively. nis_modify() can change specific attributes of an object that already exists in the namespace.
These functions should be used only with names that refer to an NIS+ Directory, NIS+ Table, NIS+ Group, or NIS+ Private object. If a name refers to an NIS+ entry object, the functions listed in nis_subr(3N) should be used.
nis_freeresult()
frees all memory associated with a
nis_result
structure. This function must be called to free the memory associated with a
NIS+
result.
nis_lookup()
,
nis_add()
,
nis_remove()
, and
nis_modify()
all return a pointer to a
nis_result
structure which
must
be freed by calling
nis_freeresult()
when you have finished using it. If one or more of the objects returned in the structure need to be
retained, they can be copied with
nis_clone_object(3N)
(see
nis_subr(3N)
). To succeed,
nis_add()
,
nis_modify()
, and
nis_remove()
must inherit the
PAF_TRUSTED_PATH
attribute.
nis_lookup() takes two parameters, the name of the object to be resolved in name , and a flags parameter, flags , which is defined below. The object name is expected to correspond to the syntax of a non-indexed NIS+ name (see nis_tables(3N) ). The nis_lookup() function is the only function from this group that can use a non-fully qualified name. If the parameter name is not a fully qualified name, then the flag EXPAND_NAME must be specified in the call. If this flag is not specified, the function will fail with the error NIS_BADNAME.
The flags parameter is constructed by logically OR ing zero or more flags from the following list.
When specified, the client library will ``follow'' links by issuing another NIS+ lookup call for the object named by the link. If the linked object is itself a link, then this process will iterate until the either a object is found that is not a LINK type object, or the library has followed 16 links.
When specified, the client library will retry the lookup until it is answered by a server. Using this flag will cause the library to block until at least one NIS+ server is available. If the network connectivity is impaired, this can be a relatively long time.
When specified, the client library will bypass any object caches and will get the object from either the master NIS+ server or one of its replicas.
When specified, the client library will bypass any object caches and any domain replicas and fetch the object from the NIS+ master server for the object's domain. This insures that the object returned is up to date at the cost of a possible performance degradation and failure if the master server is unavailable or physically distant.
When specified, the client library will attempt to expand a partially qualified name by calling the function nis_getnames() (see nis_subr(3N) ) which uses the environment variable NIS_PATH .
The status value may be translated to ascii text using the function nis_sperrno() [see nis_error(3N) ].
On return, the objects array in the result will contain one and possibly several objects that were resolved by the request. If the FOLLOW_LINKS flag was present, on success the function could return several entry objects if the link in question pointed within a table. If an error occurred when following a link, the objects array will contain a copy of the link object itself.
The function nis_add() will take the object obj and add it to the NIS+ namespace with the name name . This operation will fail if the client making the request does not have the create access right for the domain in which this object will be added. The parameter name must contain a fully qualified NIS+ name. The object members zo_name and zo_domain will be constructed from this name. This operation will fail if the object already exists. This feature prevents the accidental addition of objects over another object that has been added by another process.
The function nis_remove() will remove the object with name name from the NIS+ namespace. The client making this request must have the destroy access right for the domain in which this object resides. If the named object is a link, the link is removed and not the object that it points to. If the parameter obj is not NULL , it is assumed to point to a copy of the object being removed. In this case, if the object on the server does not have the same object identifier as the object being passed, the operation will fail with the NIS_NOTSAMEOBJ error. This feature allows the client to insure that it is removing the desired object. The parameter name must contain a fully qualified NIS+ name.
The function nis_modify() will modify the object named by name to the field values in the object pointed to by obj . This object should contain a copy of the object from the name space that is being modified. This operation will fail with the error NIS_NOTSAMEOBJ if the object identifier of the passed object does not match that of the object being modified in the namespace.
Normally the contents of the member zo_name in the nis_object structure would be constructed from the name passed in the name parameter. However, if it is non-null the client library will use the name in the zo_name member to perform a rename operation on the object. This name must not contain any unquoted `.'(dot) characters. If these conditions are not met the operation will fail and return the NIS_BADNAME error code.
struct nis_result { nis_error status; struct { uint_t objects_len; nis_object *objects_val; } objects; netobj cookie; uint32_t zticks; uint32_t dticks; uint32_t aticks; uint32_t cticks; };
The status member contains the error status of the the operation. A text message that describes the error can be obtained by calling the function nis_sperrno() (see nis_error(3N) ).
The objects structure contains two members. objects_val is an array of nis_object structures; objects_len is the number of cells in the array. These objects will be freed by the call to nis_freeresult() . If you need to keep a copy of one or more objects, they can be copied with the function nis_clone_object() and freed with the function nis_destroy_object() (see nis_server(3N) ). Refer to nis_objects(3N) for a description of the nis_object structure.
The various ticks contain details of where the time was taken during a request. They can be used to tune one's data organization for faster access and to compare different database implementations (see nis_db(3N) ).
The time spent in the NIS+ service itself. This count starts when the server receives the request and stops when it sends the reply.
The time spent in the database backend. This time is measured from the time a database call starts, until the result is returned. If the request results in multiple calls to the database, this is the sum of all the time spent in those calls.
The time spent in any ``accelerators'' or caches. This includes the time required to locate the server needed to resolve the request.
The total time spent in the request. This clock starts when you enter the client library and stops when a result is returned. By subtracting the sum of the other ticks values from this value, you can obtain the local overhead of generating a NIS+ request.
Subtracting the value in dticks from the value in zticks will yield the time spent in the service code itself. Subtracting the sum of the values in zticks and aticks from the value in cticks will yield the time spent in the client library itself. Note: all of the tick times are measured in microseconds.
The client library can return a variety of error returns and diagnostics. The more salient ones are documented below.
The request was successful.
The request was successful, however the object returned came from an object cache and not directly from the server. If you do not wish to see objects from object caches you must specify the flag NO_CACHE when you call the lookup function.
The named object does not exist in the namespace.
The object returned came from an object cache that has expired . The time to live value has gone to zero and the object may have changed. If the flag NO_CACHE was passed to the lookup function then the lookup function will retry the operation to get an unexpired copy of the object.
A server for the directory of the named object could not be reached. This can occur when there is a network partition or all servers have crashed. See the HARD_LOOKUP flag.
The object returned is of an unknown type.
The server connected to was too busy to handle your request. For the add , remove , and modify operations this is returned when either the master server for a directory is unavailable or it is in the process of checkpointing its database. It can also be returned when the server is updating it's internal state. And in the case of nis_list() if the client specifies a callback and the server does not have enough resources to handle the callback.
A generic system error occurred while attempting the request. Most commonly the server has crashed or the database has become corrupted. Check the syslog record for error messages from the server.
A request was made to a server that does not serve the name in question. Normally this will not occur, however if you are not using the built in location mechanism for servers you may see this if your mechanism is broken.
Generally a fatal result. It means that the service ran out of heap space.
An attempt was made to add a name that already exists. To add the name, first remove the existing name and then add the new object or modify the existing named object.
An attempt was made to update the database on a replica server.
The object pointed to by obj is not a valid NIS+ object.
The name passed to the function is not a legal NIS+ name.
The name passed resolved to a LINK type object and the contents of the link pointed to an invalid name.
An attempt to remove an object from the namespace was aborted because the object that would have been removed was not the same object that was passed in the request.
This hard error indicates that the named directory of the table object does not exist. This occurs when the server that should be the parent of the server that serves the table, does not know about the directory in which the table resides.
The named table does not exist.
The attempted modification failed.
The name could not be completely resolved. When the name passed to the function would resolve in a namespace that is outside the NIS+ name tree, this error is returned with a NIS+ object of type DIRECTORY , which contains the type of namespace and contact information for a server within that namespace.
This fatal error indicates the RPC subsystem failed in some way. Generally there will be a syslog(3) message indicating why the RPC request failed.
See attributes(5) for descriptions of the following attributes:
ATTRIBUTE TYPE | ATTRIBUTE VALUE |
---|---|
MT-Level | MT-Safe |
To succeed,
nis_add()
,
nis_modify()
, and
nis_remove()
must inherit the
PAF_TRUSTED_PATH
attribute.
You cannot modify the name of an object if that modification would cause the object to reside in a different domain.
You cannot modify the schema of a table object.
NAME | SYNOPSIS | DESCRIPTION | RETURN VALUES | ENVIRONMENT VARIABLES | ATTRIBUTES | SUMMARY OF TRUSTED SOLARIS CHANGES | SEE ALSO | NOTES