blob: 4b7870686f4654055b9d5a2f18afa60d94e7c7cd [file] [log] [blame]
/*
* Host Resources 'wrapper' interface
* calls the per-group interfaces from 'hr_*.h'
*/
config_require(host/hr_system)
config_require(host/hr_storage)
config_require(host/hr_device)
config_require(host/hr_other)
config_require(host/hr_proc)
config_require(host/hr_network)
config_require(host/hr_print)
config_require(host/hr_disk)
config_require(host/hr_partition)
config_require(host/hr_filesys)
config_require(host/hr_swrun)
config_require(host/hr_swinst)
/* add the host resources mib to the default mibs to load */
config_add_mib(HOST-RESOURCES-MIB)
config_add_mib(HOST-RESOURCES-TYPES)
/*********************************************
*
* A few words about the design of the Host Resources
* implementation - particularly as regards the hrDevice
* group and hrDeviceIndex. This (and hrStorageIndex) make
* use of the non-consecutive nature of SNMP instance identifiers.
*
* hrDeviceIndex is structured in a 'major/minor' form,
* with the high end indicating the type of device
* (following the enumerations of hrDeviceType) and the low
* end being used to differentiate between devices of that type.
*
* The implementation of walking through the available devices
* uses a pair of arrays of functions - indexed by hrDeviceType
* These are used to perform any initialisation needed for that
* type of device, and to step through the instances of that type.
* This latter 'get_next' routing returns the hrDeviceIndex (including
* the hrDeviceType major number), or -1 if there are no further
* instances of that type.
* When all devices of a particular type have been processed, the
* initialisation function for the next device type is called,
* and so on until all devices have been proceesed.
*
* Similar arrays are used to provide type-specific functions to
* return the "common" device information (description, ID, status
* and error count), and to save any internal structures needed
* to provide these.
* A final array is used to indicate whether hrDeviceIndexes are
* returned in a random order, or strictly increasing. In the
* latter case, this allows the search for a particular index to
* terminate as soon as the 'next' device is reached, without needing
* to check the rest of them. Similarly, once a particular type of
* device has been scanned, further types will not be examined unless
* a suitable index has not yet been found.
*
* The index used for hrFSIndex is also used as hrStorageIndex,
* for those storage areas corresponding to filestore.
* Additional storage areas (such as memory or swap space) are
* distinguished by index values greater than a defined constant.
* Currently these are individually defined entries, but other
* ranges could be implemented similarly.
* If hrFSIndex was re-implemented to reflect internal identifiers,
* it would be possible to reverse the sense of the current
* implementation, with non-filestore storage having indices
* less than a defined constant, rather than greater.
*
*
* Much of the detailed implementation of this group (as opposed to
* the implementation infrastructure outlined about) is likely to
* be very system-specific.
* The initial implementation (for HP-UX 9 and Linux) should be
* regarded as a 'proof of concept' example, rather than as
* finished, releasable code. This particularly hold for the
* disk device discovery code, which is gross in the extreme,
* and should never have seen the light of day!
* Hopefully this can be ripped out and given a quiet burial as
* soon as is decently possible.
*
* Now it's up to the rest of you to hammer this into some sort of
* sensible shape.
* Dave Shield
*
*********************************************/