Abstract An improved method of having an SCSI controller provide a logical connect between a plurality of host processors using a single SCSI initiator. The controller contains removable host adapters for connection to any type of host processor interface. When a host processor requests status information of a target device, that status information is passed to the requesting processor and subsequently stored in the controller and made available to all other host processors. In addition, a fault tolerant performance monitoring algorithm, executing within the controller, is scheduled to non-intrusively and periodically check the status of each SCSI target device and to store that status information for all host processors. The advantage to such a controller is that by using only a single SCSI initiator, each host processor can independently and logically connect through the controller and SCSI initiator to a target without clearing target status information for other host processors. This is particularly advantageous when adapted to function in a local area network (LAN) arrangement wherein a significant number of host processors are requesting data.
A method of a controller non-intrusively interrogating the status in each of a plurality of SCSI targets, said controller including at least one host adapter, a microprocessor, only one SCSI initiator and memory, with said memory comprising separate buffers for PM data from each SCSI target, said method comprising the steps of: issuing a test unit ready command from said controller sequentially to each of said plurality of SCSI targets; determining if the status is returned; determining if the status is a CHECK CONDITION if the status is returned; determining if the status is good if the status is not a CHECK CONDITION; storing the status in a PM buffer of said controller if the status is not good; generating an error message if the status is not returned; storing data in a PM buffer of said controller if the status is not returned; issuing a REQUEST SENSE command from said controller to said each of said plurality of SCSI targets when the status returned by said each of said plurality of SCSI targets is a CHECK CONDITION; returning of SENSE DATA from said each of said plurality of SCSI targets issued a REQUEST SENSE command; determining if a sense key of said SENSE DATA is UNIT ATTENTION; storing ATTENTION DATA at said controller when the sense key is UNIT ATTENTION; and storing said SENSE DATA and SCSI status as PM DATA in PM buffer of said controller when the sense key is not UNIT ATTENTION.
Claim 2
Independent Claim
246 words
A method of a controller non-intrusively interrogating the status in each of a plurality of SCSI targets, said controller including at least one host adapter, a microprocessor, only one SCSI initiator and memory, with said memory comprising separate buffers for PM data from each SCSI target, said method comprising the steps of: issuing a test unit ready command from said controller sequentially to each of said plurality of SCSI targets; determining if the status is returned; determining if the status is a CHECK CONDITION if the status is returned; determining if the status is good if the status is not a CHECK CONDITION; storing the status in a PM buffer of said controller if the status is not good; generating an error message if the status is not returned; storing data in a PM buffer of said controller if the status is not returned; issuing a REQUEST SENSE command from said controller to said each of said plurality of SCSI targets when the status returned by said each of said plurality of SCSI targets is a CHECK CONDITION; returning of SENSE DATA from said each of said plurality of SCSI targets issued a REQUEST SENSE command; determining if a sense key of said SENSE DATA is UNIT ATTENTION; storing ATTENTION DATA at said controller when the sense key is UNIT ATTENTION; and storing said SENSE DATA and SCSI status as PM DATA in PM buffer of said controller when the sense key is not UNIT ATTENTION.
Claim 3
Dependent Claim
38 words
A method of a controller non-intrusively interrogating the status in each of a plurality of SCSI targets according to claim 1 further comprising the step of: refraining from additional action on the status if the status is good.