LUN management with distributed RAID controllers
View Patent ↗A system and method is described for operating a computer memory system having a plurality of controllers capable of accessing a common set of memory modules. Access to the physical storage of the memory modules may be managed by configuration logical units (LUNs) addressable by the users. The amount of memory associated with each LUN may be managed in units of memory (LMA) from a same free LMA table maintained in each controller of the plurality of controllers. A request for maintenance of a LUN may be received from any user through any controller and results in the association of a free memory area with the LUN, and the remaining controllers perform the same operation. A test for misallocation of a free memory area is performed and when such misallocation occurs, the situation is corrected in accordance with a policy.
1. A method of managing logical units in a multicontroller memory system, comprising:
providing a plurality of controllers in communication with a plurality of memory modules, each memory module having memory addresses associated with logical units and memory addresses not associated with logical units;
maintaining a same pool of memory addresses not associated with logical units at each of the plurality of controllers;
servicing a request to perform a maintenance action for a logical unit that is initiated by a controller of the plurality of controllers;
sending a request by the initiating controller that each other controller of the plurality of controllers perform a same maintenance action as the maintenance action performed at the initiating controller;
receiving at the initiating controller a status message from each other controller relating to the maintenance action; and
initiating a fault recovery procedure if at least one of the status messages is a failure message.
2. The method of claim 1 , wherein in response to receiving a request to perform the same maintenance action, each other controller receiving the request determines if the same maintenance action can be performed,
wherein the same maintenance action can be performed if the same pool of memory addresses at the other controller contains the memory addresses contained in the request for maintenance action; and
if the same maintenance action can be performed, performing the same maintenance action and returning the status message to the initiating controller indicating success; or,
If the same maintenance action cannot be performed, returning the status message to the initiating controller indicating failure.
3. The method of claim 1 , further comprising:
assigning a global sequence number to the request when the request was initiated by the initiating controller; and
using the global sequence number associated with the failure messages to determine a corrective action to be performed.
4. The method of claim 3 , wherein the global sequence number is a time stamp.
5. The method of claim 1 , wherein the maintenance action comprises a characteristic in addition to the memory addresses.
6. The method of claim 1 , wherein memory addresses assigned by the maintenance action are deleted from the same pool of memory addresses at each other controller performing the maintenance action.
7. The method of claim 1 , wherein each controller of a group of the plurality of controllers has a network interface and the initiation of the maintenance action for the logical unit is responsive to a request received by the network interface.