Servicing of Network Software Components of Nodes of a Cluster Storage System

Abstract

Described herein are method and apparatus for servicing software components of nodes of a cluster storage system. During data-access sessions with clients, client IDs and file handles for accessing files are produced and stored to clients and stored (as session data) to each node. A serviced node is taken offline, whereby network connections to clients are disconnected. Each disconnected client is configured to retain its client ID and file handles and attempt reconnections. Session data of the serviced node is made available to a partner node (by transferring session data to the partner node). After clients have reconnected to the partner node, the clients may use the retained client IDs and file handles to continue a data-access session with the partner node since the partner node has access to the session data of the serviced node and thus will recognize and accept the retained client ID and file handles.

We claim: 
     
         1 . A cluster storage system for providing data-sessions with clients that access data containers of a shared storage of the cluster storage system, the cluster storage system comprising:
 a plurality of nodes comprising a serviced node and a partner node, each node comprising a disk element configured for receiving access requests for data containers of the shared storage the serviced node comprising a serviced disk element and the partner node comprising a partner disk element;   the serviced disk element configured for:
 receiving, from a first client through a network connection, a first access request for a first data container stored on the shared storage, the first client previously being verified as having permission to connect with the serviced disk element through a connection authentication procedure, the first client having an associated first user identifier (ID); 
 using the first user ID and permission metadata associated with the first data container, performing a first access request validation procedure on the first access request to verify that the first client has permission to access the first data container; 
 after verifying that the first client has permission to access the first data:
 producing a first data container handle comprising a physical address for the first data container and performing the first access request on the first data container; 
 storing session data comprising a plurality of data container handles, including the first data container handle, and the first user ID; 
 
   a servicing module configured for servicing the serviced disk element by:
 taking the serviced disk element offline by forcing a network disconnection with each client, including the first client, currently accessing a data container using the serviced disk element; 
 beginning servicing of the serviced disk element, the servicing comprising upgrading, reinstalling, performing maintenance, or repairing the serviced disk element; 
 providing access of the session data to the partner disk element; and 
   the partner disk element configured for:
 receiving, from the first client, a second access request for the first data container, the second access request specifying the first data container handle and the first user ID; and 
 upon determining that the session data comprises a data container handle matching the first data container handle and a user ID matching the first user ID, performing the second access request on the first data container based on the first access request validation procedure on the first user ID. 
   
     
     
         2 . The cluster storage system of  claim 1 , wherein:
 the first access request comprises a virtual address of the first data container, whereby the serviced disk element translates the virtual address to a physical address of the first data container, the physical address comprising the first data container handle for use by the first client in subsequent access requests for the first data container.   
     
     
         3 . The cluster storage system of  claim 2 , wherein the first client accesses the first data container using the partner disk element without re-submitting the virtual address of the first data container. 
     
     
         4 . The cluster storage system of  claim 1 , wherein the partner disk element is further configured for:
 validating access requests received from disconnected clients using the data container handles produced by the serviced disk element and the session data.   
     
     
         5 . The cluster storage system of  claim 4 , wherein:
 the session data further comprises permission and lock state data associated with each data container handle; and   the partner disk element is configured for validating the access requests by determining permission and lock state data stored in the session data associated with the data container handles.   
     
     
         6 . The cluster storage system of  claim 1 , wherein, upon a disconnection, each client is configured for:
 retaining each data container handle; and   upon reconnection, automatically sending retained data container handles to access previously accessed data containers using the partner disk element.   
     
     
         7 . A method for servicing a cluster storage system providing data access sessions with clients that access data containers of a shared storage of the cluster storage system, the cluster storage system comprising a plurality of nodes comprising a serviced node and a partner node, each node comprising a disk element configured for receiving access requests for data containers of the shared storage, the serviced node comprising a serviced disk element and the partner node comprising a partner disk element, the method comprising:
 at the serviced disk element:
 receiving, from a first client through a network connection, a first access request for a first data container stored on the shared storage, the first client previously being verified as having permission to connect with the serviced disk element through a connection authentication procedure, the first client having an associated first user identifier (ID); 
 using the first user ID and permission metadata associated with the first data container, performing a first access request validation procedure on the first access request to verify that the first client has permission to access the first data container; 
 after verifying the first client has permission to access the first data container, producing a first data container handle comprising a physical address for the first data container and performing the first access request on the first data container; 
 sending the first data container handle to the first client; and 
 storing session data comprising a plurality of data container handles, including the first data container handle, and the first user ID; 
   providing a servicing module configured for servicing the serviced disk element by:
 taking the serviced disk element offline by forcing a network disconnection with each client, including the first client, currently accessing a data container using the serviced disk element; 
 beginning servicing of the serviced disk element, the servicing comprising upgrading, re-installing, performing maintenance, or repairing the serviced disk element; and 
 providing access of the session data to the partner disk element; and 
   wherein the partner disk element is configured for:
 receiving, from the first client, a second access request for the first data container, the second access request specifying the first data container handle and the first user ID; and 
 upon determining that the session data comprises a data container handle matching the first data container handle and a user ID matching the first user ID, performing the second access request on the first data container based on the first access request validation procedure on the first user ID. 
   
     
     
         8 . The method of  claim 7 , wherein, upon a disconnection, each client is configured for:
 retaining each data container handle; and   upon reconnection, automatically sending retained data container handles to access previously accessed data containers using the partner disk element.   
     
     
         9 . The method of  claim 8 , wherein:
 the first access request comprises a virtual address of the first data container, whereby the serviced disk element translates the virtual address to a physical address of the first data container, the physical address comprising the first data container handle for use by the first client in subsequent access requests for the first data container.   
     
     
         10 . The method of  claim 9 , wherein the first client accesses the first data container using the partner disk element without re-submitting the virtual address of the first data container. 
     
     
         11 . The method of  claim 7 , wherein the partner disk element is further configured for:
 validating access requests received from disconnected clients using the data container handles produced by the serviced disk element and the session data.   
     
     
         12 . The method of  claim 11 , wherein:
 the session data further comprises permission and lock state data associated with each data container handle; and   the partner disk element is configured for validating the access requests by determining permission and lock state data stored in the session data associated with the data container handles.