公开(公告)号：GB2496840A

公开(公告)日：2013-05-29

申请号：GB201119660

申请日：2011-11-15

Applicant: IBM

Inventor： FUENTE CARLOS FRANCISCO ,  GUNDA KALYAN CHAKRAVARTHY ,  MUELLER-FRIEDT WOLFGANG ,  CURRAN ROBERT

IPC: G06F21/60 ,  G06F3/06 ,  H04L29/06 ,  H04L29/08

Abstract: A shared storage system 100 is in communication with one or more clusters 120 of host nodes 121-124. Access to the storage system is controlled by: storing cluster membership metadata 141 in a centralized storage 140 independent of the shared storage; updating the cluster membership metadata in response to an inclusion or exclusion request from a node identifying one or more nodes to include or exclude from a cluster; and managing access to storage volumes in the shared storage in response to each update to the cluster membership metadata. The method may also include resolving conflicting requests to update the cluster membership metadata. Responding to an exclusion request may include purging the shared storage of traffic from the excluded node, and/or establishing a fence to deny future access. The storage system does not have to rely on processing and maintaining a cluster membership list per logical unit, reducing system complexity, and provides a tie-breaking function.

公开(公告)号：GB2497167B

公开(公告)日：2014-03-05

申请号：GB201219154

申请日：2012-10-25

Applicant: IBM

Inventor： SRINIVASAN SRIKANTH ,  GUNDA KALYAN CHAKRAVARTHY

IPC: G06F11/07 ,  G06F3/06 ,  G06F17/30

公开(公告)号：GB2497167A

公开(公告)日：2013-06-05

申请号：GB201219154

申请日：2012-10-25

Applicant: IBM

Inventor： SRINIVASAN SRIKANTH ,  GUNDA KALYAN CHAKRAVARTHY

IPC: G06F11/07 ,  G06F3/06 ,  G06F17/30

Abstract: A mechanism is provided for cross-allocated block repair in a mounted file system. A set of cross-allocated blocks are identified from a plurality of blocks within an inode of the mounted file system, using a block bitmap where each cross-allocated block is associated with a corresponding bit in a designated state. The set of cross-allocated blocks are repaired 628 using a user-defined repair process, which may be marking all such blocks as free. One or more of the set of cross-allocated blocks are then deallocated 630 based on the results of the repair process. This means that data corruption can be detected and fixed without taking the file system offline. The bitmap may be populated by capturing a snapshot 606 of the file system, scanning an inode range 608 and identifying blocks associated with each inode in the range 614, determining whether a block has previously been marked as allocated 616, and, if so, placing the bit associated with the block in the block bitmap in the designated state 618.