source: gpfs_3.1_ker2.6.20/lpp/mmfs/samples/gpfsrecover.sample @ 86

#!/bin/ksh
# @(#)40  1.1  src/avs/fs/mmfs/ts/config/gpfsrecover.sample, mmfs, avs_rgpfs24, rgpfs240610b 10/31/05 10:18:06 
##############################################################################
#
# When properly installed, this script is invoked by the GPFS daemon during
# node failure recovery before any distributed locks held by failed nodes
# are released.  This script is invoked once for each file system on all
# nodes on which the file system is mounted or otherwise in use (e.g., due
# to a pending file system management command).
#
# The script is invoked with the following parameters:
#  - The first parameter is the name of the file system.
#  - The second parameter is the "recovery phase".  In this release,
#    the recovery phase parameter will always be zero.
#  - The remaining parameters are the node numbers of all failed nodes
#    for which recovery is being run.
#
# Note: Should additional node failures occur while node failure recovery
# is in progress, this script may be invoked more than once per file system.
# In this case, the list of failed nodes will not necessarily be the same for
# all file systems for which this script is invoked, and multiple invocations
# for the same file system may contain non-disjoint sets of failed nodes.
# Hence, all actions invoked by this script should be idempotent.
#
# Example: Say there are two file systems, gpfs1 and gpfs2, and nodes 10, 11,
# and 12 fail, in that order.  This may result in the following sequence of
# calls to this script:
#   gpfsrecover gpfs1 0 10
#   gpfsrecover gpfs2 0 10 11
#   gpfsrecover gpfs1 0 10 11
#   gpfsrecover gpfs2 0 10 11 12
#   gpfsrecover gpfs1 0 12
# The only guarantee is that for each file system and each failed node, this
# script will be invoked at least once with that file system and node as
# parameters.
#
# IMPORTANT: This script is invoked synchronously, i.e., gpfs recovery will
# not progress until this script finishes.  Therefore, it should not invoke
# any long-running actions, and it MUST NOT attempt to access any files or
# directories in a gpfs file system.
#
#
# To activate the code:
#
#   a) edit this script and make the appropriate changes
#   b) copy this script to /var/mmfs/etc/gpfsrecover
#   c) ensure this script is executable (chmod +x /var/mmfs/etc/gpfsrecover)
#
##############################################################################

# Get the input parameters.
fsname=$1
phase=$2
shift 2
failednodes=$*


return 0