排查存储问题

非屏蔽模式脚本：

#!/bin/bash

# UNCOMMENT THE FOLLOWING LINES:
#
# TRY_COUNT=0
# MAX_MIGRATE_ATTEMPTS=100
# OSTS="fsname-OST0000_UUID"
# DIR_OR_FILE_MIGRATED="/mnt/subdir/"
# BATCH_SIZE=10
# PARALLEL_JOBS=16 # up to max-procs processes, set to 16 if client is c5n.4xlarge with 16 vcpu
# LUSTRE_STRIPING_CONFIG="-E 100M -c 1 -E 10G -c 8 -E 100G -c 16 -E -1 -c 32" # should be consistent with the existing striping setup
#

if [ -z "$TRY_COUNT" -o -z "$MAX_MIGRATE_ATTEMPTS" -o -z "$OSTS" -o -z "$DIR_OR_FILE_MIGRATED" -o -z "$BATCH_SIZE" -o -z "$PARALLEL_JOBS" -o -z "$LUSTRE_STRIPING_CONFIG" ]; then
    echo "Some variables are not set."
    exit 1
fi

echo "lfs migrate starts"
while true; do
    output=$(sudo lfs find ! -L released --ost $OSTS --print0 $DIR_OR_FILE_MIGRATED | shuf -z | /bin/xargs -0 -P $PARALLEL_JOBS -n $BATCH_SIZE sudo lfs migrate -n $LUSTRE_STRIPING_CONFIG 2>&1)
    if [[ $? -eq 0 ]]; then
        echo "lfs migrate succeeds for $DIR_OR_FILE_MIGRATED at the $TRY_COUNT attempt, exiting."
        exit 0
    elif [[ $? -eq 123 ]]; then 
        echo "WARN: Target data objects are not located on these OSTs. Skipping lfs migrate"
        exit 1
    else
        echo "lfs migrate fails for $DIR_OR_FILE_MIGRATED at the $TRY_COUNT attempt, retrying..."
        if (( ++TRY_COUNT >= MAX_MIGRATE_ATTEMPTS )); then
            echo "WARN: Exceeds max retry attempt. Skipping lfs migrate for $DIR_OR_FILE_MIGRATED. Failed with the following error"
            echo $output
            exit 1
        fi
    fi
done

屏蔽模式脚本：

# find all OSTs with usage above a certain threshold; for example, greater than or equal to 85% full 
for OST in $(lfs df -h |egrep '( 8[5-9]| 9[0-9]|100)%'|cut -d' ' -f1); do echo ${OST};done|tr '\012' ','

# customer can also just pass OST values directly to OSTS variable
OSTS='dzfevbmv-OST0000_UUID,dzfevbmv-OST0002_UUID,dzfevbmv-OST0004_UUID,dzfevbmv-OST0005_UUID,dzfevbmv-OST0006_UUID,dzfevbmv-OST0008_UUID'

nproc=<Run up to max-procs processes if client is c5n.4xlarge with 16 vcpu, this value can be set to 16>

mnt_dir_path=<mount dir, e.g. '/my_mnt'>

lfs find ${mnt_dir_path} --ost ${OSTS}| xargs -P ${nproc} -n2 lfs migrate -E 100M -c 1 -E 10G -c 8 -E 100G -c 16 -E -1 -c 32