#!/bin/bash
#################################################################
# Author: Alessandra Forti
# Created: 14/01/2012
#################################################################

# Local settings 
fs_path=/raid
head_node=bohr3226.tier2.hep.manchester.ac.uk
dry_run='yes'

# Variable necessary to the DPM admin tools
export PATH=/opt/lcg/bin:$PATH
export PYTHONPATH=/opt/lcg/lib64/python
export DPNS_HOST="$head_node"
export DPM_HOST=$DPNS_HOST

#########################################
host=`hostname`
date_ext=`date +%Y%m%d`

# Creating DB files list
echo "Creating DB files list"
db_file=db-files-${date_ext}.txt
dpm-list-disk -s $host |sed -r 's/.*://g'|sed -r 's/ .*//g' |grep -v Found> db.tmp

# Checking if there are any entries if not something gone wrong exit
if [ `grep $fs_path db.tmp |wc -l` != 0 ] ; then
    sort db.tmp > $db_file
else
    echo "dpm-list-disk failed."
    exit
fi

# Creating FS files list
echo "Creating FS files list"
fs_file=fs-files-${date_ext}.txt

# Checking mountpoint in case by some freak accident it's not mounted
mountpoint -q $fs_path
if [ $? -ne 0 ] ; then
    echo "The path $fs_path is not mounted"
    exit
fi

find $fs_path -name '*' -type f > fs.tmp

# Checking if there are any entries if not something really wrong with the FS
if [ `grep $fs_path fs.tmp |wc -l` -ne 0 ] ; then
    sort fs.tmp > $fs_file
else
    echo "I can't get the file list in $fs_path"
    exit
fi

# Diffing the lists
echo "Creating diff-dump-fs.txt"
diff_file=diff-db-fs-${date_ext}.txt
diff -bB $db_file $fs_file | grep -v lost+found |egrep '>|<' > $diff_file

# Creating each separate list of orphane DB entries and orphan files
echo "Creating indb-not-ondisk.txt and ondisk-not-indb.txt"
grep '<' $diff_file | sed "s/< /$host:/g" > indb-not-ondisk.txt
grep '>' $diff_file | sed 's/> //g' > ondisk-not-indb.txt

# Cleaning up orphan files 
echo "Cleaning up orphan files"
for a in `cat ondisk-not-indb.txt`; do 

    # Dry run?
    if [ "$dry_run" == "yes" ]; then
	echo "To be deleted $dpns_path"
    elif [ "$dry_run" == "no" ]; then
	echo "Deleting $dpns_path"
	rm -f $a
    else
	echo "$dry_run option not understood"
    fi

done

# Cleaning up orphan DB entries
echo "Cleaning up orphan DB entries"

surl_file=`hostname -s`-surl-lost-${date_ext}.txt
for a in `cat indb-not-ondisk.txt`; do
  
  dpns_path=`dpm-sql-pfn-to-dpns $a | awk '{print $2}' | sed -r 's%/$%%'`
  
  # Create SURL file for atlas
  surl="srm://$head_node${dpns_path}"
  echo $surl >> $surl_file

  # Dry run?
  if [ "$dry_run" == "yes" ]; then
      echo "To be deleted $dpns_path"
  elif [ "$dry_run" == "no" ]; then
      echo "Deleting $dpns_path"
      rfrm $dpns_path
  else
      echo "$dry_run option not understood"
  fi
done

wc -l indb-not-ondisk.txt ondisk-not-indb.txt

rm -f db.tmp fs.tmp