scripts/check_cache.pl

   1 #!/usr/local/bin/perl
   2
   3 # check_cache.pl
   4 #
   5 # Squid-1.0 version by martin hamilton <m.t.hamilton@lut.ac.uk>
   6 # Squid-1.1 version by Bertold Kolics <bertold@tohotom.vein.hu>
   7 #
   8 # Check the Squid-1.1.x cache directory for stale objects - i.e. those
   9 # which exist on disk but aren't listed in cached's log file.
  10
  11 require "getopts.pl";
  12 &Getopts("c:drt:vh");
  13 # -c            : the full path to squid.conf
  14 # -d            : turn on debugging
  15 # -r            : actually remove stale files
  16 # -t tmpdir     : temporary directory
  17 # -v            : list stale files
  18 # -h            : print the help
  19
  20 if ($opt_h) {
  21         print "Usage: check_cache.pl -drvh -c squid.conf\n";
  22         print "\t-c the full path to squid.conf\n";
  23         print "\t-d turn on debugging\n";
  24         print "\t-r actually remove stale files\n";
  25         print "\t-t temporary directory\n";
  26         print "\t-v list stale files\n";
  27         print "\t-h print the help\n";
  28         exit;
  29 }
  30
  31 $squidconf = $opt_c || "/usr/local/squid/etc/squid.conf";
  32 open (squidconf) || die "$squidconf: $!\n";
  33 $no_cachedir = 0;
  34 $swaplog = '';
  35 $level1dirno = 16;
  36 $level2dirno = 256;
  37 while (<squidconf>) {
  38         chop;
  39         if (/^cache_dir\s+(.*)/) {
  40                 push (@cachedir, $1);
  41         } elsif (/cache_swap_log\s+(.*)/) {
  42                 $swaplog = $1;
  43         } elsif (/swap_level1_dirs/) {
  44                 $level1dirno = $1;
  45         } elsif (/swap_level21_dirs/) {
  46                 $level2dirno = $1;
  47         }
  48 }
  49 close (squidconf);
  50 push (@cachedir, '/usr/local/squid/cache') unless ($#cachedir > $[-1);
  51 $swaplog = $cachedir[0] . '/log' unless ($swaplog);
  52 $no_cachedir = $#cachedir + 1;
  53 print "$no_cachedir CACHE DIRS: ", join(' ', @cachedir), "\n" if ($opt_d);
  54 print "SWAP LOG: $swaplog\n" if ($opt_d);
  55
  56 $tmpdir = $opt_t || $ENV{TMPDIR} || "/var/tmp";
  57 chdir($tmpdir);
  58
  59 # snarf file numbers from Squid log & sort em
  60 system("cut -f1 -d' ' $swaplog |tr [a-z] [A-Z] >pl$$");
  61 system("sort -T $tmpdir pl$$ >spl$$; rm pl$$");
  62
  63 # get list of files in cache & sort em
  64 for ($i = 0 ; $i < $no_cachedir; $i++) {
  65         chdir($cachedir[i]);
  66         system("find ./ -print -type f > $tmpdir/fp$$");
  67         chdir($tmpdir);
  68 # this cut prints only the lines with 4 fields so unnecessary lines
  69 # are supressed
  70         system("cut -d'/' -f4 -s fp$$ >> cd$$ ; rm fp$$")
  71 }
  72 system("sort -T $tmpdir cd$$ >scd$$; rm cd$$");
  73
  74 # get list of objects on disk (scd$$) but not in the log (spl$$)
  75 system("comm -13 spl$$ scd$$ >comm$$; rm spl$$ scd$$");
  76
  77 chdir($tmpdir);
  78 # iterate through it
  79 open(IN, "comm$$") || die "Can't open temporary file $tmpdir/comm$$: $!";
  80 unlink("comm$$");
  81 while(<IN>) {
  82         chop;
  83         $filename = $_;
  84
  85 # calculate the full path of the current filename
  86         $fileno = hex($filename);
  87         $dirno = $fileno % $no_cachedir;
  88         $a = $fileno / $no_cachedir;
  89         $level1 = sprintf("%02X", $a % $level1dirno);
  90         $level2 = sprintf("%02X", $a / $level1dirno % $level2dirno);
  91         $filename = "$cachedir[dirno]/$level1/$level2/$filename";
  92
  93         next if -d "$filename"; # don't want directories
  94
  95         print "$filename\n" if $opt_v; # print filename if asked
  96
  97         # skip if cached file appeared since script started running
  98         if (-M $filename < 0) {
  99                 print STDERR "skipping $filename\n" if $opt_d;
 100                 next;
 101         }
 102         print "Orphan: $filename\n";
 103         unlink($filename) if $opt_r; # only remove if asked!
 104 }
 105 close(IN);