#!/usr/bin/perl -w
# NEMo Version Information
$VERSION='1.0.0';
use Sys::Hostname;

$TcpdumpFlags = "-vvv -n -tt -u -s 1500";
$TcpdumpFilter = "port 2049 and src or dst ";

%NfsV3Ops = ("null", 0,
             "getattr", 1,
             "setattr", 2,
             "lookup", 3,
             "access", 4,
             "readlink", 5,
             "read", 6,
             "write", 7,
             "create", 8,
             "mkdir", 9,
             "symlink", 10,
             "mknod", 11,
             "remove", 12,
             "rmdir", 13,
             "rename", 14,
             "link", 15,
             "readdir", 16,
             "readdirplus", 17,
             "fsstat", 18,
             "fsinfo", 19,
             "pathconf", 20,
             "commit", 21);
 
# Verify license
&VerifyLicense;

# Grab the input parameters
&ParseInput;



print "Starting NEMo Collector with arguments:\n";
print "         Output directory   : $OutputDir\n";
print "         Max data size (MB) : $MaxDataSize\n";
print "         Duration (min)     : $Duration\n";
print "         Network            : $NetworkInterface\n";
print "         Server             : $ServerIPAddress\n";
print "         Tcpdump v$TcpdumpVersion       : $Tcpdump\n";
print "         Nfsstat            : $Nfsstat\n";


sleep 1;

$TimeStr = time();
# convert max data size into bytes
$MaxDataSize *= 1000000;

# convert duration into seconds 
$Duration *= 60; 


$host = hostname;

# Grab nfsstat pre-profile before launching tcpdump

system("$Nfsstat > $OutputDir/nemo_nfsstat_$host\_$TimeStr\_$ServerIPAddress\_pre.log");

# direct STDERR (for both parent and child process to dedicated log file.
open (STDERR, ">$OutputDir/nemo_tcpd_$host\_$TimeStr\_$ServerIPAddress.log") 
                          || die "Cant redirect stderr";
# record arguments to tcpdump log file
   print STDERR "Starting NEMo Collector with arguments:\n";
   print STDERR "         Output directory   : $OutputDir\n";
   print STDERR "         Max data size (MB) : $MaxDataSize\n";
   print STDERR "         Duration (min)     : $Duration\n";
   print STDERR "         Network            : $NetworkInterface\n";
   print STDERR "         Server             : $ServerIPAddress\n";
   print STDERR "         Tcpdump v$TcpdumpVersion       : $Tcpdump\n";
   print STDERR "         Nfsstat            : $Nfsstat\n";
   print STDERR "\n\n\n";
   print STDERR "Tcpdump raw output follows:\n";


# launch tcpdump 

$pid  = open(TRACELOGOUT, "-|");

if ($pid == 0) #child
{
   system("$Tcpdump  -i $NetworkInterface $TcpdumpFlags $TcpdumpFilter $ServerIPAddress"); 

} else {       #parent

   $i = 0;
   $size = 0;
   $total_size = 0;
   open (OUTFILE, ">$OutputDir/nemo_workload_$host\_$TimeStr\_$ServerIPAddress\_$i.log")
                          || die "Cant open workload file";


   #print "I am parent\n";
   print "Launching tcpdump..\n";

   $RunTime = 0;
   $PrevTime = time();
   $Timeout = 60; 
   $SIG{'ALRM'} = \&EvalTimeout;
   $SIG{'INT'} = \&CleanUp;
   
   alarm $Timeout;

   while ( <TRACELOGOUT> ) {
       # tcpdump versions 3.8 and 3.9 appear to generate the same format
       # output under the specified options, so use the same tokenize 
       # routine.
       if (($TcpdumpVersion eq 3.8) ||
          ($TcpdumpVersion eq 3.9)) {
           &Tokenize_3_8($_);
       }
       # tcpdump versions 3.6 and 3.7 appear to generate the same format
       # output under the specified options, so use the same tokenize 
       # routine.
       if (($TcpdumpVersion eq 3.6) ||
	   ($TcpdumpVersion eq 3.7)) {
           &Tokenize_3_6($_);
       }
       $line_size = length $_;
       if ($line_size > 0) {
           print OUTFILE $_;
           $size += $line_size;
           if ($size > 100000000) {
               #Passed the 100MB mark, switch files
               close (OUTFILE);
               $i++;
               $TimeStr = time();
               open (OUTFILE, ">$OutputDir/nemo_workload_$host\_$TimeStr\_$ServerIPAddress\_$i.log")
                          || die "Cant open workload file";
               $size = 0;
           }
           $total_size += $line_size;
           if (($total_size) > $MaxDataSize) {
               last;
           }
       }
   }

   system("kill -SIGINT $pid");
   &ReportEnd;
   close (OUTFILE);
}


# Collector usage
sub Usage 
{
  print "Usage\n";
  print "$0\n";
  print "\t --ifc  NetworkInterface\n";
  print "\t --ip   FilerIPAddr\n";
  print "\t --tcpdump  TcpdumpPath\n";
  print "\t --nfsstat  NfsstatPath\n";
  print "\t [--dir OutputDir]\n"; 
  print "\t [--duration Duration(Min)]\n";
  print "\t [--maxdata MaxDataSize(MB)]\n";
  print "\t [--help]\n";
  print "\t [--version]\n";

  print "\n\n";

  print " --ifc: Interface to listen on. Ex. eth0 \n";
  print " --ip: Filer IP address.\n";
  print " --tcpdump: Pathname of local tcpdump utility. Ex. /usr/sbin/tcpdump \n";
  print " --nfsstat: Pathname of local nfsstat utility. Ex. /usr/sbin/nfsstat \n";
  print " --dir: Directory to which all output files are sent.\n";
  print "        Default is /tmp\n";
  print " --duration: Max number of minutes to run collector\n";
  print "        Default is 60 min\n";
  print " --maxdata: Maximum quantity of data gathered by collector\n";
  print "        Default is 100MB\n";
}

# Version Information
sub Version
{
  local $_=$0;
  s/.*?\///;
  print "$_ $VERSION\n";
}

# Verify license
sub VerifyLicense
{
    print "Do you accept the terms in the Gear6 software license agreement (Gear6_license.txt) (Yes/No): ";
    chomp ($LicenseAccept= <STDIN>);
    if (($LicenseAccept ne "Yes") && ($LicenseAccept ne "yes")) {
        print "Unable to continue unless license terms are accepted.\n";
        exit;
    }
}


# Parse input parameters 
sub ParseInput
{
    # Set up defaults

    $OutputDir = "/tmp";
    $MaxDataSize = 100;
    $Duration = 60;

    $ServerIPAddress = "";
    $NetworkInterface = "";
    $Tcpdump = "";
    $Nfsstat = "";

    # Read all available arguments
    print "\nProcessing arguments..\n";

    $i = 0;
    while ($i < $#ARGV+1)
    {
        if ($ARGV[$i]  eq "--ifc") {
            $NetworkInterface = $ARGV[$i+1];
            $i = $i+2;
            next;
        }
        if ($ARGV[$i] eq "--ip") {
            $ServerIPAddress = $ARGV[$i+1]; 
            $i = $i+2;
            next;
        }
        if ($ARGV[$i] eq "--maxdata") {
            $MaxDataSize = $ARGV[$i+1]; 
            $i = $i+2;
            next;
        }
        if ($ARGV[$i] eq "--duration") {
            $Duration = $ARGV[$i+1]; 
            $i = $i+2;
            next;
        }
        if ($ARGV[$i] eq "--tcpdump") {
            $Tcpdump = $ARGV[$i+1]; 
            # Locate tcpdump and determine version
            &VerifyTcpdump;
            $i = $i+2;
            next;
        }
        if ($ARGV[$i] eq "--nfsstat") {
            $Nfsstat = $ARGV[$i+1]; 
            # Locate nfsstat
            &VerifyNfsstat;
            $i = $i+2;
            next;
        }
        if ($ARGV[$i] eq "--dir") {
            $OutputDir = $ARGV[$i+1]; 
            &VerifyOutputDirectory;
            $i = $i+2;
            next;
        }
        if ($ARGV[$i] eq "--version") {
           &Version;
           exit; 
        }
        &Usage;
        exit;
    }
    
    #Query for any required arguments that are missing

    if ($ServerIPAddress eq "") {
        print "Filer IP address: ";
        chomp ($ServerIPAddress = <STDIN>);
        if ($ServerIPAddress eq "") {
            print "Unable to continue without a filer IP address\n";
            exit;
        }
    }
    if ($NetworkInterface eq "") {
        print "Network Interface (eth0): ";
        chomp ($NetworkInterface = <STDIN>);
        if ($NetworkInterface eq "") {
            print "Using eth0 as default\n";
            $NetworkInterface = "eth0";
        }
    }
    if ($Tcpdump eq "") {
        print "tcpdump path (/usr/sbin/tcpdump): ";
        chomp ($Tcpdump = <STDIN>);
        if ($Tcpdump eq "") {
            print "Using /usr/sbin/tcpdump as default\n";
            $Tcpdump = "/usr/sbin/tcpdump";
        }
        # Locate tcpdump and determine version
        &VerifyTcpdump;
    }
    if ($Nfsstat eq "") {
        print "nfsstat path (/usr/sbin/nfsstat): ";
        chomp ($Nfsstat = <STDIN>);
        if ($Nfsstat eq "") {
            print "Using /usr/sbin/nfsstat as default\n";
            $Nfsstat = "/usr/sbin/nfsstat";
        }
        # Locate nfsstat
        &VerifyNfsstat;
    }
}

# Make sure the output directory is on a local file system
sub VerifyOutputDirectory
{
  if (!(-d $OutputDir)) {
      print "$OutputDir is not a directory\n\n";
      &Usage;
      exit;
  }
  
  $stat_cmd = `which stat`;
  chomp($stat_cmd);
  
  if ( -e $stat_cmd) { 
      $stat_output = `$stat_cmd -f $OutputDir`;
      if ($stat_output =~ /nfs/) {
          print "$OutputDir is on an NFS file system\nPlease provide a directory on a local file system\n";
          exit;
      } 
  } else {
      print "Unable to find stat command: Will not verify that output directory is on a local file system\n";
  }
   
}


# Locate tcpdump and determine version

sub VerifyTcpdump 
{

  if (!(-e $Tcpdump)) {
     print "Unable to locate tcpdump at $Tcpdump\n";
     exit;
  } else {
     $vers = `$Tcpdump --version 2>&1`;
     $vers =~ /tcpdump version (\d.\d)/;
     $TcpdumpVersion = $1; 

     # Crosscheck tcpdump version against supported versions
     if (($TcpdumpVersion eq 3.6) || 
         ($TcpdumpVersion eq 3.7) ||
         ($TcpdumpVersion eq 3.8) ||
         ($TcpdumpVersion eq 3.9)) {
        print "Tcpdump version $TcpdumpVersion supported\n";
     } else {
        print "Tcpdump version $TcpdumpVersion not supported\n";
        exit;
     }
  }

}

# Locate nfsstat

sub VerifyNfsstat
{
  if (!(-e $Nfsstat)) {
     print "Unable to locate nfsstat at $Nfsstat\n";
     exit;
  } 
}


# Check to see if we've run for enough time

sub EvalTimeout 
{
  $timeval = time();
  $RunTime = $RunTime + $timeval - $PrevTime;
  $PrevTime = $timeval;
  #print "Timer fired: Runtime is $RunTime\n";
  if ($RunTime >= $Duration) {
     system("kill -SIGINT $pid");
     &ReportEnd;
     close (OUTFILE);
     exit;
  } else {
     alarm $Timeout;
  }
}

# Report time run and data gathered

sub ReportEnd 
{
  # Grab Nfsstat post profile before ending
  $TimeStr = time();
  system("$Nfsstat > $OutputDir/nemo_nfsstat_$host\_$TimeStr\_$ServerIPAddress\_post.log");
  
  $RunTime = $RunTime/60.0;
  $total_size = $total_size/1000000.0;
  print "NEMo Collector: Run ending after $RunTime min with $total_size MB data\n";
  close (OUTFILE);
  exit;
}


# End tcpdump run if we are interrupted..
sub CleanUp
{
  system("kill -SIGINT $pid");
  # Grab Nfsstat post profile before ending
  $TimeStr = time();
  system("$Nfsstat > $OutputDir/nemo_nfsstat_$host\_$TimeStr\_$ServerIPAddress\_post.log");

  $RunTime = $RunTime/60.0;
  $total_size = $total_size/1000000.0;
  print "NEMo Collector: interrupted .. Run exiting after $RunTime min with $total_size MB data\n";
  close (OUTFILE);
  exit;
}


# Filter tcpdump output
# Tokenize routine for tcpdump version 3.8
sub Tokenize_3_8
{

  # If this line contains no NFS data, remove it entirely
  if ($_[0] =~ /: \. \[tcp/) {
      #print STDERR "Removing line: $_[0]\n";
      $_[0] = "";
      return;
  }

  # If this is a tcpdump packet misinterpret, remove it also
  if ($_[0] =~ /proc-|bad-len/) {
      $_[0] = "";
      return;
  }

 
  # Remove IP information from all remaining lines
  $_[0] =~ s/(IP \(.*\) )//;
  
  # Remove source and destination IP addresses
  $_[0] =~ s/( \d+\.\d+\.\d+\.\d+\.2049[ :][ >])//;
  $_[0] =~ s/(\d+\.\d+\.\d+\.\d+\.)(\d+)([ :][ >])/$2 /;
 
  # Remove filenames from lookup/remove/create. We will 
  # keep the request itself, to include in request type 
  # distribution during static analysis

  if ($_[0] =~ /lookup|remove|create/) {
      $_[0] =~ s/("[^\s]+")//;
  }
 
  # Replace NFS op names with op numbers
  if ($_[0] =~ /\d+.\d+ \d+ reply ok \d+ \w+/ ) {
      if ($_[0] =~ /getattr|setattr|read |write|remove/) {
          $_[0] =~ s/(\d+.\d+ \d+) reply ok (\d+) (\w+) /$1 $2 o $NfsV3Ops{$3} /;
      } else { #Remove additional reply text
          $_[0] =~ s/(\d+.\d+ \d+) reply ok (\d+) (\w+) (.*)/$1 $2 o $NfsV3Ops{$3} /;
      }
  } else {
      if ($_[0] =~ /\d+.\d+ \d+ reply ERR \d+/) {
         $_[0] =~ s/(\d+.\d+ \d+) reply ERR (\d+)/$1 $2 e/; 
      } else {
         if ($_[0] =~ /getattr|setattr|read |write|remove/) {
             $_[0] =~ s/(\d+.\d+ \d+) (\d+) (\w+) /$1 $2 r $NfsV3Ops{$3} /;
         } else { #Remove additional arguments past file handle
             if (!($_[0] =~ s/(\d+.\d+ \d+) (\d+) (\w+) ([^\s]+) (.*)/$1 $2 r $NfsV3Ops{$3} $4/)) {
                $_[0] = "";  #Remove any lines that dont conform to at least
                             #one of these formats
             }
         }
      }
  }
   
  # Remove redundant text from requests/replies
  $_[0] =~ s/\@ |ids |bytes |sz |nlink |rdev |fsid |fileid |a\/m\/ctime |bytes \@ |fh|max |verf //g;
 
  return; 
}

# Filter tcpdump output
# Tokenize routine for tcpdump version 3.6
sub Tokenize_3_6
{

  # If this line contains no NFS data, remove it entirely
  if ($_[0] =~ /: \. \[tcp/) {
      #print STDERR "Removing line: $_[0]\n";
      $_[0] = "";
      return;
  }

  # If this is a tcpdump packet misinterpret, remove it also
  if ($_[0] =~ /proc-|bad-len/) {
      $_[0] = "";
      return;
  }

 
  # Remove extraneous information from all remaining lines
  $_[0] =~ s/( \(DF\) \(.*\))//;
  
  # Remove source and destination IP addresses
  $_[0] =~ s/( \d+\.\d+\.\d+\.\d+\.2049[ :][ >])//;
  $_[0] =~ s/(\d+\.\d+\.\d+\.\d+\.)(\d+)([ :][ >])/$2 /;
 
  # Remove filenames from lookup/remove/create. We will 
  # keep the request itself, to include in request type 
  # distribution during static analysis

  if ($_[0] =~ /lookup|remove|create/) {
      $_[0] =~ s/("[^\s]+")//;
  }
 
  # Replace NFS op names with op numbers
  if ($_[0] =~ /\d+.\d+ \d+ reply ok \d+ \w+/ ) {
      if ($_[0] =~ /getattr|setattr|read |write|remove/) {
          $_[0] =~ s/(\d+.\d+ \d+) reply ok (\d+) (\w+) /$1 $2 o $NfsV3Ops{$3} /;
      } else { #Remove additional reply text
          $_[0] =~ s/(\d+.\d+ \d+) reply ok (\d+) (\w+) (.*)/$1 $2 o $NfsV3Ops{$3} /;
      }
  } else {
      if ($_[0] =~ /\d+.\d+ \d+ reply ERR \d+/) {
         $_[0] =~ s/(\d+.\d+ \d+) reply ERR (\d+)/$1 $2 e/; 
      } else {
         if ($_[0] =~ /getattr|setattr|read |write|remove/) {
             $_[0] =~ s/(\d+.\d+ \d+) (\d+) (\w+) /$1 $2 r $NfsV3Ops{$3} /;
         } else { #Remove additional arguments past file handle
             if (!($_[0] =~ s/(\d+.\d+ \d+) (\d+) (\w+) ([^\s]+) (.*)/$1 $2 r $NfsV3Ops{$3} $4/)) {
                $_[0] = "";  #Remove any lines that dont conform to at least
                             #one of these formats
             }
         }
      }
  }
   
  # Remove redundant text from requests/replies
  $_[0] =~ s/\@ |ids |bytes |bytes|sz |nlink |rdev |fsid |fileid |a\/m\/ctime |bytes \@ |fh|Unknown|max |verf //g;
 
  return; 
}