Autobackup.pl

From MythTV Official Wiki
Revision as of 09:24, 28 November 2012 by PhilB (Talk | contribs)

Jump to: navigation, search

Important.png Note: The correct title of this article is autobackup.pl. It appears incorrectly here due to technical restrictions.


Author Phil Brady
Description Maintains a backup subset of recordings, taking into account age and whether they are part of a series.
Supports Version25.png  



Description

autobackup.pl selects a subset of mythtv recordings and makes a backup of them to a destination partition which is typically smaller than the total size of all recordings and which is on a separate disk. It allows recovery of service with a 'useful' set of recordings in case of primary disk failure. It should be augmented by appropriate database and system backups.

Options

--preview or -p
to inhibit file copy/deletes. This shows what action would be taken with a full run but without changing the filestore.
--loglevel n to set logging levels.
n=0 for only errors
n=1 for calling info,
n=2 for stats,
n=3 for a recordings list
n=4 for protocol diagnostics
Default is 3.
--help or -h Help text
--delete <filename>
deletes file from destination directory. For use with the recording delete system event, though use is optional.

Details

autobackup.pl is intended to be run as a regular job when the system is otherwise idle. It checks:

  • the destination partition to determine how full it is,
  • the database to extract information about recordings
  • the source directories as a consistency check
  • The destination directory to identify recordings which have alreay been backed up and others which are obsolete and need removing.

It is then in a state to identify series recordings, allocate priorities and determine the action for each file. Files are selected for backup in date/time order (newest first) but:

  • recordings marked as non-expirable have a 20 year priority boost
  • recordings which are recognised as episodes of a series can be bunched together with the latest episode and/or have a number of days added as a boost. The author finds bunching with zero days boost to be satisfactory.

Three or more recordings with the same name and an average interval between them of less than 8 days are deemed to be a series.

Files will be promoted from the head of the list, and files demoted as necessary from the end in order to prevent over-filling the partition.

The script should handle recording groups with more than one source directory, though there has only been limited testing of the code for this.

The script was designed to be run automatically at system closedown and to maintain a permanently fitted destination partition, though it should backup satisfactorily to a mounted external eSATA or USB drive if started manually.


Partition Limits

New recordings will be promoted and old ones demoted whilst observing the partition limit of 90%. Older recordings will only be considered if the partition is below a partition threshold of 85% full and this is only likely happen after a spate of manual deletions. The two limits prevent a recording from repeatedly falling in and out of favour as higher priority recordings are created and removed. Threshold should be a few days worth of recordings below limit for stability. The order in which files are copied over by the scipt is chosen for robustness of the algorithm even after an unexpected closedown during copying.


Deleted recordings

Recordings can optionally be deleted from the destination directory with the RecordingDelete system event mechanism by calling this script with --delete %FILE%. This will remove the file more quickly.

However, such files will be routinely deleted by this script even if the event system is not used. Recordings deleted routinely by the script have the advantage of appearing in the log file, complete with recording name, so giving a fuller picture of the change in the state of files.

Recordings which have been deleted from the database but still remain on the destination partition no longer have their name in the database. In order to give these a name in listings, the script writes a list of today's secure recordings to the file to enable the names to be established next day. The file is /var/lib/db_backups/autobackup.list. This is placed in the directory which is used to hold backups and control files from the sister database backup routine autobackup.sh.


Working parameters

See the parameters (%params) at the head of the program defining parameters such as the destination partition, maximum percentage used, source directories, destination directories, default logging level, series action, threshold setting and name of recordings list file.

On the author's system, the script is placed in /usr/bin/autobackup.pl. Logging is directed by autobackup.sh to /var/log/mythtv/autobackup.log The source directory is /var/lib/mythtv/recordings and the destination /var/lib/mythtv2/recordings The destination partition is /var/lib/mythtv2; the limit is 90% and the threshold 85%.


Logging

If log files are required then they need to be redirected from STDOUT and STDERR and rotated as necessary by a calling script.


Database backups

The perl script has a sister routine autobackup.sh which takes a database backup at last closedown of the day. The sister routine will trigger a recordings backup if it finds this perl script present.


Performance

The time taken by this script is dominted by file copying time. With SATA2 interfaces 6Gb/min is a rule of thumb. If the script is run daily, it will typically run in under a minute.


Glossary

The listing includes these terms:

del deleted from database by the user but still present on the backup partition.

dem or demoted selected for removal from the backup partiion in order to ake way for a promotion.

insecure or ins a file which has not been backed up and does not merit it.

promoted or pro selected for copying to the backup partition.

secure or sec a file which is already securely backed up and still merits that.

ser a recording deemed to be one of a series and given a priority boost.

vip a recording marked in the database as non expirable.


Development regime

The script has been developed and tested in the UK on a 64 bit Mythbuntu 12.04 (Mythtv 0.25) dedicated frontend/backend system with ACPI__Wakeup, DVB-T receivers, a single directory for recordings, and a local time close to or equal to GMT.

A 500Gb disk holds root, swap and backup recording partitions; a 2Tb drive holds emergency root and swap and the primary recording partition /var/lib/mythtv.

The intention is to reboot from the 2Tb and run a database restore should the 500Gb fail, and to edit fstab and run Find_orphans.py should the 2Tb fail..

This system has been developed with version 72 of Myth Protocol and it will need validating carefully against other implementations. See Myth Protocol/Guide. To do so, you will need to add a new version and its key in %MythProtocolVersion and test thoroughly with --loglevel 4 and then with --preview.


Example logging

This shows the statistics produced, and a (severely cut) list of recordings. The error has been deliberately introduced. See glossary for meaning of status.

2012-11-27 20:55:06 /usr/bin/autobackup.pl --preview
Version 1.0  26 Nov 2012. running with loglevel=3, preview=yes

Partition /var/lib/mythtv2 info
          size                      452.50G
    88%   used                      401.03G
    90%   limit                     407.25G
    85%   threshold                 384.63G

Database scan found 240 recordings

Source directory /var/lib/mythtv/recordings
      3   vip recordings              4.67G
    100   series recordings         161.08G
    138   expirable recordings      246.29G
      0   obsolete recordings         0.00G
    241   --- total ---             412.04G

There were 1 database inconsistencies.
Suggest using find_orphans.py

Destination directory /var/lib/mythtv2/recordings
      3   vip recordings              4.67G
    100   series recordings         161.08G
    131   expirable recordings      233.50G
      0   obsolete recordings         0.00G
    234   --- total ---             399.26G

Changes proposed:
      0   delete obsolete             0.00G
      0   demote                      0.00G
      1   promote                     0.91G

Proposed final state of partition:
    88%   used                      401.94G

All files (priority order) Status    Size   Name (if known)
-----------------------------------------------------------
1704_20121113205700.mpg    Sec Vip   0.4Gb  All in the Mind
1003_20121127192600.mpg    Pro       0.9Gb  The Martin Lewis Money Show
1001_20121126223600.mpg    Sec       1.5Gb  Have I Got a Bit More News for You
1001_20121111205600.mpg    Sec Ser   2.3Gb  Andrew Marr's History of the World
1001_20121104205600.mpg    Sec Ser   2.1Gb  Andrew Marr's History of the World
1001_20121028205600.mpg    Sec Ser   2.2Gb  Andrew Marr's History of the World
1002_20121027212600.mpg    Sec       1.3Gb  The Thick of It
1704_20121026182600.mpg    Sec       0.4Gb  The News Quiz
1003_20121025223300.mpg    Sec       0.8Gb  Corfu
1003_20111028195600.mpg    Sec       0.8Gb  Wild Britain With Ray Mears
1009_20111024225600.mpg    Ins       2.4Gb  The Secret Life of Waves
1002_20110620205800.mpg    Ins       2.3Gb  Made in Britain
1009_20110322225600.mpg    Ins       2.2Gb  The Secret World of Whitehall

Finished: time taken    0 seconds

                                                                  

The Script

In order to use this script, you will need to:

  • create a destination partition and directory (I understand top level directories are not usable)
  • set permissions to allow access by root and mythtv.
  • Ensure you can access it too if you wish to test it.
  • Copy this code to /usr/bin/autobackup.pl (or elsewhere) and chmod + x it.
  • Do test with --preview first!
  • Research protocol versions and test toroughly if necessary.
  • trigger it daily - see autobackup.sh as a possible mechanism.


Application-x-perl.png autobackup.pl

#!/usr/bin/perl -w
use strict;
use IO::Socket;
use Time::Local;

#
# Select recordings for backup to a small partition and copy/delete as necessary.
# See --help for details
#
# Written by Phil Brady  November 2012.
#

#program version
use constant VERSION  => '1.0  26 Nov 2012';

#control parameters
my %params =(
  "partition"	=> "/var/lib/mythtv2",
  "destdir"		=> "/var/lib/mythtv2/recordings",

  "sourcedir1"	=> "/var/lib/mythtv/recordings",  #first or only dir in recording group
# "sourcedir2"	=> "/var/lib/mythtv/other",	#add others like this. Note sourcedir1 to n, not 0 to n

  "percent"		=> 90,   	# max used% of partition
  "threshold"	=> 85,		# percentage for older files to be promoted.
  "SeriesPolicy"=>  1,		# if set, use priority of newest member of a series.	]  Use either, neither
  "SeriesBoost"	=>	0,		# boost in days for series.  							]  or both of these boosts.
  "preview"		=>  0,		# command line parameter -p or --preview.
  "filelist"	=>	"/var/lib/mythtv/db_backups/autobackup.list", #list recordings here so we can identify deleted ones next day
);


# default --loglevel parameter
my $loglevel=3;		# 0=errors only, 1=calling params, 2=stats, 3=recordings details, 4=protocol diagnostic trace.


# backend parameters
use constant myipaddress	=> '127.0.0.1';
use constant backendip		=> '127.0.0.1';
my $sock;
my %MythProtocolVersion	= (
	'72' => 'D78EFD6F',
	);     #add other versions here but do sanity check with loglevel 4 before use.


#constants for extracting details from program record
use constant recstart        => 1;  #start of first record
use constant reclength       => 44; #length of each record
use constant recname         => 0;  #name of program
use constant recfilename     => 10;
use constant recsize         => 11;  #recording size
use constant recflags        => 27;  #flags
use constant    flagautoexp       => 4;
use constant    flagdeletepending => 128;	#this bit utilised but never seen set!
use constant    flagpriority      => 1024;  #not used.
use constant recgroup		 => 28;		#more accurate indicator of pending delete

my $gig= 1024*1024*1024;

#info on recorded programs. Hashes all use filename as key.
my %frecname;   #program name 
my %fsize;      #file size
my %fstatus;    #recording status: 0=obsolete/deletepending,1=normal,2=series record, 3=vip recording
                 # +4 =secure before, +8=secure after.
my %fpriority;  #hold merit for saving the recording
my %newest;	 #time of most recent episode in series
my @deleteq;    #list of files for deletion
my @promoteq;   #ditto promotion

#statistics
my %stats;

#---------ok, let's start!---------

&parse_command_line;

&check_partition;

&scan_database;	#to get list of recordings

&mark_series;	#identify series recordings and mark episodes

#scan source directories and check consistency with database
&scansources;

#scan destination directory.  Identify which recordings are backed up and find obsolete ones. 
&scandir($params{'destdir'},1);

&maintain_list('read');		#obsolete recordings are anonymous.  Get names by checking yesterday's list

&decide_priorities;			#based on expirability, obsolescence, series episode and recording time.

#now sort them.  
my @sorted = sort {$fpriority{$b} <=> $fpriority{$a} or			#priority
              $frecname{$b} cmp $frecname{$a} or   				#keeps series episodes together
              substr($b,5,14) <=> substr($a,5,14)} keys %fstatus;	#in date order
my @sorted2=@sorted;

&decide_action;				#decide promotions/demotions
							#note:  operation depends heavily on allocated priorities & sort order.

&do_action;					#now do the copies/deletes

&showall if $loglevel==3;	#list all known recordings

&maintain_list('write');	#make new list so we can identify deleted recordings tomorrow

#almost done!
printf "\nFinished: time taken %4d seconds\n",time-$^T  if $loglevel;
exit 0;

#-------------------
sub maintain_list{

#	Called with read or write.

#	This routine writes a list of saved recording file names and program names to a file.
#	It checks the list the following day to put a name to deleted recordings which 
#	would otherwise be anonymous (named '??').

#	If you are using the recordingdelete event to remove backup recordings then you will probably
#	not see these anonymous recordings.  

#	To remove this functionality, uncomment the next line
#	return;

#	Documentation/reminder:  Myth delete mechanism seems to be:
#	1. changes group to 'Deleted'
#	2. Some while later, backend deletes file and removes entry from database. 
#	3. Deletepending flag not seen in use.

	if ($_[0] eq 'read'){

		#check yesterday's list to name deleted recordings
		my $nameyesterday;
		if ($loglevel != 3){return};			#nobody listening!
		if ($stats{'noname'}==0){return};		#no anonymous ones to check
		unless (open FH, "<$params{'filelist'}"){ 
			print "Warning:  cannot read $params{'filelist'} $!\n";
			return;   #not fatal
		}
		while (<FH>){
			chomp;		
			($_,$nameyesterday)=split /\[\]:\[\]/,$_;
			if (exists $frecname{$_}){
				if ($frecname{$_} eq '??'){$frecname{$_} = ".$nameyesterday"};
			}
		}
		close FH;

	}elsif ($_[0] eq 'write'){
		if ($params{'preview'}) {return};

		#write list of backed up recordings for use tomorrow
		unless (open FH, ">$params{'filelist'}"){
			print "Cannot write $params{'filelist'} $!" if $loglevel >1;
			return;
		} 
		for (keys %fstatus){
			if (($fstatus{$_} & 11) > 8){
				#this is a backed up file so remember for tomorrow
				print FH $_ . '[]:[]' . $frecname{$_}."\n";
			}
		} 
		close FH;
	}else{
		print "bad maintain_list $_[0]\n";
		exit 1;
	}
}
#-------------------
sub decide_priorities{
	my $st;
	foreach (keys %fstatus){
		$st=($fstatus{$_} & 3);
		if ($st==0){
			$fpriority{$_}=0;
		}elsif ($st==3){
			$fpriority{$_} +=7305;		#20 years boost for vip=non expirable
		}elsif ($st==2){				#boost episodes of series
			if ($params{'SeriesPolicy'}){
				$fpriority{$_}=$newest{$frecname{$_}};
			}
			$fpriority{$_} += $params{'SeriesBoost'};
		}
	}
}
#-------------------
sub timestamp{
	#extract timestamp from filename - return epoch DAYS as a real.

	(my $in)=@_;	#filename
	$in=substr($in,5,14);
	my @bits=($in =~ m/../g);
	my $out = timelocal($bits[6],$bits[5],$bits[4],$bits[3],$bits[2]-1,$bits[0].$bits[1]);
	$out/86400;
}
#-------------------
sub mark_series{

	#identify series of recordings; mark the episodes

	my %namecount;
	my $tmp;
	my %oldest;	# %newest is global as needed later in decide_priorities.


	#count instances of names, keep newest/oldest
	foreach (keys %frecname){
		$tmp=&timestamp($_);
		$fpriority{$_}=$tmp;		
		$_=$frecname{$_};

		#keep count, oldest, newest for each name found
		$namecount{$_}++;
		if (exists $newest{$_}) {
			if ($tmp < $oldest{$_}) {$oldest{$_}=$tmp};
			if ($tmp > $newest{$_}) {$newest{$_}=$tmp};
		}else{
			$oldest{$_}=$tmp;
			$newest{$_}=$tmp;
		}
	}

	#mark as series if more than 2 episodes and average interval no more than 8 days.
	my $count;
	my $avinterval;
	foreach (keys %namecount){
		$count=$namecount{$_};
		$avinterval=0;
		if ($count>1){$avinterval=($newest{$_}-$oldest{$_})/($count-1)};
		if (($count>2) && ($avinterval<8)){
				#this is a series
				$namecount{$_}=-1;
		}
	}

	#mark the recordings as series episodes
	foreach (keys %frecname){
		if ($namecount{$frecname{$_}} <0) {
			if ($fstatus{$_}==1){$fstatus{$_}=2}; 
		}
	}
}
#--------------------
sub do_action{

	if ($params{'preview'}){return 0};
	my $command;
	my $qsize;
	my $fname;

	#do deletes first
	$qsize=@deleteq;
	print "\nNow deleting $qsize files\n" if $loglevel>2;
	if ($qsize){   
		foreach $_ (@deleteq){
			$command= "rm -f $params{'destdir'}/$_";
			system $command;
		}
	}

	#now do promotions
	$qsize=@promoteq;
	print "Now promoting $qsize files\n" if $loglevel>2;
	if ($qsize==0){return};

	#count the source directories
	my $dirct=0;
	while (exists $params{'sourcedir'.++$dirct}){
	}
	$dirct--;

	#do copies
	for $fname (@promoteq){
		for (1..$dirct){		
			if ( -e "$params{'sourcedir'.$_}/$fname"){
				$command="cp -p $params{'sourcedir'.$_}/$fname $params{'destdir'}";
				#print "$command\n";				
				system $command;
				last;
			}
		}
	}
	
}
#-------------------
sub decide_action{

	$stats{'promote'}=0;
	$stats{'promote size'}=0;
	$stats{'demote'}=0;
	$stats{'demote size'}=0;

	#queue obsolete files for deletion
	my $finished=0;
	while ($finished==0){
		if (@sorted==0){     #array empty
			$finished=1;
		}else{
			$_=pop @sorted;
			if ($fstatus{$_} & 3){         #not obsolete - we overshot!
				push @sorted,$_;          #put it back
				$finished=1;
			}elsif ($fstatus{$_} ==12){   #obsolete file for deletion
				push @deleteq,$_;
				$fstatus{$_} -=8;         #demoted
			}
		}
	}

	$stats{'used size'} -=$stats{'obsolete recordings size'};
	#report
	$stats{'delete obsolete'}=$stats{'obsolete recordings'};    #cleaner name for user
	$stats{'delete obsolete size'}=$stats{'obsolete recordings size'};

	#now decide promotions/deletion
	my $stale=0;   #become stale once we find a secure normal recording
	while ($_ = shift @sorted){   #newest
		if ($fstatus{$_}==0){       #skip deletions
		}elsif ($fstatus{$_} <4){    #an unsaved recording - try to save it
			if (&freespace($fsize{$_},$stale)) {
				#space made - mark for promotion
				if ($stale){				#put in order: oldest fresh, other fresh, newest stale, oldest stale
					push @promoteq,$_;
				}else{
					unshift @promoteq,$_;
				}
				$fstatus{$_} +=8;
				$stats{'promote'}++;
				$stats{'promote size'} +=$fsize{$_};
				$stats{'used size'} +=$fsize{$_};
			}
		}elsif ($fstatus{$_} == 13){
        	#non vip already backed up.
			$stale=1;
		}
	}

	&show_stats('Changes proposed:','delete obsolete','demote','promote');
	&show_stats('Proposed final state of partition:','used');
}    

#--------------------
sub freespace{
	#free up space by removing oldest secured files
	#return 1 if successful, 0 if not
	my ($needed,$stale)=@_;

	#check against lower threshold first
	if ($needed < ($stats{'threshold size'} - $stats{'used size'})) {return 1};
	if ($stale){return 0};

	#now against higher
	my $loopstate = -1;
	while ($loopstate == -1){
		if ($needed < ($stats{'limit size'} - $stats{'used size'})) {
			$loopstate=1;   #will fit now
		}else{
			if (@sorted){
				my $d = pop(@sorted);
				if ($fstatus{$d} >12) {
					#demote this one
					push @deleteq,$d;
					$fstatus{$d} -=8;
					$stats{'demote'}++;
					$stats{'demote size'} += $fsize{$d};
					$stats{'used size'} -= $fsize{$d};
				}	
			}else{
				#no files left to remove
				$loopstate=0;
			}
		}
	}
	$loopstate;
}
#---------------------------
sub opensocket{
	print "Opening socket to backend\n" if $loglevel>=4;
	unless ($sock = IO::Socket::INET->new(PeerAddr => backendip,
		PeerPort => 6543, 
		Proto => 'tcp',
		timeout => 25) ) {
			print "cannot open socket to backend.\n";
			print "Wrong backend IP address?  Backend not running?\n";
			exit 0;
	}
}
#---------------------------
sub scan_database {

	#probe for protocol version
	&opensocket;
	print "Probing for protocol version\n" if $loglevel>=4;
	my $r= &getreply("MYTH_PROTO_VERSION 1");
	unless ($r =~ /^REJECT/){
		die "Odd response from MYTH_PROTO_VERSION";
	}
	my $VersionFound=substr($r,11,);
	close($sock);
	unless (exists $MythProtocolVersion{$VersionFound}){
		die "Mythprotocol version not known or validated";
	}

	#now open properly
	&opensocket;
	$r= &getreply("MYTH_PROTO_VERSION ". $VersionFound .' '. $MythProtocolVersion{$VersionFound});
	unless (substr($r,0,6) eq 'ACCEPT') {
		die "ERROR:  Wrong protocol version:  $r";
	}

	print "ANNouncing - if no reply, suspect myipaddress\n" if $loglevel>=4;
	$r= &getreply('ANN Monitor '.myipaddress . " 0");
	unless (substr($r,0,2) eq 'OK') {
		die "ERROR:  ANN rejected:  $r";
	}

	print "Getting list of recordings\n" if $loglevel>=4;
	$r= &getreply('QUERY_RECORDINGS Unsorted');
	my $sep = qr/\[\]:\[\]/;
	my @recs = split($sep, $r);

	$stats{'DBrecordings'}=$recs[0];  #no of recordings
	print "\nDatabase scan found $recs[0] recordings\n" if $loglevel>1;

	#extract info from recordings
	my ($fname, $fstatus);
	for (my $i= recstart; $i <= $#recs; $i += reclength) {
		(undef, undef, undef, $fname)=split /\//, $recs[$i+recfilename];
		$frecname{$fname} = $recs[$i+recname];
		$fsize{$fname}=$recs[$i+recsize];
		my $flags=$recs[$i+recflags];		
		if ($recs[$i+recgroup] eq 'Deleted'){    #just deleted
			$fstatus{$fname}=0;
		}elsif (($flags & flagdeletepending)>0){
			$fstatus{$fname}=0;    #Delete pending flag (not used?)
		}elsif (($flags & flagautoexp)==0){
			$fstatus{$fname}=3;    #non expirable so vip
		}else{
			$fstatus{$fname}=1;    #expirable
		}
		if ($loglevel==4){
			print "$fname $flags $frecname{$fname}\n";
		}
	}

	close($sock);
	if ($loglevel==4){
		print "Scan finishing because loglevel set to 4 for protocol diagnostics.\n";
		exit 0;
	}
}

#------------------
sub getreply {
	my ($command) = @_;
	my ($length, $ret, $bytes, $data);
	my $com = length($command). ' ' x (8 - length(length($command))). $command;

	if ($loglevel==4){print "sending $com\n"};
	print $sock $com;

	#read length of reply
	read($sock, $length, 8); 
	$length = int($length);

	# Read and return any data that was returned
	$ret = '';
	while ($length > 0) {
		$bytes = read($sock, $data, ($length < 8192 ? $length : 8192)); 
		last if ($bytes < 1);   # EOF
		# On to the next
		$ret .= $data;
		$length -= $bytes;
	}
	print "Reply starts:". substr($ret,0,200)."\n" if $loglevel >=4;
	return $ret;
}


#-------------------------
sub check_partition{
	my $reply = `df $params{'partition'}`;
	if (length $reply == 0){die 'cannot find partition'};
	my @fields = split /\s+/,$reply;
	$stats{'partition size'}=$fields[8]*1024;
	$stats{'used size'}=$fields[9]*1024;
	$stats{'used'} = sprintf "%2d%%", $stats{'used size'} *100 / $stats{'partition size'};
	$stats{'limit size'} = ($stats{'partition size'} * $params{'percent'} /100);
	$stats{'limit'}= $params{percent} . '%';
	$stats{'size size'}=$stats{'partition size'};  #same but more readable in output
	$stats{'threshold'}=$params{'threshold'} .'%';
	$stats{'threshold size'}= $params{'threshold'} * $stats{'partition size'}/100;
	&show_stats ("Partition $params{'partition'} info", 'size', 'used', 'limit','threshold');
}

#----------------------------

sub show_stats{
# print item, number & size
	if ($loglevel<2) {return 0};
	my @items = @_;
	$_=shift @items;
	if ($_) {print "\n$_\n"};    #optional heading
	my $item_value;
	while (@items){
		$_=shift @items;
		if (exists $stats{$_}) {
			if ($_ eq 'used'){
				#update used percentage
				$stats{'used'} = sprintf "%2d%%", $stats{'used size'} *100 / $stats{'partition size'};
			}
			$item_value=$stats{$_};
			#show item and number
			if ($item_value =~/%/){
				printf "%7s",$item_value;
			}else{
				printf "%7d",$item_value;
			}
		}else{
			print "       ";
		}
		printf "   %-25s",$_;

		#show size if present
		$_ .= ' size';
		if (exists $stats{"$_"}) {
			$item_value= $stats{$_};
			printf "%7.2fG",$item_value/$gig;
		}
		print "\n";
	}
}
#------------
sub showall{
	#show all recordings
#	print "\nfile status is Deleted, Secure, Insecure, Promote or Demote followed by vip (non expirable), series or deleted.\n"; 
	print "\nAll files (priority order) Status    Size   Name (if known)\n";
	print "-----------------------------------------------------------\n";   
	foreach (@sorted2){
		printf "$_  %9s   %3.1fGb  $frecname{$_}\n",&statusname($fstatus{$_}),$fsize{$_}/$gig;
	}
}
#------------
sub scansources{
	#scan source directories, report only.
	$stats{'missing'}=0;
	$stats{'found'}=0;

	my $i=0;
	while (exists $params{'sourcedir'.++$i}){
		&scandir($params{'sourcedir'.$i},0);
	}

	#counts consistent with d/base?
	$i=$stats{'DBrecordings'}-$stats{'found'};
	if ($i<0){$i=-$i};
	$i += $stats{'missing'};	
	if (($i>0) && ($loglevel>1)) {
		print "\nThere were $i database inconsistencies.\n";
		print "Suggest using find_orphans.py\n";
	}
}
#-------------
sub scandir{
	my ($dir, $destination)=@_;
	my ($size, $category);
	#scan a folder, show stats for it
 
	$stats{'vip recordings'}=0;
	$stats{'vip recordings size'}=0;
	$stats{'expirable recordings'}=0;
	$stats{'expirable recordings size'}=0;
	$stats{'obsolete recordings'}=0;
	$stats{'obsolete recordings size'}=0;

	$stats{'series recordings'}=0;
	$stats{'series recordings size'}=0;
	$stats{'noname'}=0;

	opendir DH,$dir or die "Cannot open $dir: $!";
	foreach $_ (readdir DH) {
		if (/^\d{4}_\d{14}\.mpg$/) {
			if (exists $fstatus{$_}) {
				$stats{'found'}++;
				if ($destination){
					$fstatus{$_} +=12;
				}              
			}else{
				$stats{'missing'}++;
				if ($destination){
					$fsize{$_}= -s "$dir\/$_";
					$fstatus{$_} = 12;
					$frecname{$_}='??';
					$stats{'noname'}++;
				}
			}
			
			#collect stats
			if (exists $fstatus{$_}){      
				$size=$fsize{$_};
				$_=($fstatus{$_} & 3);
				$category=qw/obsolete expirable series vip/[$_];
				$stats{$category.' recordings'}++;
				$stats{$category.' recordings size'} += $size;
			}
		}
	}
	closedir DH;
  
	#summary
	$stats{'--- total ---'}=$stats{'series recordings'}+
		$stats{'expirable recordings'} + $stats{'vip recordings'} + $stats{'obsolete recordings'};
	$stats{'--- total --- size'}=$stats{'series recordings size'}
		+$stats{'expirable recordings size'} + $stats{'vip recordings size'} + $stats{'obsolete recordings size'};
	$_=$destination?'Destination':'Source';
	&show_stats("$_ directory $dir",'vip recordings', 'series recordings', 'expirable recordings','obsolete recordings','--- total ---');
}
   
#------------
sub statusname{
	#return text showing status 
	(my $st)=@_;
	my @ftype=('Del','   ','Ser','Vip');
	my $st2=($st & 12)/4;
	qw/Ins Dem Pro Sec/[$st2] .' '. $ftype[$st & 3];
}

#------------------
sub parse_command_line{
	if (scalar @ARGV){
		my @argcopy=@ARGV;
		while (@argcopy){
			$_ = shift @argcopy;
			if (/^--preview$|^-p$/) {
				$params{'preview'}=1;

			}elsif (/^--loglevel$/) {
				my $bad=1;
				if (scalar @argcopy){
					$loglevel=shift @argcopy;
					if ($loglevel =~/\d/){
						if (($loglevel>=0) && ($loglevel<5)) {$bad=0};
					}
				}
				if ($bad){
					print "bad loglevel needs 0 to 4\n";
					exit 0;
				}
			}elsif (/^--help$|^-h$/){
				&givehelp;
				exit 0;
			}elsif (/^--delete$/){
				#recording delete event
				$_ = "$params{'destdir'}/" . shift @argcopy;
				if ( -e $_){system "rm -f $_"};
				exit 0;
			}else{
				print "parameter $_ not recognised.  Try $0 --help\n";
				exit 0;
			}
		}
	}
	if ($loglevel){
		$_= `date +%F" "%T`;
		chomp;
		print "\n$_ $0 @ARGV\n";
		print	"Version ". VERSION . ". running with loglevel=$loglevel, preview=". qw/no yes/[$params{'preview'}] ."\n";
	}

}
#------------
sub givehelp{

print "

NAME:      autobackup.pl

SYNOPSIS:  autobackup.pl [options]

DESCRIPTION:

	autobackup.pl selects a subset of mythtv recordings 
	and makes a backup of them to a destination partition which 
	is typically smaller than the total size of all recordings
	and which is on a separate disk.  It allows recovery of 
	service with a 'useful' set of recordings in case of 
	primary disk failure. 

	It should be augmented by appropriate database and system
	backups.

OPTIONS:

	--preview or -p
		to inhibit file copy/deletes.  This shows what 
		action would be taken with a full run.

	--loglevel n	to set logging levels.  
				1 for calling info, 
				2 for stats, 
				3 for a recordings list
				4 for protocol diagnostics 
				Default is 3.

	--help or -h		this text

	--delete <filename>
		deletes file from destination directory.  For use with
		recording delete system event, though use optional.

DETAILS:

	autobackup.pl is intended to be run as a regular job at system
	closedown when the system is otherwise idle.  It checks 
	the destination partition to determine how full it is, 
	then scans the database and the source and destination 
	directories to check the state of all recordings.
	It is then in a state to identify series recordings, 
	allocate priorities and determine the action for each file.

	Files are selected for backup in date/time order (newest 
	first) but: 

	-	recordings marked as non-expirable have a 20 year 
		priority boost

	-	recordings which are recognised as episodes of a series 
		can be bunched together with the latest episode
		and/or have a number of days added as a boost.
		The author uses bunching with zero days boost.

	Three or more recordings with the same name and an average 
	interval between them of less than 8 days are deemed to 
	be a series.

	Files will be promoted from the head of the list, and files 
	demoted as necessary from the end in order to prevent 
	over-filling the partition.

PARTITION LIMITS

	New recordings will be promoted whilst observing the partition
	limit of $params{'percent'}%.  Older recordings will only be considered if the 
	partition is below a partition threshold of $params{'threshold'}% full.  
	This is only likely happen after a spate of manual deletions.  

	The two limits prevent a recording from falling in and out
	of favour as higher priority recordings are created and removed.
	Threshold should be a few days worth of recordings below limit.

DELETED RECORDINGS

	Recordings can optionally be deleted from the destination directory 
	with the RecordingDelete system event mechanism by calling this program 
	with --delete %FILE%.  This will remove the file more quickly.

	However, such files will be routinely deleted by this program if the event
	system is not used.
 
	Recordings which have been deleted from the database but still 
	remain on the destination partition no longer have their name in
	the database.  In order to give these a name in listings, the script 
	writes a list of today's recordings to the file to enable the names 
	to be established next day.  The file is 
	$params{'filelist'} 
	 
WORKING PARAMETERS

	See the parameters (\%params) at the head of the program defining 
	parameters such as the destination partition, percentage used,
	source and destination directories, default logging level, series 
	action, threshold setting and name of recordings list.

LOGGING

	If log files are required then they need to be redirected from STDOUT
	and STDERR and rotated as necessary by a calling script eg autobackup.sh

 
DATABASE BACKUPS

	The perl script has a sister routine autobackup.sh which
	takes a database backup and triggers this script at last
	closedown of the day.  See autobackup.sh.

 
DEVELOPMENT REGIME

	The script has been developed and tested in the UK on a 
	Mythbuntu 12.04 dedicated frontend/backend system with 
	ACPI wakeup, DVB-T receivers, a single directory for 
	recordings, and a local	time close to or equal to GMT.  

	A 500Gb disk holds root, swap and backup recording partitions.
	A 2Tb drive holds emergency root and swap and the primary 
	recording partition /var/lib/mythtv.

	Assumptions may well have been inadvertently made which are 
	not valid in all circumstances and the author can accept 
	no responsibility for the results of this script.
	In particular, this system has been developed with version 
	72 of mythtv protocol and it will need validating carefully 
	against other implementations.  

	To use with other versions you will need to add a new version
	and its key in \%MythProtocolVersion and test thoroughly 
	with --loglevel 4 and then with --preview.

 
AUTHOR:

	Written by Phil Brady, November 2012.


COPYRIGHT

	Copyright   ©   2012 Phil Brady.
	License  GPLv3+: GNU GPL version 3 or later
	See http://gnu.org/licenses/gpl.html.

	This is free software: you are free to change and redistribute 
	it.  There is NO WARRANTY, to the extent permitted by law.

";

exit 0;
}