btrbk: action "diff": always diff between src/target subvolume, as this works identically for snapshots as well as for received backups; enhanced visual output

pull/30/head
Axel Burri 2015-01-03 21:25:46 +01:00
parent 7db7c2c485
commit 272fb6db29
1 changed files with 132 additions and 55 deletions

187
btrbk
View File

@ -86,7 +86,7 @@ sub HELP_MESSAGE
print STDERR " tree shows backup tree\n"; print STDERR " tree shows backup tree\n";
print STDERR " execute perform all backups\n"; print STDERR " execute perform all backups\n";
print STDERR " dryrun don't run btrfs commands, just show what would be executed\n"; print STDERR " dryrun don't run btrfs commands, just show what would be executed\n";
print STDERR " diff <subvol> [parent_subvol] shows new files for subvol, optionally against a parent subvolume\n"; print STDERR " diff <from> <to> shows new files for subvolume <from>, against subvolume <to>\n";
print STDERR "\n"; print STDERR "\n";
print STDERR "For additional information, see $PROJECT_HOME\n"; print STDERR "For additional information, see $PROJECT_HOME\n";
} }
@ -292,8 +292,53 @@ sub btr_subvolume_find_new($$)
my $vol = shift; my $vol = shift;
my $lastgen = shift; my $lastgen = shift;
my $ret = run_cmd("/sbin/btrfs subvolume find-new $vol $lastgen"); my $ret = run_cmd("/sbin/btrfs subvolume find-new $vol $lastgen");
ERROR "Failed to fetch modified files for: $vol" unless(defined($ret)); unless(defined($ret)) {
return $ret; ERROR "Failed to fetch modified files for: $vol";
return undef;
}
my %files;
my $parse_errors = 0;
my $transid_marker;
foreach (split(/\n/, $ret))
{
if(/^inode \S+ file offset (\S+) len (\S+) disk start \S+ offset \S+ gen (\S+) flags (\S+) (\S+)$/) {
my $file_offset = $1;
my $len = $2;
my $gen = $3;
my $flags = $4;
my $name = $5;
$files{$name}->{len} += $len;
$files{$name}->{new} = 1 if($file_offset == 0);
$files{$name}->{gen}->{$gen} = 1; # count the generations
if($flags eq "COMPRESS") {
$files{$name}->{flags}->{compress} = 1;
}
elsif($flags eq "COMPRESS|INLINE") {
$files{$name}->{flags}->{compress} = 1;
$files{$name}->{flags}->{inline} = 1;
}
elsif($flags eq "INLINE") {
$files{$name}->{flags}->{inline} = 1;
}
elsif($flags eq "NONE") {
}
else {
WARN "unparsed flags: $flags";
}
}
elsif(/^transid marker was (\S+)$/) {
$transid_marker = $1;
}
else {
$parse_errors++;
}
}
return { files => \%files,
transid_marker => $transid_marker,
parse_errors => $parse_errors,
};
} }
@ -538,7 +583,6 @@ MAIN:
my $action_execute; my $action_execute;
my $action_tree; my $action_tree;
my $action_diff; my $action_diff;
my $target_diff;
if(($command eq "execute") || ($command eq "dryrun")) { if(($command eq "execute") || ($command eq "dryrun")) {
$action_execute = 1; $action_execute = 1;
$dryrun = 1 if($command eq "dryrun"); $dryrun = 1 if($command eq "dryrun");
@ -547,13 +591,7 @@ MAIN:
$action_tree = 1; $action_tree = 1;
} }
elsif($command eq "diff") { elsif($command eq "diff") {
$action_diff = shift @ARGV; $action_diff = 1;
$target_diff = shift @ARGV;
unless($action_diff) {
ERROR "Missing subvolume argument for \"diff\" command";
HELP_MESSAGE(0);
exit 1;
}
} }
else { else {
ERROR "Unrecognized command: $command"; ERROR "Unrecognized command: $command";
@ -567,54 +605,92 @@ MAIN:
# #
if($action_diff) if($action_diff)
{ {
my $vol = $action_diff; my $src_vol = shift @ARGV;
my $detail = btr_subvolume_detail($vol); my $target_vol = shift @ARGV;
unless($detail) { exit 1; } unless($src_vol && $target_vol) {
if($detail->{is_root}) { ERROR "subvolume at \"$vol\" is btrfs root!"; exit 1; } ERROR "Missing subvolume argument for \"diff\" command";
unless($detail->{cgen}) { ERROR "subvolume at \"$vol\" does not provide cgen"; exit 1; } HELP_MESSAGE(0);
if($detail->{parent_uuid} eq "-") { ERROR "subvolume at \"$vol\" has no parent, aborting."; exit 1; }
my $info = btr_tree($vol);
my $node = $uuid_info{$detail->{uuid}};
my $target = $uuid_info{$detail->{parent_uuid}};
die unless($node->{cgen} == $detail->{cgen}); # my paranoia
unless($node->{cgen} == $target->{gen}) { # this should always match as far as i understand btrfs send -p
WARN "generation mismatch: cgen=$node->{cgen} != parent_gen=$target->{gen}";
}
my $lastgen = $detail->{cgen};
if($target_diff) {
my $target_detail = btr_subvolume_detail($target_diff);
exit 1 unless($target_detail);
$target = $uuid_info{$target_detail->{uuid}};
# check if given parent is really a parent
my $cur = $node;
while($cur->{PARENT}) {
$cur = $cur->{PARENT} || last;
last if($cur->{uuid} eq $target_detail->{uuid});
my $count++; die if($count == 1000); # just in case we parsed crappy input
}
unless($cur->{id} == $target->{id}) {
ERROR "subvolume at \"$target_diff\" is not an ancestor of \"$action_diff\"";
exit 1; exit 1;
} }
$lastgen = $target->{gen};
my $src_detail = btr_subvolume_detail($src_vol);
unless($src_detail) { exit 1; }
if($src_detail->{is_root}) { ERROR "subvolume at \"$src_vol\" is btrfs root!"; exit 1; }
unless($src_detail->{cgen}) { ERROR "subvolume at \"$src_vol\" does not provide cgen"; exit 1; }
# if($src_detail->{parent_uuid} eq "-") { ERROR "subvolume at \"$src_vol\" has no parent, aborting."; exit 1; }
my $target_detail = btr_subvolume_detail($target_vol);
unless($target_detail) { exit 1; }
unless($src_detail->{cgen}) { ERROR "subvolume at \"$src_vol\" does not provide cgen"; exit 1; }
# if($src_detail->{parent_uuid} eq "-") { ERROR "subvolume at \"$src_vol\" has no parent, aborting."; exit 1; }
my $info = btr_tree($src_vol);
my $src = $uuid_info{$src_detail->{uuid}} || die;
my $target = $uuid_info{$target_detail->{uuid}};
unless($target) { ERROR "target subvolume is not on the same btrfs filesystem!"; exit 1; }
my $lastgen;
# check if given src and target share same parent
if(ref($src->{PARENT}) && ($src->{PARENT}->{uuid} eq $target->{uuid})) {
DEBUG "target subvolume is direct parent of source subvolume";
}
elsif(ref($src->{PARENT}) && ref($target->{PARENT}) && ($src->{PARENT}->{uuid} eq $target->{PARENT}->{uuid})) {
DEBUG "target subvolume and source subvolume share same parent";
}
else {
# TODO: this rule only applies to snapshots. find a way to distinguish snapshots from received backups
# ERROR "subvolumes \"$target_vol\" and \"$src_vol\" do not share the same parents";
# exit 1;
} }
# NOTE: in some cases "cgen" differs from "gen", even for read-only snapshots (observed: gen=cgen+1)
$lastgen = $src->{cgen} + 1;
# dump files, sorted and unique # dump files, sorted and unique
my $ret = btr_subvolume_find_new($vol, $lastgen); my $ret = btr_subvolume_find_new($target_vol, $lastgen);
exit 1 unless(defined($ret)); exit 1 unless(ref($ret));
my %files;
foreach (split(/\n/, $ret)) { print "--------------------------------------------------------------------------------\n";
/\S+ \S+ \S+ \S+ \S+ \S+ \S+ \S+ \S+ \S+ \S+ \S+ \S+ \S+ \S+ \S+ (\S+)/; print "Showing changed files for subvolume:\n $target->{path} (gen=$target->{gen})\n";
$files{$1} = 1; print "\nStarting at creation generation from subvolume:\n $src->{path} (cgen=$src->{cgen})\n";
print "\nThis will show all files modified within generation range: [$lastgen..$target->{gen}]\n";
print "Newest file generation (transid marker) was: $ret->{transid_marker}\n";
print "Parse errors: $ret->{parse_errors}\n" if($ret->{parse_errors});
print "\nLegend: <flags> <gen> <size> <filename>\n";
print " +.. file accessed at offset 0 (at least once)\n";
print " .c. flags COMPRESS or COMPRESS|INLINE set (at least once)\n";
print " ..i flags INLINE or COMPRESS|INLINE set (at least once)\n";
print " <gen> file was modified in <gen> generations\n";
print " <size> file was modified for a total of <size> bytes\n";
print "--------------------------------------------------------------------------------\n";
my $files = $ret->{files};
# calculate the character offsets
my $len_charlen = 0;
my $gen_charlen = 0;
foreach (values %$files) {
my $len = length($_->{len});
my $gen = length(scalar(keys($_->{gen})));
$len_charlen = $len if($len > $len_charlen);
$gen_charlen = $gen if($gen > $gen_charlen);
} }
print "--------------------------------------------------------------------------------\n";
print "Showing diff for: $node->{path}\n"; # finally print the output
print "Using parent at : $target->{path}\n"; foreach my $name (sort keys %$files) {
print "--------------------------------------------------------------------------------\n"; print ($files->{$name}->{new} ? '+' : '.');
print "$_\n" foreach(sort keys %files); print ($files->{$name}->{flags}->{compress} ? 'c' : '.');
print ($files->{$name}->{flags}->{inline} ? 'i' : '.');
# make nice table
my $gens = scalar(keys($files->{$name}->{gen}));
my $len = $files->{$name}->{len};
print " " . (' ' x ($gen_charlen - length($gens))) . $gens;
print " " . (' ' x ($len_charlen - length($len))) . $len;
print " $name\n";
}
exit 0; exit 0;
} }
@ -689,11 +765,12 @@ MAIN:
} }
} }
if($action_execute)
{
# #
# create snapshots # create snapshots
# #
if($action_execute)
{
my $timestamp = strftime($time_format, localtime); my $timestamp = strftime($time_format, localtime);
my %snapshot_cache; my %snapshot_cache;
foreach my $job (@$jobs) foreach my $job (@$jobs)