#!/usr/bin/env perl # This program is part of Percona Toolkit: http://www.percona.com/software/ # See "COPYRIGHT, LICENSE, AND WARRANTY" at the end of this file for legal # notices and disclaimers. use strict; use warnings FATAL => 'all'; use constant PTDEBUG => $ENV{PTDEBUG} || 0; # ########################################################################### # OptionParser package # This package is a copy without comments from the original. The original # with comments and its test file can be found in the Bazaar repository at, # lib/OptionParser.pm # t/lib/OptionParser.t # See https://launchpad.net/percona-toolkit for more information. # ########################################################################### { package OptionParser; use strict; use warnings FATAL => 'all'; use English qw(-no_match_vars); use constant PTDEBUG => $ENV{PTDEBUG} || 0; use List::Util qw(max); use Getopt::Long; my $POD_link_re = '[LC]<"?([^">]+)"?>'; sub new { my ( $class, %args ) = @_; my @required_args = qw(); foreach my $arg ( @required_args ) { die "I need a $arg argument" unless $args{$arg}; } my ($program_name) = $PROGRAM_NAME =~ m/([.A-Za-z-]+)$/; $program_name ||= $PROGRAM_NAME; my $home = $ENV{HOME} || $ENV{HOMEPATH} || $ENV{USERPROFILE} || '.'; my %attributes = ( 'type' => 1, 'short form' => 1, 'group' => 1, 'default' => 1, 'cumulative' => 1, 'negatable' => 1, ); my $self = { head1 => 'OPTIONS', # These args are used internally skip_rules => 0, # to instantiate another Option- item => '--(.*)', # Parser obj that parses the attributes => \%attributes, # DSN OPTIONS section. Tools parse_attributes => \&_parse_attribs, # don't tinker with these args. %args, strict => 1, # disabled by a special rule program_name => $program_name, opts => {}, got_opts => 0, short_opts => {}, defaults => {}, groups => {}, allowed_groups => {}, errors => [], rules => [], # desc of rules for --help mutex => [], # rule: opts are mutually exclusive atleast1 => [], # rule: at least one opt is required disables => {}, # rule: opt disables other opts defaults_to => {}, # rule: opt defaults to value of other opt DSNParser => undef, default_files => [ "/etc/percona-toolkit/percona-toolkit.conf", "/etc/percona-toolkit/$program_name.conf", "$home/.percona-toolkit.conf", "$home/.$program_name.conf", ], types => { string => 's', # standard Getopt type int => 'i', # standard Getopt type float => 'f', # standard Getopt type Hash => 'H', # hash, formed from a comma-separated list hash => 'h', # hash as above, but only if a value is given Array => 'A', # array, similar to Hash array => 'a', # array, similar to hash DSN => 'd', # DSN size => 'z', # size with kMG suffix (powers of 2^10) time => 'm', # time, with an optional suffix of s/h/m/d }, }; return bless $self, $class; } sub get_specs { my ( $self, $file ) = @_; $file ||= $self->{file} || __FILE__; my @specs = $self->_pod_to_specs($file); $self->_parse_specs(@specs); open my $fh, "<", $file or die "Cannot open $file: $OS_ERROR"; my $contents = do { local $/ = undef; <$fh> }; close $fh; if ( $contents =~ m/^=head1 DSN OPTIONS/m ) { PTDEBUG && _d('Parsing DSN OPTIONS'); my $dsn_attribs = { dsn => 1, copy => 1, }; my $parse_dsn_attribs = sub { my ( $self, $option, $attribs ) = @_; map { my $val = $attribs->{$_}; if ( $val ) { $val = $val eq 'yes' ? 1 : $val eq 'no' ? 0 : $val; $attribs->{$_} = $val; } } keys %$attribs; return { key => $option, %$attribs, }; }; my $dsn_o = new OptionParser( description => 'DSN OPTIONS', head1 => 'DSN OPTIONS', dsn => 0, # XXX don't infinitely recurse! item => '\* (.)', # key opts are a single character skip_rules => 1, # no rules before opts attributes => $dsn_attribs, parse_attributes => $parse_dsn_attribs, ); my @dsn_opts = map { my $opts = { key => $_->{spec}->{key}, dsn => $_->{spec}->{dsn}, copy => $_->{spec}->{copy}, desc => $_->{desc}, }; $opts; } $dsn_o->_pod_to_specs($file); $self->{DSNParser} = DSNParser->new(opts => \@dsn_opts); } if ( $contents =~ m/^=head1 VERSION\n\n^(.+)$/m ) { $self->{version} = $1; PTDEBUG && _d($self->{version}); } return; } sub DSNParser { my ( $self ) = @_; return $self->{DSNParser}; }; sub get_defaults_files { my ( $self ) = @_; return @{$self->{default_files}}; } sub _pod_to_specs { my ( $self, $file ) = @_; $file ||= $self->{file} || __FILE__; open my $fh, '<', $file or die "Cannot open $file: $OS_ERROR"; my @specs = (); my @rules = (); my $para; local $INPUT_RECORD_SEPARATOR = ''; while ( $para = <$fh> ) { next unless $para =~ m/^=head1 $self->{head1}/; last; } while ( $para = <$fh> ) { last if $para =~ m/^=over/; next if $self->{skip_rules}; chomp $para; $para =~ s/\s+/ /g; $para =~ s/$POD_link_re/$1/go; PTDEBUG && _d('Option rule:', $para); push @rules, $para; } die "POD has no $self->{head1} section" unless $para; do { if ( my ($option) = $para =~ m/^=item $self->{item}/ ) { chomp $para; PTDEBUG && _d($para); my %attribs; $para = <$fh>; # read next paragraph, possibly attributes if ( $para =~ m/: / ) { # attributes $para =~ s/\s+\Z//g; %attribs = map { my ( $attrib, $val) = split(/: /, $_); die "Unrecognized attribute for --$option: $attrib" unless $self->{attributes}->{$attrib}; ($attrib, $val); } split(/; /, $para); if ( $attribs{'short form'} ) { $attribs{'short form'} =~ s/-//; } $para = <$fh>; # read next paragraph, probably short help desc } else { PTDEBUG && _d('Option has no attributes'); } $para =~ s/\s+\Z//g; $para =~ s/\s+/ /g; $para =~ s/$POD_link_re/$1/go; $para =~ s/\.(?:\n.*| [A-Z].*|\Z)//s; PTDEBUG && _d('Short help:', $para); die "No description after option spec $option" if $para =~ m/^=item/; if ( my ($base_option) = $option =~ m/^\[no\](.*)/ ) { $option = $base_option; $attribs{'negatable'} = 1; } push @specs, { spec => $self->{parse_attributes}->($self, $option, \%attribs), desc => $para . (defined $attribs{default} ? " (default $attribs{default})" : ''), group => ($attribs{'group'} ? $attribs{'group'} : 'default'), }; } while ( $para = <$fh> ) { last unless $para; if ( $para =~ m/^=head1/ ) { $para = undef; # Can't 'last' out of a do {} block. last; } last if $para =~ m/^=item /; } } while ( $para ); die "No valid specs in $self->{head1}" unless @specs; close $fh; return @specs, @rules; } sub _parse_specs { my ( $self, @specs ) = @_; my %disables; # special rule that requires deferred checking foreach my $opt ( @specs ) { if ( ref $opt ) { # It's an option spec, not a rule. PTDEBUG && _d('Parsing opt spec:', map { ($_, '=>', $opt->{$_}) } keys %$opt); my ( $long, $short ) = $opt->{spec} =~ m/^([\w-]+)(?:\|([^!+=]*))?/; if ( !$long ) { die "Cannot parse long option from spec $opt->{spec}"; } $opt->{long} = $long; die "Duplicate long option --$long" if exists $self->{opts}->{$long}; $self->{opts}->{$long} = $opt; if ( length $long == 1 ) { PTDEBUG && _d('Long opt', $long, 'looks like short opt'); $self->{short_opts}->{$long} = $long; } if ( $short ) { die "Duplicate short option -$short" if exists $self->{short_opts}->{$short}; $self->{short_opts}->{$short} = $long; $opt->{short} = $short; } else { $opt->{short} = undef; } $opt->{is_negatable} = $opt->{spec} =~ m/!/ ? 1 : 0; $opt->{is_cumulative} = $opt->{spec} =~ m/\+/ ? 1 : 0; $opt->{is_required} = $opt->{desc} =~ m/required/ ? 1 : 0; $opt->{group} ||= 'default'; $self->{groups}->{ $opt->{group} }->{$long} = 1; $opt->{value} = undef; $opt->{got} = 0; my ( $type ) = $opt->{spec} =~ m/=(.)/; $opt->{type} = $type; PTDEBUG && _d($long, 'type:', $type); $opt->{spec} =~ s/=./=s/ if ( $type && $type =~ m/[HhAadzm]/ ); if ( (my ($def) = $opt->{desc} =~ m/default\b(?: ([^)]+))?/) ) { $self->{defaults}->{$long} = defined $def ? $def : 1; PTDEBUG && _d($long, 'default:', $def); } if ( $long eq 'config' ) { $self->{defaults}->{$long} = join(',', $self->get_defaults_files()); } if ( (my ($dis) = $opt->{desc} =~ m/(disables .*)/) ) { $disables{$long} = $dis; PTDEBUG && _d('Deferring check of disables rule for', $opt, $dis); } $self->{opts}->{$long} = $opt; } else { # It's an option rule, not a spec. PTDEBUG && _d('Parsing rule:', $opt); push @{$self->{rules}}, $opt; my @participants = $self->_get_participants($opt); my $rule_ok = 0; if ( $opt =~ m/mutually exclusive|one and only one/ ) { $rule_ok = 1; push @{$self->{mutex}}, \@participants; PTDEBUG && _d(@participants, 'are mutually exclusive'); } if ( $opt =~ m/at least one|one and only one/ ) { $rule_ok = 1; push @{$self->{atleast1}}, \@participants; PTDEBUG && _d(@participants, 'require at least one'); } if ( $opt =~ m/default to/ ) { $rule_ok = 1; $self->{defaults_to}->{$participants[0]} = $participants[1]; PTDEBUG && _d($participants[0], 'defaults to', $participants[1]); } if ( $opt =~ m/restricted to option groups/ ) { $rule_ok = 1; my ($groups) = $opt =~ m/groups ([\w\s\,]+)/; my @groups = split(',', $groups); %{$self->{allowed_groups}->{$participants[0]}} = map { s/\s+//; $_ => 1; } @groups; } if( $opt =~ m/accepts additional command-line arguments/ ) { $rule_ok = 1; $self->{strict} = 0; PTDEBUG && _d("Strict mode disabled by rule"); } die "Unrecognized option rule: $opt" unless $rule_ok; } } foreach my $long ( keys %disables ) { my @participants = $self->_get_participants($disables{$long}); $self->{disables}->{$long} = \@participants; PTDEBUG && _d('Option', $long, 'disables', @participants); } return; } sub _get_participants { my ( $self, $str ) = @_; my @participants; foreach my $long ( $str =~ m/--(?:\[no\])?([\w-]+)/g ) { die "Option --$long does not exist while processing rule $str" unless exists $self->{opts}->{$long}; push @participants, $long; } PTDEBUG && _d('Participants for', $str, ':', @participants); return @participants; } sub opts { my ( $self ) = @_; my %opts = %{$self->{opts}}; return %opts; } sub short_opts { my ( $self ) = @_; my %short_opts = %{$self->{short_opts}}; return %short_opts; } sub set_defaults { my ( $self, %defaults ) = @_; $self->{defaults} = {}; foreach my $long ( keys %defaults ) { die "Cannot set default for nonexistent option $long" unless exists $self->{opts}->{$long}; $self->{defaults}->{$long} = $defaults{$long}; PTDEBUG && _d('Default val for', $long, ':', $defaults{$long}); } return; } sub get_defaults { my ( $self ) = @_; return $self->{defaults}; } sub get_groups { my ( $self ) = @_; return $self->{groups}; } sub _set_option { my ( $self, $opt, $val ) = @_; my $long = exists $self->{opts}->{$opt} ? $opt : exists $self->{short_opts}->{$opt} ? $self->{short_opts}->{$opt} : die "Getopt::Long gave a nonexistent option: $opt"; $opt = $self->{opts}->{$long}; if ( $opt->{is_cumulative} ) { $opt->{value}++; } else { $opt->{value} = $val; } $opt->{got} = 1; PTDEBUG && _d('Got option', $long, '=', $val); } sub get_opts { my ( $self ) = @_; foreach my $long ( keys %{$self->{opts}} ) { $self->{opts}->{$long}->{got} = 0; $self->{opts}->{$long}->{value} = exists $self->{defaults}->{$long} ? $self->{defaults}->{$long} : $self->{opts}->{$long}->{is_cumulative} ? 0 : undef; } $self->{got_opts} = 0; $self->{errors} = []; if ( @ARGV && $ARGV[0] eq "--config" ) { shift @ARGV; $self->_set_option('config', shift @ARGV); } if ( $self->has('config') ) { my @extra_args; foreach my $filename ( split(',', $self->get('config')) ) { eval { push @extra_args, $self->_read_config_file($filename); }; if ( $EVAL_ERROR ) { if ( $self->got('config') ) { die $EVAL_ERROR; } elsif ( PTDEBUG ) { _d($EVAL_ERROR); } } } unshift @ARGV, @extra_args; } Getopt::Long::Configure('no_ignore_case', 'bundling'); GetOptions( map { $_->{spec} => sub { $self->_set_option(@_); } } grep { $_->{long} ne 'config' } # --config is handled specially above. values %{$self->{opts}} ) or $self->save_error('Error parsing options'); if ( exists $self->{opts}->{version} && $self->{opts}->{version}->{got} ) { if ( $self->{version} ) { print $self->{version}, "\n"; } else { print "Error parsing version. See the VERSION section of the tool's documentation.\n"; } exit 0; } if ( @ARGV && $self->{strict} ) { $self->save_error("Unrecognized command-line options @ARGV"); } foreach my $mutex ( @{$self->{mutex}} ) { my @set = grep { $self->{opts}->{$_}->{got} } @$mutex; if ( @set > 1 ) { my $err = join(', ', map { "--$self->{opts}->{$_}->{long}" } @{$mutex}[ 0 .. scalar(@$mutex) - 2] ) . ' and --'.$self->{opts}->{$mutex->[-1]}->{long} . ' are mutually exclusive.'; $self->save_error($err); } } foreach my $required ( @{$self->{atleast1}} ) { my @set = grep { $self->{opts}->{$_}->{got} } @$required; if ( @set == 0 ) { my $err = join(', ', map { "--$self->{opts}->{$_}->{long}" } @{$required}[ 0 .. scalar(@$required) - 2] ) .' or --'.$self->{opts}->{$required->[-1]}->{long}; $self->save_error("Specify at least one of $err"); } } $self->_check_opts( keys %{$self->{opts}} ); $self->{got_opts} = 1; return; } sub _check_opts { my ( $self, @long ) = @_; my $long_last = scalar @long; while ( @long ) { foreach my $i ( 0..$#long ) { my $long = $long[$i]; next unless $long; my $opt = $self->{opts}->{$long}; if ( $opt->{got} ) { if ( exists $self->{disables}->{$long} ) { my @disable_opts = @{$self->{disables}->{$long}}; map { $self->{opts}->{$_}->{value} = undef; } @disable_opts; PTDEBUG && _d('Unset options', @disable_opts, 'because', $long,'disables them'); } if ( exists $self->{allowed_groups}->{$long} ) { my @restricted_groups = grep { !exists $self->{allowed_groups}->{$long}->{$_} } keys %{$self->{groups}}; my @restricted_opts; foreach my $restricted_group ( @restricted_groups ) { RESTRICTED_OPT: foreach my $restricted_opt ( keys %{$self->{groups}->{$restricted_group}} ) { next RESTRICTED_OPT if $restricted_opt eq $long; push @restricted_opts, $restricted_opt if $self->{opts}->{$restricted_opt}->{got}; } } if ( @restricted_opts ) { my $err; if ( @restricted_opts == 1 ) { $err = "--$restricted_opts[0]"; } else { $err = join(', ', map { "--$self->{opts}->{$_}->{long}" } grep { $_ } @restricted_opts[0..scalar(@restricted_opts) - 2] ) . ' or --'.$self->{opts}->{$restricted_opts[-1]}->{long}; } $self->save_error("--$long is not allowed with $err"); } } } elsif ( $opt->{is_required} ) { $self->save_error("Required option --$long must be specified"); } $self->_validate_type($opt); if ( $opt->{parsed} ) { delete $long[$i]; } else { PTDEBUG && _d('Temporarily failed to parse', $long); } } die "Failed to parse options, possibly due to circular dependencies" if @long == $long_last; $long_last = @long; } return; } sub _validate_type { my ( $self, $opt ) = @_; return unless $opt; if ( !$opt->{type} ) { $opt->{parsed} = 1; return; } my $val = $opt->{value}; if ( $val && $opt->{type} eq 'm' ) { # type time PTDEBUG && _d('Parsing option', $opt->{long}, 'as a time value'); my ( $prefix, $num, $suffix ) = $val =~ m/([+-]?)(\d+)([a-z])?$/; if ( !$suffix ) { my ( $s ) = $opt->{desc} =~ m/\(suffix (.)\)/; $suffix = $s || 's'; PTDEBUG && _d('No suffix given; using', $suffix, 'for', $opt->{long}, '(value:', $val, ')'); } if ( $suffix =~ m/[smhd]/ ) { $val = $suffix eq 's' ? $num # Seconds : $suffix eq 'm' ? $num * 60 # Minutes : $suffix eq 'h' ? $num * 3600 # Hours : $num * 86400; # Days $opt->{value} = ($prefix || '') . $val; PTDEBUG && _d('Setting option', $opt->{long}, 'to', $val); } else { $self->save_error("Invalid time suffix for --$opt->{long}"); } } elsif ( $val && $opt->{type} eq 'd' ) { # type DSN PTDEBUG && _d('Parsing option', $opt->{long}, 'as a DSN'); my $prev = {}; my $from_key = $self->{defaults_to}->{ $opt->{long} }; if ( $from_key ) { PTDEBUG && _d($opt->{long}, 'DSN copies from', $from_key, 'DSN'); if ( $self->{opts}->{$from_key}->{parsed} ) { $prev = $self->{opts}->{$from_key}->{value}; } else { PTDEBUG && _d('Cannot parse', $opt->{long}, 'until', $from_key, 'parsed'); return; } } my $defaults = $self->{DSNParser}->parse_options($self); $opt->{value} = $self->{DSNParser}->parse($val, $prev, $defaults); } elsif ( $val && $opt->{type} eq 'z' ) { # type size PTDEBUG && _d('Parsing option', $opt->{long}, 'as a size value'); $self->_parse_size($opt, $val); } elsif ( $opt->{type} eq 'H' || (defined $val && $opt->{type} eq 'h') ) { $opt->{value} = { map { $_ => 1 } split(/(?{type} eq 'A' || (defined $val && $opt->{type} eq 'a') ) { $opt->{value} = [ split(/(?{long}, 'type', $opt->{type}, 'value', $val); } $opt->{parsed} = 1; return; } sub get { my ( $self, $opt ) = @_; my $long = (length $opt == 1 ? $self->{short_opts}->{$opt} : $opt); die "Option $opt does not exist" unless $long && exists $self->{opts}->{$long}; return $self->{opts}->{$long}->{value}; } sub got { my ( $self, $opt ) = @_; my $long = (length $opt == 1 ? $self->{short_opts}->{$opt} : $opt); die "Option $opt does not exist" unless $long && exists $self->{opts}->{$long}; return $self->{opts}->{$long}->{got}; } sub has { my ( $self, $opt ) = @_; my $long = (length $opt == 1 ? $self->{short_opts}->{$opt} : $opt); return defined $long ? exists $self->{opts}->{$long} : 0; } sub set { my ( $self, $opt, $val ) = @_; my $long = (length $opt == 1 ? $self->{short_opts}->{$opt} : $opt); die "Option $opt does not exist" unless $long && exists $self->{opts}->{$long}; $self->{opts}->{$long}->{value} = $val; return; } sub save_error { my ( $self, $error ) = @_; push @{$self->{errors}}, $error; return; } sub errors { my ( $self ) = @_; return $self->{errors}; } sub usage { my ( $self ) = @_; warn "No usage string is set" unless $self->{usage}; # XXX return "Usage: " . ($self->{usage} || '') . "\n"; } sub descr { my ( $self ) = @_; warn "No description string is set" unless $self->{description}; # XXX my $descr = ($self->{description} || $self->{program_name} || '') . " For more details, please use the --help option, " . "or try 'perldoc $PROGRAM_NAME' " . "for complete documentation."; $descr = join("\n", $descr =~ m/(.{0,80})(?:\s+|$)/g) unless $ENV{DONT_BREAK_LINES}; $descr =~ s/ +$//mg; return $descr; } sub usage_or_errors { my ( $self, $file, $return ) = @_; $file ||= $self->{file} || __FILE__; if ( !$self->{description} || !$self->{usage} ) { PTDEBUG && _d("Getting description and usage from SYNOPSIS in", $file); my %synop = $self->_parse_synopsis($file); $self->{description} ||= $synop{description}; $self->{usage} ||= $synop{usage}; PTDEBUG && _d("Description:", $self->{description}, "\nUsage:", $self->{usage}); } if ( $self->{opts}->{help}->{got} ) { print $self->print_usage() or die "Cannot print usage: $OS_ERROR"; exit 0 unless $return; } elsif ( scalar @{$self->{errors}} ) { print $self->print_errors() or die "Cannot print errors: $OS_ERROR"; exit 0 unless $return; } return; } sub print_errors { my ( $self ) = @_; my $usage = $self->usage() . "\n"; if ( (my @errors = @{$self->{errors}}) ) { $usage .= join("\n * ", 'Errors in command-line arguments:', @errors) . "\n"; } return $usage . "\n" . $self->descr(); } sub print_usage { my ( $self ) = @_; die "Run get_opts() before print_usage()" unless $self->{got_opts}; my @opts = values %{$self->{opts}}; my $maxl = max( map { length($_->{long}) # option long name + ($_->{is_negatable} ? 4 : 0) # "[no]" if opt is negatable + ($_->{type} ? 2 : 0) # "=x" where x is the opt type } @opts); my $maxs = max(0, map { length($_) + ($self->{opts}->{$_}->{is_negatable} ? 4 : 0) + ($self->{opts}->{$_}->{type} ? 2 : 0) } values %{$self->{short_opts}}); my $lcol = max($maxl, ($maxs + 3)); my $rcol = 80 - $lcol - 6; my $rpad = ' ' x ( 80 - $rcol ); $maxs = max($lcol - 3, $maxs); my $usage = $self->descr() . "\n" . $self->usage(); my @groups = reverse sort grep { $_ ne 'default'; } keys %{$self->{groups}}; push @groups, 'default'; foreach my $group ( reverse @groups ) { $usage .= "\n".($group eq 'default' ? 'Options' : $group).":\n\n"; foreach my $opt ( sort { $a->{long} cmp $b->{long} } grep { $_->{group} eq $group } @opts ) { my $long = $opt->{is_negatable} ? "[no]$opt->{long}" : $opt->{long}; my $short = $opt->{short}; my $desc = $opt->{desc}; $long .= $opt->{type} ? "=$opt->{type}" : ""; if ( $opt->{type} && $opt->{type} eq 'm' ) { my ($s) = $desc =~ m/\(suffix (.)\)/; $s ||= 's'; $desc =~ s/\s+\(suffix .\)//; $desc .= ". Optional suffix s=seconds, m=minutes, h=hours, " . "d=days; if no suffix, $s is used."; } $desc = join("\n$rpad", grep { $_ } $desc =~ m/(.{0,$rcol})(?:\s+|$)/g); $desc =~ s/ +$//mg; if ( $short ) { $usage .= sprintf(" --%-${maxs}s -%s %s\n", $long, $short, $desc); } else { $usage .= sprintf(" --%-${lcol}s %s\n", $long, $desc); } } } $usage .= "\nOption types: s=string, i=integer, f=float, h/H/a/A=comma-separated list, d=DSN, z=size, m=time\n"; if ( (my @rules = @{$self->{rules}}) ) { $usage .= "\nRules:\n\n"; $usage .= join("\n", map { " $_" } @rules) . "\n"; } if ( $self->{DSNParser} ) { $usage .= "\n" . $self->{DSNParser}->usage(); } $usage .= "\nOptions and values after processing arguments:\n\n"; foreach my $opt ( sort { $a->{long} cmp $b->{long} } @opts ) { my $val = $opt->{value}; my $type = $opt->{type} || ''; my $bool = $opt->{spec} =~ m/^[\w-]+(?:\|[\w-])?!?$/; $val = $bool ? ( $val ? 'TRUE' : 'FALSE' ) : !defined $val ? '(No value)' : $type eq 'd' ? $self->{DSNParser}->as_string($val) : $type =~ m/H|h/ ? join(',', sort keys %$val) : $type =~ m/A|a/ ? join(',', @$val) : $val; $usage .= sprintf(" --%-${lcol}s %s\n", $opt->{long}, $val); } return $usage; } sub prompt_noecho { shift @_ if ref $_[0] eq __PACKAGE__; my ( $prompt ) = @_; local $OUTPUT_AUTOFLUSH = 1; print $prompt or die "Cannot print: $OS_ERROR"; my $response; eval { require Term::ReadKey; Term::ReadKey::ReadMode('noecho'); chomp($response = ); Term::ReadKey::ReadMode('normal'); print "\n" or die "Cannot print: $OS_ERROR"; }; if ( $EVAL_ERROR ) { die "Cannot read response; is Term::ReadKey installed? $EVAL_ERROR"; } return $response; } sub _read_config_file { my ( $self, $filename ) = @_; open my $fh, "<", $filename or die "Cannot open $filename: $OS_ERROR\n"; my @args; my $prefix = '--'; my $parse = 1; LINE: while ( my $line = <$fh> ) { chomp $line; next LINE if $line =~ m/^\s*(?:\#|\;|$)/; $line =~ s/\s+#.*$//g; $line =~ s/^\s+|\s+$//g; if ( $line eq '--' ) { $prefix = ''; $parse = 0; next LINE; } if ( $parse && (my($opt, $arg) = $line =~ m/^\s*([^=\s]+?)(?:\s*=\s*(.*?)\s*)?$/) ) { push @args, grep { defined $_ } ("$prefix$opt", $arg); } elsif ( $line =~ m/./ ) { push @args, $line; } else { die "Syntax error in file $filename at line $INPUT_LINE_NUMBER"; } } close $fh; return @args; } sub read_para_after { my ( $self, $file, $regex ) = @_; open my $fh, "<", $file or die "Can't open $file: $OS_ERROR"; local $INPUT_RECORD_SEPARATOR = ''; my $para; while ( $para = <$fh> ) { next unless $para =~ m/^=pod$/m; last; } while ( $para = <$fh> ) { next unless $para =~ m/$regex/; last; } $para = <$fh>; chomp($para); close $fh or die "Can't close $file: $OS_ERROR"; return $para; } sub clone { my ( $self ) = @_; my %clone = map { my $hashref = $self->{$_}; my $val_copy = {}; foreach my $key ( keys %$hashref ) { my $ref = ref $hashref->{$key}; $val_copy->{$key} = !$ref ? $hashref->{$key} : $ref eq 'HASH' ? { %{$hashref->{$key}} } : $ref eq 'ARRAY' ? [ @{$hashref->{$key}} ] : $hashref->{$key}; } $_ => $val_copy; } qw(opts short_opts defaults); foreach my $scalar ( qw(got_opts) ) { $clone{$scalar} = $self->{$scalar}; } return bless \%clone; } sub _parse_size { my ( $self, $opt, $val ) = @_; if ( lc($val || '') eq 'null' ) { PTDEBUG && _d('NULL size for', $opt->{long}); $opt->{value} = 'null'; return; } my %factor_for = (k => 1_024, M => 1_048_576, G => 1_073_741_824); my ($pre, $num, $factor) = $val =~ m/^([+-])?(\d+)([kMG])?$/; if ( defined $num ) { if ( $factor ) { $num *= $factor_for{$factor}; PTDEBUG && _d('Setting option', $opt->{y}, 'to num', $num, '* factor', $factor); } $opt->{value} = ($pre || '') . $num; } else { $self->save_error("Invalid size for --$opt->{long}: $val"); } return; } sub _parse_attribs { my ( $self, $option, $attribs ) = @_; my $types = $self->{types}; return $option . ($attribs->{'short form'} ? '|' . $attribs->{'short form'} : '' ) . ($attribs->{'negatable'} ? '!' : '' ) . ($attribs->{'cumulative'} ? '+' : '' ) . ($attribs->{'type'} ? '=' . $types->{$attribs->{type}} : '' ); } sub _parse_synopsis { my ( $self, $file ) = @_; $file ||= $self->{file} || __FILE__; PTDEBUG && _d("Parsing SYNOPSIS in", $file); local $INPUT_RECORD_SEPARATOR = ''; # read paragraphs open my $fh, "<", $file or die "Cannot open $file: $OS_ERROR"; my $para; 1 while defined($para = <$fh>) && $para !~ m/^=head1 SYNOPSIS/; die "$file does not contain a SYNOPSIS section" unless $para; my @synop; for ( 1..2 ) { # 1 for the usage, 2 for the description my $para = <$fh>; push @synop, $para; } close $fh; PTDEBUG && _d("Raw SYNOPSIS text:", @synop); my ($usage, $desc) = @synop; die "The SYNOPSIS section in $file is not formatted properly" unless $usage && $desc; $usage =~ s/^\s*Usage:\s+(.+)/$1/; chomp $usage; $desc =~ s/\n/ /g; $desc =~ s/\s{2,}/ /g; $desc =~ s/\. ([A-Z][a-z])/. $1/g; $desc =~ s/\s+$//; return ( description => $desc, usage => $usage, ); }; sub _d { my ($package, undef, $line) = caller 0; @_ = map { (my $temp = $_) =~ s/\n/\n# /g; $temp; } map { defined $_ ? $_ : 'undef' } @_; print STDERR "# $package:$line $PID ", join(' ', @_), "\n"; } if ( PTDEBUG ) { print '# ', $^X, ' ', $], "\n"; if ( my $uname = `uname -a` ) { $uname =~ s/\s+/ /g; print "# $uname\n"; } print '# Arguments: ', join(' ', map { my $a = "_[$_]_"; $a =~ s/\n/\n# /g; $a; } @ARGV), "\n"; } 1; } # ########################################################################### # End OptionParser package # ########################################################################### # ########################################################################### # VersionParser package # This package is a copy without comments from the original. The original # with comments and its test file can be found in the Bazaar repository at, # lib/VersionParser.pm # t/lib/VersionParser.t # See https://launchpad.net/percona-toolkit for more information. # ########################################################################### { package VersionParser; use strict; use warnings FATAL => 'all'; use English qw(-no_match_vars); use constant PTDEBUG => $ENV{PTDEBUG} || 0; sub new { my ( $class ) = @_; bless {}, $class; } sub parse { my ( $self, $str ) = @_; my $result = sprintf('%03d%03d%03d', $str =~ m/(\d+)/g); PTDEBUG && _d($str, 'parses to', $result); return $result; } sub version_ge { my ( $self, $dbh, $target ) = @_; if ( !$self->{$dbh} ) { $self->{$dbh} = $self->parse( $dbh->selectrow_array('SELECT VERSION()')); } my $result = $self->{$dbh} ge $self->parse($target) ? 1 : 0; PTDEBUG && _d($self->{$dbh}, 'ge', $target, ':', $result); return $result; } sub innodb_version { my ( $self, $dbh ) = @_; return unless $dbh; my $innodb_version = "NO"; my ($innodb) = grep { $_->{engine} =~ m/InnoDB/i } map { my %hash; @hash{ map { lc $_ } keys %$_ } = values %$_; \%hash; } @{ $dbh->selectall_arrayref("SHOW ENGINES", {Slice=>{}}) }; if ( $innodb ) { PTDEBUG && _d("InnoDB support:", $innodb->{support}); if ( $innodb->{support} =~ m/YES|DEFAULT/i ) { my $vars = $dbh->selectrow_hashref( "SHOW VARIABLES LIKE 'innodb_version'"); $innodb_version = !$vars ? "BUILTIN" : ($vars->{Value} || $vars->{value}); } else { $innodb_version = $innodb->{support}; # probably DISABLED or NO } } PTDEBUG && _d("InnoDB version:", $innodb_version); return $innodb_version; } sub _d { my ($package, undef, $line) = caller 0; @_ = map { (my $temp = $_) =~ s/\n/\n# /g; $temp; } map { defined $_ ? $_ : 'undef' } @_; print STDERR "# $package:$line $PID ", join(' ', @_), "\n"; } 1; } # ########################################################################### # End VersionParser package # ########################################################################### # ########################################################################### # DSNParser package # This package is a copy without comments from the original. The original # with comments and its test file can be found in the Bazaar repository at, # lib/DSNParser.pm # t/lib/DSNParser.t # See https://launchpad.net/percona-toolkit for more information. # ########################################################################### { package DSNParser; use strict; use warnings FATAL => 'all'; use English qw(-no_match_vars); use constant PTDEBUG => $ENV{PTDEBUG} || 0; use Data::Dumper; $Data::Dumper::Indent = 0; $Data::Dumper::Quotekeys = 0; eval { require DBI; }; my $have_dbi = $EVAL_ERROR ? 0 : 1; sub new { my ( $class, %args ) = @_; foreach my $arg ( qw(opts) ) { die "I need a $arg argument" unless $args{$arg}; } my $self = { opts => {} # h, P, u, etc. Should come from DSN OPTIONS section in POD. }; foreach my $opt ( @{$args{opts}} ) { if ( !$opt->{key} || !$opt->{desc} ) { die "Invalid DSN option: ", Dumper($opt); } PTDEBUG && _d('DSN option:', join(', ', map { "$_=" . (defined $opt->{$_} ? ($opt->{$_} || '') : 'undef') } keys %$opt ) ); $self->{opts}->{$opt->{key}} = { dsn => $opt->{dsn}, desc => $opt->{desc}, copy => $opt->{copy} || 0, }; } return bless $self, $class; } sub prop { my ( $self, $prop, $value ) = @_; if ( @_ > 2 ) { PTDEBUG && _d('Setting', $prop, 'property'); $self->{$prop} = $value; } return $self->{$prop}; } sub parse { my ( $self, $dsn, $prev, $defaults ) = @_; if ( !$dsn ) { PTDEBUG && _d('No DSN to parse'); return; } PTDEBUG && _d('Parsing', $dsn); $prev ||= {}; $defaults ||= {}; my %given_props; my %final_props; my $opts = $self->{opts}; foreach my $dsn_part ( split(/,/, $dsn) ) { if ( my ($prop_key, $prop_val) = $dsn_part =~ m/^(.)=(.*)$/ ) { $given_props{$prop_key} = $prop_val; } else { PTDEBUG && _d('Interpreting', $dsn_part, 'as h=', $dsn_part); $given_props{h} = $dsn_part; } } foreach my $key ( keys %$opts ) { PTDEBUG && _d('Finding value for', $key); $final_props{$key} = $given_props{$key}; if ( !defined $final_props{$key} && defined $prev->{$key} && $opts->{$key}->{copy} ) { $final_props{$key} = $prev->{$key}; PTDEBUG && _d('Copying value for', $key, 'from previous DSN'); } if ( !defined $final_props{$key} ) { $final_props{$key} = $defaults->{$key}; PTDEBUG && _d('Copying value for', $key, 'from defaults'); } } foreach my $key ( keys %given_props ) { die "Unknown DSN option '$key' in '$dsn'. For more details, " . "please use the --help option, or try 'perldoc $PROGRAM_NAME' " . "for complete documentation." unless exists $opts->{$key}; } if ( (my $required = $self->prop('required')) ) { foreach my $key ( keys %$required ) { die "Missing required DSN option '$key' in '$dsn'. For more details, " . "please use the --help option, or try 'perldoc $PROGRAM_NAME' " . "for complete documentation." unless $final_props{$key}; } } return \%final_props; } sub parse_options { my ( $self, $o ) = @_; die 'I need an OptionParser object' unless ref $o eq 'OptionParser'; my $dsn_string = join(',', map { "$_=".$o->get($_); } grep { $o->has($_) && $o->get($_) } keys %{$self->{opts}} ); PTDEBUG && _d('DSN string made from options:', $dsn_string); return $self->parse($dsn_string); } sub as_string { my ( $self, $dsn, $props ) = @_; return $dsn unless ref $dsn; my @keys = $props ? @$props : sort keys %$dsn; return join(',', map { "$_=" . ($_ eq 'p' ? '...' : $dsn->{$_}) } grep { exists $self->{opts}->{$_} && exists $dsn->{$_} && defined $dsn->{$_} } @keys); } sub usage { my ( $self ) = @_; my $usage = "DSN syntax is key=value[,key=value...] Allowable DSN keys:\n\n" . " KEY COPY MEANING\n" . " === ==== =============================================\n"; my %opts = %{$self->{opts}}; foreach my $key ( sort keys %opts ) { $usage .= " $key " . ($opts{$key}->{copy} ? 'yes ' : 'no ') . ($opts{$key}->{desc} || '[No description]') . "\n"; } $usage .= "\n If the DSN is a bareword, the word is treated as the 'h' key.\n"; return $usage; } sub get_cxn_params { my ( $self, $info ) = @_; my $dsn; my %opts = %{$self->{opts}}; my $driver = $self->prop('dbidriver') || ''; if ( $driver eq 'Pg' ) { $dsn = 'DBI:Pg:dbname=' . ( $info->{D} || '' ) . ';' . join(';', map { "$opts{$_}->{dsn}=$info->{$_}" } grep { defined $info->{$_} } qw(h P)); } else { $dsn = 'DBI:mysql:' . ( $info->{D} || '' ) . ';' . join(';', map { "$opts{$_}->{dsn}=$info->{$_}" } grep { defined $info->{$_} } qw(F h P S A)) . ';mysql_read_default_group=client'; } PTDEBUG && _d($dsn); return ($dsn, $info->{u}, $info->{p}); } sub fill_in_dsn { my ( $self, $dbh, $dsn ) = @_; my $vars = $dbh->selectall_hashref('SHOW VARIABLES', 'Variable_name'); my ($user, $db) = $dbh->selectrow_array('SELECT USER(), DATABASE()'); $user =~ s/@.*//; $dsn->{h} ||= $vars->{hostname}->{Value}; $dsn->{S} ||= $vars->{'socket'}->{Value}; $dsn->{P} ||= $vars->{port}->{Value}; $dsn->{u} ||= $user; $dsn->{D} ||= $db; } sub get_dbh { my ( $self, $cxn_string, $user, $pass, $opts ) = @_; $opts ||= {}; my $defaults = { AutoCommit => 0, RaiseError => 1, PrintError => 0, ShowErrorStatement => 1, mysql_enable_utf8 => ($cxn_string =~ m/charset=utf8/i ? 1 : 0), }; @{$defaults}{ keys %$opts } = values %$opts; if ( $opts->{mysql_use_result} ) { $defaults->{mysql_use_result} = 1; } if ( !$have_dbi ) { die "Cannot connect to MySQL because the Perl DBI module is not " . "installed or not found. Run 'perl -MDBI' to see the directories " . "that Perl searches for DBI. If DBI is not installed, try:\n" . " Debian/Ubuntu apt-get install libdbi-perl\n" . " RHEL/CentOS yum install perl-DBI\n" . " OpenSolaris pgk install pkg:/SUNWpmdbi\n"; } my $dbh; my $tries = 2; while ( !$dbh && $tries-- ) { PTDEBUG && _d($cxn_string, ' ', $user, ' ', $pass, join(', ', map { "$_=>$defaults->{$_}" } keys %$defaults )); eval { $dbh = DBI->connect($cxn_string, $user, $pass, $defaults); if ( $cxn_string =~ m/mysql/i ) { my $sql; $sql = 'SELECT @@SQL_MODE'; PTDEBUG && _d($dbh, $sql); my ($sql_mode) = $dbh->selectrow_array($sql); $sql = 'SET @@SQL_QUOTE_SHOW_CREATE = 1' . '/*!40101, @@SQL_MODE=\'NO_AUTO_VALUE_ON_ZERO' . ($sql_mode ? ",$sql_mode" : '') . '\'*/'; PTDEBUG && _d($dbh, $sql); $dbh->do($sql); if ( my ($charset) = $cxn_string =~ m/charset=(\w+)/ ) { $sql = "/*!40101 SET NAMES $charset*/"; PTDEBUG && _d($dbh, ':', $sql); $dbh->do($sql); PTDEBUG && _d('Enabling charset for STDOUT'); if ( $charset eq 'utf8' ) { binmode(STDOUT, ':utf8') or die "Can't binmode(STDOUT, ':utf8'): $OS_ERROR"; } else { binmode(STDOUT) or die "Can't binmode(STDOUT): $OS_ERROR"; } } if ( $self->prop('set-vars') ) { $sql = "SET " . $self->prop('set-vars'); PTDEBUG && _d($dbh, ':', $sql); $dbh->do($sql); } } }; if ( !$dbh && $EVAL_ERROR ) { PTDEBUG && _d($EVAL_ERROR); if ( $EVAL_ERROR =~ m/not a compiled character set|character set utf8/ ) { PTDEBUG && _d('Going to try again without utf8 support'); delete $defaults->{mysql_enable_utf8}; } elsif ( $EVAL_ERROR =~ m/locate DBD\/mysql/i ) { die "Cannot connect to MySQL because the Perl DBD::mysql module is " . "not installed or not found. Run 'perl -MDBD::mysql' to see " . "the directories that Perl searches for DBD::mysql. If " . "DBD::mysql is not installed, try:\n" . " Debian/Ubuntu apt-get install libdbd-mysql-perl\n" . " RHEL/CentOS yum install perl-DBD-MySQL\n" . " OpenSolaris pgk install pkg:/SUNWapu13dbd-mysql\n"; } if ( !$tries ) { die $EVAL_ERROR; } } } PTDEBUG && _d('DBH info: ', $dbh, Dumper($dbh->selectrow_hashref( 'SELECT DATABASE(), CONNECTION_ID(), VERSION()/*!50038 , @@hostname*/')), 'Connection info:', $dbh->{mysql_hostinfo}, 'Character set info:', Dumper($dbh->selectall_arrayref( 'SHOW VARIABLES LIKE "character_set%"', { Slice => {}})), '$DBD::mysql::VERSION:', $DBD::mysql::VERSION, '$DBI::VERSION:', $DBI::VERSION, ); return $dbh; } sub get_hostname { my ( $self, $dbh ) = @_; if ( my ($host) = ($dbh->{mysql_hostinfo} || '') =~ m/^(\w+) via/ ) { return $host; } my ( $hostname, $one ) = $dbh->selectrow_array( 'SELECT /*!50038 @@hostname, */ 1'); return $hostname; } sub disconnect { my ( $self, $dbh ) = @_; PTDEBUG && $self->print_active_handles($dbh); $dbh->disconnect; } sub print_active_handles { my ( $self, $thing, $level ) = @_; $level ||= 0; printf("# Active %sh: %s %s %s\n", ($thing->{Type} || 'undef'), "\t" x $level, $thing, (($thing->{Type} || '') eq 'st' ? $thing->{Statement} || '' : '')) or die "Cannot print: $OS_ERROR"; foreach my $handle ( grep {defined} @{ $thing->{ChildHandles} } ) { $self->print_active_handles( $handle, $level + 1 ); } } sub copy { my ( $self, $dsn_1, $dsn_2, %args ) = @_; die 'I need a dsn_1 argument' unless $dsn_1; die 'I need a dsn_2 argument' unless $dsn_2; my %new_dsn = map { my $key = $_; my $val; if ( $args{overwrite} ) { $val = defined $dsn_1->{$key} ? $dsn_1->{$key} : $dsn_2->{$key}; } else { $val = defined $dsn_2->{$key} ? $dsn_2->{$key} : $dsn_1->{$key}; } $key => $val; } keys %{$self->{opts}}; return \%new_dsn; } sub _d { my ($package, undef, $line) = caller 0; @_ = map { (my $temp = $_) =~ s/\n/\n# /g; $temp; } map { defined $_ ? $_ : 'undef' } @_; print STDERR "# $package:$line $PID ", join(' ', @_), "\n"; } 1; } # ########################################################################### # End DSNParser package # ########################################################################### # ########################################################################### # Daemon package # This package is a copy without comments from the original. The original # with comments and its test file can be found in the Bazaar repository at, # lib/Daemon.pm # t/lib/Daemon.t # See https://launchpad.net/percona-toolkit for more information. # ########################################################################### { package Daemon; use strict; use warnings FATAL => 'all'; use English qw(-no_match_vars); use constant PTDEBUG => $ENV{PTDEBUG} || 0; use POSIX qw(setsid); sub new { my ( $class, %args ) = @_; foreach my $arg ( qw(o) ) { die "I need a $arg argument" unless $args{$arg}; } my $o = $args{o}; my $self = { o => $o, log_file => $o->has('log') ? $o->get('log') : undef, PID_file => $o->has('pid') ? $o->get('pid') : undef, }; check_PID_file(undef, $self->{PID_file}); PTDEBUG && _d('Daemonized child will log to', $self->{log_file}); return bless $self, $class; } sub daemonize { my ( $self ) = @_; PTDEBUG && _d('About to fork and daemonize'); defined (my $pid = fork()) or die "Cannot fork: $OS_ERROR"; if ( $pid ) { PTDEBUG && _d('Parent PID', $PID, 'exiting after forking child PID',$pid); exit; } PTDEBUG && _d('Daemonizing child PID', $PID); $self->{PID_owner} = $PID; $self->{child} = 1; POSIX::setsid() or die "Cannot start a new session: $OS_ERROR"; chdir '/' or die "Cannot chdir to /: $OS_ERROR"; $self->_make_PID_file(); $OUTPUT_AUTOFLUSH = 1; PTDEBUG && _d('Redirecting STDIN to /dev/null'); close STDIN; open STDIN, '/dev/null' or die "Cannot reopen STDIN to /dev/null: $OS_ERROR"; if ( $self->{log_file} ) { PTDEBUG && _d('Redirecting STDOUT and STDERR to', $self->{log_file}); close STDOUT; open STDOUT, '>>', $self->{log_file} or die "Cannot open log file $self->{log_file}: $OS_ERROR"; close STDERR; open STDERR, ">&STDOUT" or die "Cannot dupe STDERR to STDOUT: $OS_ERROR"; } else { if ( -t STDOUT ) { PTDEBUG && _d('No log file and STDOUT is a terminal;', 'redirecting to /dev/null'); close STDOUT; open STDOUT, '>', '/dev/null' or die "Cannot reopen STDOUT to /dev/null: $OS_ERROR"; } if ( -t STDERR ) { PTDEBUG && _d('No log file and STDERR is a terminal;', 'redirecting to /dev/null'); close STDERR; open STDERR, '>', '/dev/null' or die "Cannot reopen STDERR to /dev/null: $OS_ERROR"; } } return; } sub check_PID_file { my ( $self, $file ) = @_; my $PID_file = $self ? $self->{PID_file} : $file; PTDEBUG && _d('Checking PID file', $PID_file); if ( $PID_file && -f $PID_file ) { my $pid; eval { chomp($pid = `cat $PID_file`); }; die "Cannot cat $PID_file: $OS_ERROR" if $EVAL_ERROR; PTDEBUG && _d('PID file exists; it contains PID', $pid); if ( $pid ) { my $pid_is_alive = kill 0, $pid; if ( $pid_is_alive ) { die "The PID file $PID_file already exists " . " and the PID that it contains, $pid, is running"; } else { warn "Overwriting PID file $PID_file because the PID that it " . "contains, $pid, is not running"; } } else { die "The PID file $PID_file already exists but it does not " . "contain a PID"; } } else { PTDEBUG && _d('No PID file'); } return; } sub make_PID_file { my ( $self ) = @_; if ( exists $self->{child} ) { die "Do not call Daemon::make_PID_file() for daemonized scripts"; } $self->_make_PID_file(); $self->{PID_owner} = $PID; return; } sub _make_PID_file { my ( $self ) = @_; my $PID_file = $self->{PID_file}; if ( !$PID_file ) { PTDEBUG && _d('No PID file to create'); return; } $self->check_PID_file(); open my $PID_FH, '>', $PID_file or die "Cannot open PID file $PID_file: $OS_ERROR"; print $PID_FH $PID or die "Cannot print to PID file $PID_file: $OS_ERROR"; close $PID_FH or die "Cannot close PID file $PID_file: $OS_ERROR"; PTDEBUG && _d('Created PID file:', $self->{PID_file}); return; } sub _remove_PID_file { my ( $self ) = @_; if ( $self->{PID_file} && -f $self->{PID_file} ) { unlink $self->{PID_file} or warn "Cannot remove PID file $self->{PID_file}: $OS_ERROR"; PTDEBUG && _d('Removed PID file'); } else { PTDEBUG && _d('No PID to remove'); } return; } sub DESTROY { my ( $self ) = @_; $self->_remove_PID_file() if ($self->{PID_owner} || 0) == $PID; return; } sub _d { my ($package, undef, $line) = caller 0; @_ = map { (my $temp = $_) =~ s/\n/\n# /g; $temp; } map { defined $_ ? $_ : 'undef' } @_; print STDERR "# $package:$line $PID ", join(' ', @_), "\n"; } 1; } # ########################################################################### # End Daemon package # ########################################################################### # ########################################################################### # Transformers package # This package is a copy without comments from the original. The original # with comments and its test file can be found in the Bazaar repository at, # lib/Transformers.pm # t/lib/Transformers.t # See https://launchpad.net/percona-toolkit for more information. # ########################################################################### { package Transformers; use strict; use warnings FATAL => 'all'; use English qw(-no_match_vars); use constant PTDEBUG => $ENV{PTDEBUG} || 0; use Time::Local qw(timegm timelocal); use Digest::MD5 qw(md5_hex); require Exporter; our @ISA = qw(Exporter); our %EXPORT_TAGS = (); our @EXPORT = (); our @EXPORT_OK = qw( micro_t percentage_of secs_to_time time_to_secs shorten ts parse_timestamp unix_timestamp any_unix_timestamp make_checksum crc32 ); our $mysql_ts = qr/(\d\d)(\d\d)(\d\d) +(\d+):(\d+):(\d+)(\.\d+)?/; our $proper_ts = qr/(\d\d\d\d)-(\d\d)-(\d\d)[T ](\d\d):(\d\d):(\d\d)(\.\d+)?/; our $n_ts = qr/(\d{1,5})([shmd]?)/; # Limit \d{1,5} because \d{6} looks sub micro_t { my ( $t, %args ) = @_; my $p_ms = defined $args{p_ms} ? $args{p_ms} : 0; # precision for ms vals my $p_s = defined $args{p_s} ? $args{p_s} : 0; # precision for s vals my $f; $t = 0 if $t < 0; $t = sprintf('%.17f', $t) if $t =~ /e/; $t =~ s/\.(\d{1,6})\d*/\.$1/; if ($t > 0 && $t <= 0.000999) { $f = ($t * 1000000) . 'us'; } elsif ($t >= 0.001000 && $t <= 0.999999) { $f = sprintf("%.${p_ms}f", $t * 1000); $f = ($f * 1) . 'ms'; # * 1 to remove insignificant zeros } elsif ($t >= 1) { $f = sprintf("%.${p_s}f", $t); $f = ($f * 1) . 's'; # * 1 to remove insignificant zeros } else { $f = 0; # $t should = 0 at this point } return $f; } sub percentage_of { my ( $is, $of, %args ) = @_; my $p = $args{p} || 0; # float precision my $fmt = $p ? "%.${p}f" : "%d"; return sprintf $fmt, ($is * 100) / ($of ||= 1); } sub secs_to_time { my ( $secs, $fmt ) = @_; $secs ||= 0; return '00:00' unless $secs; $fmt ||= $secs >= 86_400 ? 'd' : $secs >= 3_600 ? 'h' : 'm'; return $fmt eq 'd' ? sprintf( "%d+%02d:%02d:%02d", int($secs / 86_400), int(($secs % 86_400) / 3_600), int(($secs % 3_600) / 60), $secs % 60) : $fmt eq 'h' ? sprintf( "%02d:%02d:%02d", int(($secs % 86_400) / 3_600), int(($secs % 3_600) / 60), $secs % 60) : sprintf( "%02d:%02d", int(($secs % 3_600) / 60), $secs % 60); } sub time_to_secs { my ( $val, $default_suffix ) = @_; die "I need a val argument" unless defined $val; my $t = 0; my ( $prefix, $num, $suffix ) = $val =~ m/([+-]?)(\d+)([a-z])?$/; $suffix = $suffix || $default_suffix || 's'; if ( $suffix =~ m/[smhd]/ ) { $t = $suffix eq 's' ? $num * 1 # Seconds : $suffix eq 'm' ? $num * 60 # Minutes : $suffix eq 'h' ? $num * 3600 # Hours : $num * 86400; # Days $t *= -1 if $prefix && $prefix eq '-'; } else { die "Invalid suffix for $val: $suffix"; } return $t; } sub shorten { my ( $num, %args ) = @_; my $p = defined $args{p} ? $args{p} : 2; # float precision my $d = defined $args{d} ? $args{d} : 1_024; # divisor my $n = 0; my @units = ('', qw(k M G T P E Z Y)); while ( $num >= $d && $n < @units - 1 ) { $num /= $d; ++$n; } return sprintf( $num =~ m/\./ || $n ? "%.${p}f%s" : '%d', $num, $units[$n]); } sub ts { my ( $time, $gmt ) = @_; my ( $sec, $min, $hour, $mday, $mon, $year ) = $gmt ? gmtime($time) : localtime($time); $mon += 1; $year += 1900; my $val = sprintf("%d-%02d-%02dT%02d:%02d:%02d", $year, $mon, $mday, $hour, $min, $sec); if ( my ($us) = $time =~ m/(\.\d+)$/ ) { $us = sprintf("%.6f", $us); $us =~ s/^0\././; $val .= $us; } return $val; } sub parse_timestamp { my ( $val ) = @_; if ( my($y, $m, $d, $h, $i, $s, $f) = $val =~ m/^$mysql_ts$/ ) { return sprintf "%d-%02d-%02d %02d:%02d:" . (defined $f ? '%09.6f' : '%02d'), $y + 2000, $m, $d, $h, $i, (defined $f ? $s + $f : $s); } return $val; } sub unix_timestamp { my ( $val, $gmt ) = @_; if ( my($y, $m, $d, $h, $i, $s, $us) = $val =~ m/^$proper_ts$/ ) { $val = $gmt ? timegm($s, $i, $h, $d, $m - 1, $y) : timelocal($s, $i, $h, $d, $m - 1, $y); if ( defined $us ) { $us = sprintf('%.6f', $us); $us =~ s/^0\././; $val .= $us; } } return $val; } sub any_unix_timestamp { my ( $val, $callback ) = @_; if ( my ($n, $suffix) = $val =~ m/^$n_ts$/ ) { $n = $suffix eq 's' ? $n # Seconds : $suffix eq 'm' ? $n * 60 # Minutes : $suffix eq 'h' ? $n * 3600 # Hours : $suffix eq 'd' ? $n * 86400 # Days : $n; # default: Seconds PTDEBUG && _d('ts is now - N[shmd]:', $n); return time - $n; } elsif ( $val =~ m/^\d{9,}/ ) { PTDEBUG && _d('ts is already a unix timestamp'); return $val; } elsif ( my ($ymd, $hms) = $val =~ m/^(\d{6})(?:\s+(\d+:\d+:\d+))?/ ) { PTDEBUG && _d('ts is MySQL slow log timestamp'); $val .= ' 00:00:00' unless $hms; return unix_timestamp(parse_timestamp($val)); } elsif ( ($ymd, $hms) = $val =~ m/^(\d{4}-\d\d-\d\d)(?:[T ](\d+:\d+:\d+))?/) { PTDEBUG && _d('ts is properly formatted timestamp'); $val .= ' 00:00:00' unless $hms; return unix_timestamp($val); } else { PTDEBUG && _d('ts is MySQL expression'); return $callback->($val) if $callback && ref $callback eq 'CODE'; } PTDEBUG && _d('Unknown ts type:', $val); return; } sub make_checksum { my ( $val ) = @_; my $checksum = uc substr(md5_hex($val), -16); PTDEBUG && _d($checksum, 'checksum for', $val); return $checksum; } sub crc32 { my ( $string ) = @_; return unless $string; my $poly = 0xEDB88320; my $crc = 0xFFFFFFFF; foreach my $char ( split(//, $string) ) { my $comp = ($crc ^ ord($char)) & 0xFF; for ( 1 .. 8 ) { $comp = $comp & 1 ? $poly ^ ($comp >> 1) : $comp >> 1; } $crc = (($crc >> 8) & 0x00FFFFFF) ^ $comp; } return $crc ^ 0xFFFFFFFF; } sub _d { my ($package, undef, $line) = caller 0; @_ = map { (my $temp = $_) =~ s/\n/\n# /g; $temp; } map { defined $_ ? $_ : 'undef' } @_; print STDERR "# $package:$line $PID ", join(' ', @_), "\n"; } 1; } # ########################################################################### # End Transformers package # ########################################################################### # ########################################################################### # Processlist package # This package is a copy without comments from the original. The original # with comments and its test file can be found in the Bazaar repository at, # lib/Processlist.pm # t/lib/Processlist.t # See https://launchpad.net/percona-toolkit for more information. # ########################################################################### { package Processlist; use strict; use warnings FATAL => 'all'; use English qw(-no_match_vars); use Time::HiRes qw(time usleep); use List::Util qw(max); use Data::Dumper; $Data::Dumper::Indent = 1; $Data::Dumper::Sortkeys = 1; $Data::Dumper::Quotekeys = 0; use constant PTDEBUG => $ENV{PTDEBUG} || 0; use constant { ID => 0, USER => 1, HOST => 2, DB => 3, COMMAND => 4, TIME => 5, STATE => 6, INFO => 7, START => 8, # Calculated start time of statement ($start - TIME) ETIME => 9, # Exec time of SHOW PROCESSLIST (margin of error in START) FSEEN => 10, # First time ever seen PROFILE => 11, # Profile of individual STATE times }; sub new { my ( $class, %args ) = @_; foreach my $arg ( qw(MasterSlave) ) { die "I need a $arg argument" unless $args{$arg}; } my $self = { %args, polls => 0, last_poll => 0, active_cxn => {}, # keyed off ID event_cache => [], }; return bless $self, $class; } sub parse_event { my ( $self, %args ) = @_; my @required_args = qw(code); foreach my $arg ( @required_args ) { die "I need a $arg argument" unless $args{$arg}; } my ($code) = @args{@required_args}; if ( @{$self->{event_cache}} ) { PTDEBUG && _d("Returning cached event"); return shift @{$self->{event_cache}}; } if ( $self->{interval} && $self->{polls} ) { PTDEBUG && _d("Sleeping between polls"); usleep($self->{interval}); } PTDEBUG && _d("Polling PROCESSLIST"); my ($time, $etime) = @args{qw(time etime)}; my $start = $etime ? 0 : time; # don't need start if etime given my $rows = $code->(); if ( !$rows ) { warn "Processlist callback did not return an arrayref"; return; } $time = time unless $time; $etime = $time - $start unless $etime; $self->{polls}++; PTDEBUG && _d('Rows:', ($rows ? scalar @$rows : 0), 'in', $etime, 'seconds'); my $active_cxn = $self->{active_cxn}; my $curr_cxn = {}; my @new_cxn = (); CURRENTLY_ACTIVE_CXN: foreach my $curr ( @$rows ) { $curr_cxn->{$curr->[ID]} = $curr; my $query_start = $time - ($curr->[TIME] || 0); if ( $active_cxn->{$curr->[ID]} ) { PTDEBUG && _d('Checking existing cxn', $curr->[ID]); my $prev = $active_cxn->{$curr->[ID]}; # previous state of cxn my $new_query = 0; my $fudge = ($curr->[TIME] || 0) =~ m/\D/ ? 0.001 : 1; # micro-t? if ( $prev->[INFO] ) { if ( !$curr->[INFO] || $prev->[INFO] ne $curr->[INFO] ) { PTDEBUG && _d('Info is different; new query'); $new_query = 1; } elsif ( defined $curr->[TIME] && $curr->[TIME] < $prev->[TIME] ) { PTDEBUG && _d('Time is less than previous; new query'); $new_query = 1; } elsif ( $curr->[INFO] && defined $curr->[TIME] && $query_start - $etime - $prev->[START] > $fudge ) { PTDEBUG && _d('Query restarted; new query', $query_start, $etime, $prev->[START], $fudge); $new_query = 1; } if ( $new_query ) { $self->_update_profile($prev, $curr, $time); push @{$self->{event_cache}}, $self->make_event($prev, $time); } } if ( $curr->[INFO] ) { if ( $prev->[INFO] && !$new_query ) { PTDEBUG && _d("Query on cxn", $curr->[ID], "hasn't changed"); $self->_update_profile($prev, $curr, $time); } else { PTDEBUG && _d('Saving new query, state', $curr->[STATE]); push @new_cxn, [ @$curr, # proc info int($query_start), # START $etime, # ETIME $time, # FSEEN { $curr->[STATE] => 0 }, # PROFILE ]; } } } else { PTDEBUG && _d('New cxn', $curr->[ID]); if ( $curr->[INFO] && defined $curr->[TIME] ) { PTDEBUG && _d('Saving query of new cxn, state', $curr->[STATE]); push @new_cxn, [ @$curr, # proc info int($query_start), # START $etime, # ETIME $time, # FSEEN { $curr->[STATE] => 0 }, # PROFILE ]; } } } # CURRENTLY_ACTIVE_CXN PREVIOUSLY_ACTIVE_CXN: foreach my $prev ( values %$active_cxn ) { if ( !$curr_cxn->{$prev->[ID]} ) { PTDEBUG && _d('cxn', $prev->[ID], 'ended'); push @{$self->{event_cache}}, $self->make_event($prev, $time); delete $active_cxn->{$prev->[ID]}; } elsif ( ($curr_cxn->{$prev->[ID]}->[COMMAND] || "") eq 'Sleep' || !$curr_cxn->{$prev->[ID]}->[STATE] || !$curr_cxn->{$prev->[ID]}->[INFO] ) { PTDEBUG && _d('cxn', $prev->[ID], 'became idle'); delete $active_cxn->{$prev->[ID]}; } } map { $active_cxn->{$_->[ID]} = $_; } @new_cxn; $self->{last_poll} = $time; my $event = shift @{$self->{event_cache}}; PTDEBUG && _d(scalar @{$self->{event_cache}}, "events in cache"); return $event; } sub make_event { my ( $self, $row, $time ) = @_; my $observed_time = $time - $row->[FSEEN]; my $Query_time = max($row->[TIME], $observed_time); my $event = { id => $row->[ID], db => $row->[DB], user => $row->[USER], host => $row->[HOST], arg => $row->[INFO], bytes => length($row->[INFO]), ts => Transformers::ts($row->[START] + $row->[TIME]), # Query END time Query_time => $Query_time, Lock_time => $row->[PROFILE]->{Locked} || 0, }; PTDEBUG && _d('Properties of event:', Dumper($event)); return $event; } sub _get_active_cxn { my ( $self ) = @_; PTDEBUG && _d("Active cxn:", Dumper($self->{active_cxn})); return $self->{active_cxn}; } sub _update_profile { my ( $self, $prev, $curr, $time ) = @_; return unless $prev && $curr; my $time_elapsed = $time - $self->{last_poll}; if ( ($prev->[STATE] || "") eq ($curr->[STATE] || "") ) { PTDEBUG && _d("Query is still in", $curr->[STATE], "state"); $prev->[PROFILE]->{$prev->[STATE] || ""} += $time_elapsed; } else { PTDEBUG && _d("Query changed from state", $prev->[STATE], "to", $curr->[STATE]); my $half_time = ($time_elapsed || 0) / 2; $prev->[PROFILE]->{$prev->[STATE] || ""} += $half_time; $prev->[STATE] = $curr->[STATE]; $prev->[PROFILE]->{$curr->[STATE] || ""} = $half_time; } return; } sub find { my ( $self, $proclist, %find_spec ) = @_; PTDEBUG && _d('find specs:', Dumper(\%find_spec)); my $ms = $self->{MasterSlave}; my @matches; QUERY: foreach my $query ( @$proclist ) { PTDEBUG && _d('Checking query', Dumper($query)); my $matched = 0; if ( !$find_spec{replication_threads} && $ms->is_replication_thread($query) ) { PTDEBUG && _d('Skipping replication thread'); next QUERY; } if ( $find_spec{busy_time} && ($query->{Command} || '') eq 'Query' ) { if ( $query->{Time} < $find_spec{busy_time} ) { PTDEBUG && _d("Query isn't running long enough"); next QUERY; } PTDEBUG && _d('Exceeds busy time'); $matched++; } if ( $find_spec{idle_time} && ($query->{Command} || '') eq 'Sleep' ) { if ( $query->{Time} < $find_spec{idle_time} ) { PTDEBUG && _d("Query isn't idle long enough"); next QUERY; } PTDEBUG && _d('Exceeds idle time'); $matched++; } PROPERTY: foreach my $property ( qw(Id User Host db State Command Info) ) { my $filter = "_find_match_$property"; if ( defined $find_spec{ignore}->{$property} && $self->$filter($query, $find_spec{ignore}->{$property}) ) { PTDEBUG && _d('Query matches ignore', $property, 'spec'); next QUERY; } if ( defined $find_spec{match}->{$property} ) { if ( !$self->$filter($query, $find_spec{match}->{$property}) ) { PTDEBUG && _d('Query does not match', $property, 'spec'); next QUERY; } PTDEBUG && _d('Query matches', $property, 'spec'); $matched++; } } if ( $matched || $find_spec{all} ) { PTDEBUG && _d("Query matched one or more specs, adding"); push @matches, $query; next QUERY; } PTDEBUG && _d('Query does not match any specs, ignoring'); } # QUERY return @matches; } sub _find_match_Id { my ( $self, $query, $property ) = @_; return defined $property && defined $query->{Id} && $query->{Id} == $property; } sub _find_match_User { my ( $self, $query, $property ) = @_; return defined $property && defined $query->{User} && $query->{User} =~ m/$property/; } sub _find_match_Host { my ( $self, $query, $property ) = @_; return defined $property && defined $query->{Host} && $query->{Host} =~ m/$property/; } sub _find_match_db { my ( $self, $query, $property ) = @_; return defined $property && defined $query->{db} && $query->{db} =~ m/$property/; } sub _find_match_State { my ( $self, $query, $property ) = @_; return defined $property && defined $query->{State} && $query->{State} =~ m/$property/; } sub _find_match_Command { my ( $self, $query, $property ) = @_; return defined $property && defined $query->{Command} && $query->{Command} =~ m/$property/; } sub _find_match_Info { my ( $self, $query, $property ) = @_; return defined $property && defined $query->{Info} && $query->{Info} =~ m/$property/; } sub _d { my ($package, undef, $line) = caller 0; @_ = map { (my $temp = $_) =~ s/\n/\n# /g; $temp; } map { defined $_ ? $_ : 'undef' } @_; print STDERR "# $package:$line $PID ", join(' ', @_), "\n"; } 1; } # ########################################################################### # End Processlist package # ########################################################################### # ########################################################################### # TextResultSetParser package # This package is a copy without comments from the original. The original # with comments and its test file can be found in the Bazaar repository at, # lib/TextResultSetParser.pm # t/lib/TextResultSetParser.t # See https://launchpad.net/percona-toolkit for more information. # ########################################################################### { package TextResultSetParser; use strict; use warnings FATAL => 'all'; use English qw(-no_match_vars); use constant PTDEBUG => $ENV{PTDEBUG} || 0; sub new { my ( $class, %args ) = @_; my %value_for = ( 'NULL' => undef, # DBI::selectall_arrayref() does this ($args{value_for} ? %{$args{value_for}} : ()), ); my $self = { %args, value_for => \%value_for, }; return bless $self, $class; } sub _parse_tabular { my ( $text, @cols ) = @_; my %row; my @vals = $text =~ m/\| +([^\|]*?)(?= +\|)/msg; return (undef, \@vals) unless @cols; @row{@cols} = @vals; return (\%row, undef); } sub _parse_tab_sep { my ( $text, @cols ) = @_; my %row; my @vals = split(/\t/, $text); return (undef, \@vals) unless @cols; @row{@cols} = @vals; return (\%row, undef); } sub parse_vertical_row { my ( $self, $text ) = @_; my %row = $text =~ m/^\s*(\w+):(?: ([^\n]*))?/msg; if ( $self->{NAME_lc} ) { my %lc_row = map { my $key = lc $_; $key => $row{$_}; } keys %row; return \%lc_row; } else { return \%row; } } sub parse { my ( $self, $text ) = @_; my $result_set; if ( $text =~ m/^\+---/m ) { # standard "tabular" output PTDEBUG && _d('Result set text is standard tabular'); my $line_pattern = qr/^(\| .*)[\r\n]+/m; $result_set = $self->parse_horizontal_row($text, $line_pattern, \&_parse_tabular); } elsif ( $text =~ m/^\w+\t\w+/m ) { # tab-separated PTDEBUG && _d('Result set text is tab-separated'); my $line_pattern = qr/^(.*?\t.*)[\r\n]+/m; $result_set = $self->parse_horizontal_row($text, $line_pattern, \&_parse_tab_sep); } elsif ( $text =~ m/\*\*\* \d+\. row/ ) { # "vertical" output PTDEBUG && _d('Result set text is vertical (\G)'); foreach my $row ( split_vertical_rows($text) ) { push @$result_set, $self->parse_vertical_row($row); } } else { my $text_sample = substr $text, 0, 300; my $remaining = length $text > 300 ? (length $text) - 300 : 0; chomp $text_sample; die "Cannot determine if text is tabular, tab-separated or vertical:\n" . "$text_sample\n" . ($remaining ? "(not showing last $remaining bytes of text)\n" : ""); } if ( $self->{value_for} ) { foreach my $result_set ( @$result_set ) { foreach my $key ( keys %$result_set ) { next unless defined $result_set->{$key}; $result_set->{$key} = $self->{value_for}->{ $result_set->{$key} } if exists $self->{value_for}->{ $result_set->{$key} }; } } } return $result_set; } sub parse_horizontal_row { my ( $self, $text, $line_pattern, $sub ) = @_; my @result_sets = (); my @cols = (); foreach my $line ( $text =~ m/$line_pattern/g ) { my ( $row, $cols ) = $sub->($line, @cols); if ( $row ) { push @result_sets, $row; } else { @cols = map { $self->{NAME_lc} ? lc $_ : $_ } @$cols; } } return \@result_sets; } sub split_vertical_rows { my ( $text ) = @_; my $ROW_HEADER = '\*{3,} \d+\. row \*{3,}'; my @rows = $text =~ m/($ROW_HEADER.*?)(?=$ROW_HEADER|\z)/omgs; return @rows; } sub _d { my ($package, undef, $line) = caller 0; @_ = map { (my $temp = $_) =~ s/\n/\n# /g; $temp; } map { defined $_ ? $_ : 'undef' } @_; print STDERR "# $package:$line $PID ", join(' ', @_), "\n"; } 1; } # ########################################################################### # End TextResultSetParser package # ########################################################################### # ########################################################################### # MasterSlave package # This package is a copy without comments from the original. The original # with comments and its test file can be found in the Bazaar repository at, # lib/MasterSlave.pm # t/lib/MasterSlave.t # See https://launchpad.net/percona-toolkit for more information. # ########################################################################### { package MasterSlave; use strict; use warnings FATAL => 'all'; use English qw(-no_match_vars); use constant PTDEBUG => $ENV{PTDEBUG} || 0; sub new { my ( $class, %args ) = @_; my $self = { %args, replication_thread => {}, }; return bless $self, $class; } sub get_slaves { my ($self, %args) = @_; my @required_args = qw(make_cxn OptionParser DSNParser Quoter); foreach my $arg ( @required_args ) { die "I need a $arg argument" unless $args{$arg}; } my ($make_cxn, $o, $dp) = @args{@required_args}; my $slaves = []; my $method = $o->get('recursion-method'); PTDEBUG && _d('Slave recursion method:', $method); if ( !$method || $method =~ m/processlist|hosts/i ) { my @required_args = qw(dbh dsn); foreach my $arg ( @required_args ) { die "I need a $arg argument" unless $args{$arg}; } my ($dbh, $dsn) = @args{@required_args}; $self->recurse_to_slaves( { dbh => $dbh, dsn => $dsn, dsn_parser => $dp, recurse => $o->get('recurse'), method => $o->get('recursion-method'), callback => sub { my ( $dsn, $dbh, $level, $parent ) = @_; return unless $level; PTDEBUG && _d('Found slave:', $dp->as_string($dsn)); push @$slaves, $make_cxn->(dsn => $dsn, dbh => $dbh); return; }, } ); } elsif ( $method =~ m/^dsn=/i ) { my ($dsn_table_dsn) = $method =~ m/^dsn=(.+)/i; $slaves = $self->get_cxn_from_dsn_table( %args, dsn_table_dsn => $dsn_table_dsn, ); } else { die "Invalid --recursion-method: $method. Valid values are: " . "dsn=DSN, hosts, or processlist.\n"; } return $slaves; } sub recurse_to_slaves { my ( $self, $args, $level ) = @_; $level ||= 0; my $dp = $args->{dsn_parser}; my $dsn = $args->{dsn}; my $dbh; eval { $dbh = $args->{dbh} || $dp->get_dbh( $dp->get_cxn_params($dsn), { AutoCommit => 1 }); PTDEBUG && _d('Connected to', $dp->as_string($dsn)); }; if ( $EVAL_ERROR ) { print STDERR "Cannot connect to ", $dp->as_string($dsn), "\n" or die "Cannot print: $OS_ERROR"; return; } my $sql = 'SELECT @@SERVER_ID'; PTDEBUG && _d($sql); my ($id) = $dbh->selectrow_array($sql); PTDEBUG && _d('Working on server ID', $id); my $master_thinks_i_am = $dsn->{server_id}; if ( !defined $id || ( defined $master_thinks_i_am && $master_thinks_i_am != $id ) || $args->{server_ids_seen}->{$id}++ ) { PTDEBUG && _d('Server ID seen, or not what master said'); if ( $args->{skip_callback} ) { $args->{skip_callback}->($dsn, $dbh, $level, $args->{parent}); } return; } $args->{callback}->($dsn, $dbh, $level, $args->{parent}); if ( !defined $args->{recurse} || $level < $args->{recurse} ) { my @slaves = grep { !$_->{master_id} || $_->{master_id} == $id } # Only my slaves. $self->find_slave_hosts($dp, $dbh, $dsn, $args->{method}); foreach my $slave ( @slaves ) { PTDEBUG && _d('Recursing from', $dp->as_string($dsn), 'to', $dp->as_string($slave)); $self->recurse_to_slaves( { %$args, dsn => $slave, dbh => undef, parent => $dsn }, $level + 1 ); } } } sub find_slave_hosts { my ( $self, $dsn_parser, $dbh, $dsn, $method ) = @_; my @methods = qw(processlist hosts); if ( $method ) { @methods = grep { $_ ne $method } @methods; unshift @methods, $method; } else { if ( ($dsn->{P} || 3306) != 3306 ) { PTDEBUG && _d('Port number is non-standard; using only hosts method'); @methods = qw(hosts); } } PTDEBUG && _d('Looking for slaves on', $dsn_parser->as_string($dsn), 'using methods', @methods); my @slaves; METHOD: foreach my $method ( @methods ) { my $find_slaves = "_find_slaves_by_$method"; PTDEBUG && _d('Finding slaves with', $find_slaves); @slaves = $self->$find_slaves($dsn_parser, $dbh, $dsn); last METHOD if @slaves; } PTDEBUG && _d('Found', scalar(@slaves), 'slaves'); return @slaves; } sub _find_slaves_by_processlist { my ( $self, $dsn_parser, $dbh, $dsn ) = @_; my @slaves = map { my $slave = $dsn_parser->parse("h=$_", $dsn); $slave->{source} = 'processlist'; $slave; } grep { $_ } map { my ( $host ) = $_->{host} =~ m/^([^:]+):/; if ( $host eq 'localhost' ) { $host = '127.0.0.1'; # Replication never uses sockets. } $host; } $self->get_connected_slaves($dbh); return @slaves; } sub _find_slaves_by_hosts { my ( $self, $dsn_parser, $dbh, $dsn ) = @_; my @slaves; my $sql = 'SHOW SLAVE HOSTS'; PTDEBUG && _d($dbh, $sql); @slaves = @{$dbh->selectall_arrayref($sql, { Slice => {} })}; if ( @slaves ) { PTDEBUG && _d('Found some SHOW SLAVE HOSTS info'); @slaves = map { my %hash; @hash{ map { lc $_ } keys %$_ } = values %$_; my $spec = "h=$hash{host},P=$hash{port}" . ( $hash{user} ? ",u=$hash{user}" : '') . ( $hash{password} ? ",p=$hash{password}" : ''); my $dsn = $dsn_parser->parse($spec, $dsn); $dsn->{server_id} = $hash{server_id}; $dsn->{master_id} = $hash{master_id}; $dsn->{source} = 'hosts'; $dsn; } @slaves; } return @slaves; } sub get_connected_slaves { my ( $self, $dbh ) = @_; my $show = "SHOW GRANTS FOR "; my $user = 'CURRENT_USER()'; my $vp = $self->{VersionParser}; if ( $vp && !$vp->version_ge($dbh, '4.1.2') ) { $user = $dbh->selectrow_arrayref('SELECT USER()')->[0]; $user =~ s/([^@]+)@(.+)/'$1'\@'$2'/; } my $sql = $show . $user; PTDEBUG && _d($dbh, $sql); my $proc; eval { $proc = grep { m/ALL PRIVILEGES.*?\*\.\*|PROCESS/ } @{$dbh->selectcol_arrayref($sql)}; }; if ( $EVAL_ERROR ) { if ( $EVAL_ERROR =~ m/no such grant defined for user/ ) { PTDEBUG && _d('Retrying SHOW GRANTS without host; error:', $EVAL_ERROR); ($user) = split('@', $user); $sql = $show . $user; PTDEBUG && _d($sql); eval { $proc = grep { m/ALL PRIVILEGES.*?\*\.\*|PROCESS/ } @{$dbh->selectcol_arrayref($sql)}; }; } die "Failed to $sql: $EVAL_ERROR" if $EVAL_ERROR; } if ( !$proc ) { die "You do not have the PROCESS privilege"; } $sql = 'SHOW PROCESSLIST'; PTDEBUG && _d($dbh, $sql); grep { $_->{command} =~ m/Binlog Dump/i } map { # Lowercase the column names my %hash; @hash{ map { lc $_ } keys %$_ } = values %$_; \%hash; } @{$dbh->selectall_arrayref($sql, { Slice => {} })}; } sub is_master_of { my ( $self, $master, $slave ) = @_; my $master_status = $self->get_master_status($master) or die "The server specified as a master is not a master"; my $slave_status = $self->get_slave_status($slave) or die "The server specified as a slave is not a slave"; my @connected = $self->get_connected_slaves($master) or die "The server specified as a master has no connected slaves"; my (undef, $port) = $master->selectrow_array('SHOW VARIABLES LIKE "port"'); if ( $port != $slave_status->{master_port} ) { die "The slave is connected to $slave_status->{master_port} " . "but the master's port is $port"; } if ( !grep { $slave_status->{master_user} eq $_->{user} } @connected ) { die "I don't see any slave I/O thread connected with user " . $slave_status->{master_user}; } if ( ($slave_status->{slave_io_state} || '') eq 'Waiting for master to send event' ) { my ( $master_log_name, $master_log_num ) = $master_status->{file} =~ m/^(.*?)\.0*([1-9][0-9]*)$/; my ( $slave_log_name, $slave_log_num ) = $slave_status->{master_log_file} =~ m/^(.*?)\.0*([1-9][0-9]*)$/; if ( $master_log_name ne $slave_log_name || abs($master_log_num - $slave_log_num) > 1 ) { die "The slave thinks it is reading from " . "$slave_status->{master_log_file}, but the " . "master is writing to $master_status->{file}"; } } return 1; } sub get_master_dsn { my ( $self, $dbh, $dsn, $dsn_parser ) = @_; my $master = $self->get_slave_status($dbh) or return undef; my $spec = "h=$master->{master_host},P=$master->{master_port}"; return $dsn_parser->parse($spec, $dsn); } sub get_slave_status { my ( $self, $dbh ) = @_; if ( !$self->{not_a_slave}->{$dbh} ) { my $sth = $self->{sths}->{$dbh}->{SLAVE_STATUS} ||= $dbh->prepare('SHOW SLAVE STATUS'); PTDEBUG && _d($dbh, 'SHOW SLAVE STATUS'); $sth->execute(); my ($ss) = @{$sth->fetchall_arrayref({})}; if ( $ss && %$ss ) { $ss = { map { lc($_) => $ss->{$_} } keys %$ss }; # lowercase the keys return $ss; } PTDEBUG && _d('This server returns nothing for SHOW SLAVE STATUS'); $self->{not_a_slave}->{$dbh}++; } } sub get_master_status { my ( $self, $dbh ) = @_; if ( $self->{not_a_master}->{$dbh} ) { PTDEBUG && _d('Server on dbh', $dbh, 'is not a master'); return; } my $sth = $self->{sths}->{$dbh}->{MASTER_STATUS} ||= $dbh->prepare('SHOW MASTER STATUS'); PTDEBUG && _d($dbh, 'SHOW MASTER STATUS'); $sth->execute(); my ($ms) = @{$sth->fetchall_arrayref({})}; PTDEBUG && _d( $ms ? map { "$_=" . (defined $ms->{$_} ? $ms->{$_} : '') } keys %$ms : ''); if ( !$ms || scalar keys %$ms < 2 ) { PTDEBUG && _d('Server on dbh', $dbh, 'does not seem to be a master'); $self->{not_a_master}->{$dbh}++; } return { map { lc($_) => $ms->{$_} } keys %$ms }; # lowercase the keys } sub wait_for_master { my ( $self, %args ) = @_; my @required_args = qw(master_status slave_dbh); foreach my $arg ( @required_args ) { die "I need a $arg argument" unless $args{$arg}; } my ($master_status, $slave_dbh) = @args{@required_args}; my $timeout = $args{timeout} || 60; my $result; my $waited; if ( $master_status ) { my $sql = "SELECT MASTER_POS_WAIT('$master_status->{file}', " . "$master_status->{position}, $timeout)"; PTDEBUG && _d($slave_dbh, $sql); my $start = time; ($result) = $slave_dbh->selectrow_array($sql); $waited = time - $start; PTDEBUG && _d('Result of waiting:', $result); PTDEBUG && _d("Waited", $waited, "seconds"); } else { PTDEBUG && _d('Not waiting: this server is not a master'); } return { result => $result, waited => $waited, }; } sub stop_slave { my ( $self, $dbh ) = @_; my $sth = $self->{sths}->{$dbh}->{STOP_SLAVE} ||= $dbh->prepare('STOP SLAVE'); PTDEBUG && _d($dbh, $sth->{Statement}); $sth->execute(); } sub start_slave { my ( $self, $dbh, $pos ) = @_; if ( $pos ) { my $sql = "START SLAVE UNTIL MASTER_LOG_FILE='$pos->{file}', " . "MASTER_LOG_POS=$pos->{position}"; PTDEBUG && _d($dbh, $sql); $dbh->do($sql); } else { my $sth = $self->{sths}->{$dbh}->{START_SLAVE} ||= $dbh->prepare('START SLAVE'); PTDEBUG && _d($dbh, $sth->{Statement}); $sth->execute(); } } sub catchup_to_master { my ( $self, $slave, $master, $timeout ) = @_; $self->stop_slave($master); $self->stop_slave($slave); my $slave_status = $self->get_slave_status($slave); my $slave_pos = $self->repl_posn($slave_status); my $master_status = $self->get_master_status($master); my $master_pos = $self->repl_posn($master_status); PTDEBUG && _d('Master position:', $self->pos_to_string($master_pos), 'Slave position:', $self->pos_to_string($slave_pos)); my $result; if ( $self->pos_cmp($slave_pos, $master_pos) < 0 ) { PTDEBUG && _d('Waiting for slave to catch up to master'); $self->start_slave($slave, $master_pos); $result = $self->wait_for_master( master_status => $master_status, slave_dbh => $slave, timeout => $timeout, master_status => $master_status ); if ( !defined $result->{result} ) { $slave_status = $self->get_slave_status($slave); if ( !$self->slave_is_running($slave_status) ) { PTDEBUG && _d('Master position:', $self->pos_to_string($master_pos), 'Slave position:', $self->pos_to_string($slave_pos)); $slave_pos = $self->repl_posn($slave_status); if ( $self->pos_cmp($slave_pos, $master_pos) != 0 ) { die "MASTER_POS_WAIT() returned NULL but slave has not " . "caught up to master"; } PTDEBUG && _d('Slave is caught up to master and stopped'); } else { die "Slave has not caught up to master and it is still running"; } } } else { PTDEBUG && _d("Slave is already caught up to master"); } return $result; } sub catchup_to_same_pos { my ( $self, $s1_dbh, $s2_dbh ) = @_; $self->stop_slave($s1_dbh); $self->stop_slave($s2_dbh); my $s1_status = $self->get_slave_status($s1_dbh); my $s2_status = $self->get_slave_status($s2_dbh); my $s1_pos = $self->repl_posn($s1_status); my $s2_pos = $self->repl_posn($s2_status); if ( $self->pos_cmp($s1_pos, $s2_pos) < 0 ) { $self->start_slave($s1_dbh, $s2_pos); } elsif ( $self->pos_cmp($s2_pos, $s1_pos) < 0 ) { $self->start_slave($s2_dbh, $s1_pos); } $s1_status = $self->get_slave_status($s1_dbh); $s2_status = $self->get_slave_status($s2_dbh); $s1_pos = $self->repl_posn($s1_status); $s2_pos = $self->repl_posn($s2_status); if ( $self->slave_is_running($s1_status) || $self->slave_is_running($s2_status) || $self->pos_cmp($s1_pos, $s2_pos) != 0) { die "The servers aren't both stopped at the same position"; } } sub slave_is_running { my ( $self, $slave_status ) = @_; return ($slave_status->{slave_sql_running} || 'No') eq 'Yes'; } sub has_slave_updates { my ( $self, $dbh ) = @_; my $sql = q{SHOW VARIABLES LIKE 'log_slave_updates'}; PTDEBUG && _d($dbh, $sql); my ($name, $value) = $dbh->selectrow_array($sql); return $value && $value =~ m/^(1|ON)$/; } sub repl_posn { my ( $self, $status ) = @_; if ( exists $status->{file} && exists $status->{position} ) { return { file => $status->{file}, position => $status->{position}, }; } else { return { file => $status->{relay_master_log_file}, position => $status->{exec_master_log_pos}, }; } } sub get_slave_lag { my ( $self, $dbh ) = @_; my $stat = $self->get_slave_status($dbh); return unless $stat; # server is not a slave return $stat->{seconds_behind_master}; } sub pos_cmp { my ( $self, $a, $b ) = @_; return $self->pos_to_string($a) cmp $self->pos_to_string($b); } sub short_host { my ( $self, $dsn ) = @_; my ($host, $port); if ( $dsn->{master_host} ) { $host = $dsn->{master_host}; $port = $dsn->{master_port}; } else { $host = $dsn->{h}; $port = $dsn->{P}; } return ($host || '[default]') . ( ($port || 3306) == 3306 ? '' : ":$port" ); } sub is_replication_thread { my ( $self, $query, %args ) = @_; return unless $query; my $type = lc($args{type} || 'all'); die "Invalid type: $type" unless $type =~ m/^binlog_dump|slave_io|slave_sql|all$/i; my $match = 0; if ( $type =~ m/binlog_dump|all/i ) { $match = 1 if ($query->{Command} || $query->{command} || '') eq "Binlog Dump"; } if ( !$match ) { if ( ($query->{User} || $query->{user} || '') eq "system user" ) { PTDEBUG && _d("Slave replication thread"); if ( $type ne 'all' ) { my $state = $query->{State} || $query->{state} || ''; if ( $state =~ m/^init|end$/ ) { PTDEBUG && _d("Special state:", $state); $match = 1; } else { my ($slave_sql) = $state =~ m/ ^(Waiting\sfor\sthe\snext\sevent |Reading\sevent\sfrom\sthe\srelay\slog |Has\sread\sall\srelay\slog;\swaiting |Making\stemp\sfile |Waiting\sfor\sslave\smutex\son\sexit)/xi; $match = $type eq 'slave_sql' && $slave_sql ? 1 : $type eq 'slave_io' && !$slave_sql ? 1 : 0; } } else { $match = 1; } } else { PTDEBUG && _d('Not system user'); } if ( !defined $args{check_known_ids} || $args{check_known_ids} ) { my $id = $query->{Id} || $query->{id}; if ( $match ) { $self->{replication_thread}->{$id} = 1; } else { if ( $self->{replication_thread}->{$id} ) { PTDEBUG && _d("Thread ID is a known replication thread ID"); $match = 1; } } } } PTDEBUG && _d('Matches', $type, 'replication thread:', ($match ? 'yes' : 'no'), '; match:', $match); return $match; } sub get_replication_filters { my ( $self, %args ) = @_; my @required_args = qw(dbh); foreach my $arg ( @required_args ) { die "I need a $arg argument" unless $args{$arg}; } my ($dbh) = @args{@required_args}; my %filters = (); my $status = $self->get_master_status($dbh); if ( $status ) { map { $filters{$_} = $status->{$_} } grep { defined $status->{$_} && $status->{$_} ne '' } qw( binlog_do_db binlog_ignore_db ); } $status = $self->get_slave_status($dbh); if ( $status ) { map { $filters{$_} = $status->{$_} } grep { defined $status->{$_} && $status->{$_} ne '' } qw( replicate_do_db replicate_ignore_db replicate_do_table replicate_ignore_table replicate_wild_do_table replicate_wild_ignore_table ); my $sql = "SHOW VARIABLES LIKE 'slave_skip_errors'"; PTDEBUG && _d($dbh, $sql); my $row = $dbh->selectrow_arrayref($sql); $filters{slave_skip_errors} = $row->[1] if $row->[1] && $row->[1] ne 'OFF'; } return \%filters; } sub pos_to_string { my ( $self, $pos ) = @_; my $fmt = '%s/%020d'; return sprintf($fmt, @{$pos}{qw(file position)}); } sub reset_known_replication_threads { my ( $self ) = @_; $self->{replication_thread} = {}; return; } sub get_cxn_from_dsn_table { my ($self, %args) = @_; my @required_args = qw(dsn_table_dsn make_cxn DSNParser Quoter); foreach my $arg ( @required_args ) { die "I need a $arg argument" unless $args{$arg}; } my ($dsn_table_dsn, $make_cxn, $dp, $q) = @args{@required_args}; PTDEBUG && _d('DSN table DSN:', $dsn_table_dsn); my $dsn = $dp->parse($dsn_table_dsn); my $dsn_table; if ( $dsn->{D} && $dsn->{t} ) { $dsn_table = $q->quote($dsn->{D}, $dsn->{t}); } elsif ( $dsn->{t} && $dsn->{t} =~ m/\./ ) { $dsn_table = $q->quote($q->split_unquote($dsn->{t})); } else { die "DSN table DSN does not specify a database (D) " . "or a database-qualified table (t)"; } my $dsn_tbl_cxn = $make_cxn->(dsn => $dsn); my $dbh = $dsn_tbl_cxn->connect(); my $sql = "SELECT dsn FROM $dsn_table ORDER BY id"; PTDEBUG && _d($sql); my $dsn_strings = $dbh->selectcol_arrayref($sql); my @cxn; if ( $dsn_strings ) { foreach my $dsn_string ( @$dsn_strings ) { PTDEBUG && _d('DSN from DSN table:', $dsn_string); push @cxn, $make_cxn->(dsn_string => $dsn_string); } } return \@cxn; } sub _d { my ($package, undef, $line) = caller 0; @_ = map { (my $temp = $_) =~ s/\n/\n# /g; $temp; } map { defined $_ ? $_ : 'undef' } @_; print STDERR "# $package:$line $PID ", join(' ', @_), "\n"; } 1; } # ########################################################################### # End MasterSlave package # ########################################################################### # ########################################################################### # QueryRewriter package # This package is a copy without comments from the original. The original # with comments and its test file can be found in the Bazaar repository at, # lib/QueryRewriter.pm # t/lib/QueryRewriter.t # See https://launchpad.net/percona-toolkit for more information. # ########################################################################### { package QueryRewriter; use strict; use warnings FATAL => 'all'; use English qw(-no_match_vars); use constant PTDEBUG => $ENV{PTDEBUG} || 0; our $verbs = qr{^SHOW|^FLUSH|^COMMIT|^ROLLBACK|^BEGIN|SELECT|INSERT |UPDATE|DELETE|REPLACE|^SET|UNION|^START|^LOCK}xi; my $quote_re = qr/"(?:(?!(? [^()]+ ) # Non-parens without backtracking | (??{ $bal }) # Group with matching parens )* \) /x; my $olc_re = qr/(?:--|#)[^'"\r\n]*(?=[\r\n]|\Z)/; # One-line comments my $mlc_re = qr#/\*[^!].*?\*/#sm; # But not /*!version */ my $vlc_re = qr#/\*.*?[0-9+].*?\*/#sm; # For SHOW + /*!version */ my $vlc_rf = qr#^(SHOW).*?/\*![0-9+].*?\*/#sm; # Variation for SHOW sub new { my ( $class, %args ) = @_; my $self = { %args }; return bless $self, $class; } sub strip_comments { my ( $self, $query ) = @_; return unless $query; $query =~ s/$olc_re//go; $query =~ s/$mlc_re//go; if ( $query =~ m/$vlc_rf/i ) { # contains show + version $query =~ s/$vlc_re//go; } return $query; } sub shorten { my ( $self, $query, $length ) = @_; $query =~ s{ \A( (?:INSERT|REPLACE) (?:\s+LOW_PRIORITY|DELAYED|HIGH_PRIORITY|IGNORE)? (?:\s\w+)*\s+\S+\s+VALUES\s*\(.*?\) ) \s*,\s*\(.*?(ON\s+DUPLICATE|\Z)} {$1 /*... omitted ...*/$2}xsi; return $query unless $query =~ m/IN\s*\(\s*(?!select)/i; my $last_length = 0; my $query_length = length($query); while ( $length > 0 && $query_length > $length && $query_length < ( $last_length || $query_length + 1 ) ) { $last_length = $query_length; $query =~ s{ (\bIN\s*\() # The opening of an IN list ([^\)]+) # Contents of the list, assuming no item contains paren (?=\)) # Close of the list } { $1 . __shorten($2) }gexsi; } return $query; } sub __shorten { my ( $snippet ) = @_; my @vals = split(/,/, $snippet); return $snippet unless @vals > 20; my @keep = splice(@vals, 0, 20); # Remove and save the first 20 items return join(',', @keep) . "/*... omitted " . scalar(@vals) . " items ...*/"; } sub fingerprint { my ( $self, $query ) = @_; $query =~ m#\ASELECT /\*!40001 SQL_NO_CACHE \*/ \* FROM `# # mysqldump query && return 'mysqldump'; $query =~ m#/\*\w+\.\w+:[0-9]/[0-9]\*/# # pt-table-checksum, etc query && return 'percona-toolkit'; $query =~ m/\Aadministrator command: / && return $query; $query =~ m/\A\s*(call\s+\S+)\(/i && return lc($1); # Warning! $1 used, be careful. if ( my ($beginning) = $query =~ m/\A((?:INSERT|REPLACE)(?: IGNORE)?\s+INTO.+?VALUES\s*\(.*?\))\s*,\s*\(/is ) { $query = $beginning; # Shorten multi-value INSERT statements ASAP } $query =~ s/$olc_re//go; $query =~ s/$mlc_re//go; $query =~ s/\Ause \S+\Z/use ?/i # Abstract the DB in USE && return $query; $query =~ s/\\["']//g; # quoted strings $query =~ s/".*?"/?/sg; # quoted strings $query =~ s/'.*?'/?/sg; # quoted strings $query =~ s/[0-9+-][0-9a-f.xb+-]*/?/g;# Anything vaguely resembling numbers $query =~ s/[xb.+-]\?/?/g; # Clean up leftovers $query =~ s/\A\s+//; # Chop off leading whitespace chomp $query; # Kill trailing whitespace $query =~ tr[ \n\t\r\f][ ]s; # Collapse whitespace $query = lc $query; $query =~ s/\bnull\b/?/g; # Get rid of NULLs $query =~ s{ # Collapse IN and VALUES lists \b(in|values?)(?:[\s,]*\([\s?,]*\))+ } {$1(?+)}gx; $query =~ s{ # Collapse UNION \b(select\s.*?)(?:(\sunion(?:\sall)?)\s\1)+ } {$1 /*repeat$2*/}xg; $query =~ s/\blimit \?(?:, ?\?| offset \?)?/limit ?/; # LIMIT if ( $query =~ m/\bORDER BY /gi ) { # Find, anchor on ORDER BY clause 1 while $query =~ s/\G(.+?)\s+ASC/$1/gi && pos $query; } return $query; } sub distill_verbs { my ( $self, $query ) = @_; $query =~ m/\A\s*call\s+(\S+)\(/i && return "CALL $1"; $query =~ m/\A\s*use\s+/ && return "USE"; $query =~ m/\A\s*UNLOCK TABLES/i && return "UNLOCK"; $query =~ m/\A\s*xa\s+(\S+)/i && return "XA_$1"; if ( $query =~ m/\Aadministrator command:/ ) { $query =~ s/administrator command:/ADMIN/; $query = uc $query; return $query; } $query = $self->strip_comments($query); if ( $query =~ m/\A\s*SHOW\s+/i ) { PTDEBUG && _d($query); $query = uc $query; $query =~ s/\s+(?:GLOBAL|SESSION|FULL|STORAGE|ENGINE)\b/ /g; $query =~ s/\s+COUNT[^)]+\)//g; $query =~ s/\s+(?:FOR|FROM|LIKE|WHERE|LIMIT|IN)\b.+//ms; $query =~ s/\A(SHOW(?:\s+\S+){1,2}).*\Z/$1/s; $query =~ s/\s+/ /g; PTDEBUG && _d($query); return $query; } eval $QueryParser::data_def_stmts; eval $QueryParser::tbl_ident; my ( $dds ) = $query =~ /^\s*($QueryParser::data_def_stmts)\b/i; if ( $dds) { my ( $obj ) = $query =~ m/$dds.+(DATABASE|TABLE)\b/i; $obj = uc $obj if $obj; PTDEBUG && _d('Data def statment:', $dds, 'obj:', $obj); my ($db_or_tbl) = $query =~ m/(?:TABLE|DATABASE)\s+($QueryParser::tbl_ident)(\s+.*)?/i; PTDEBUG && _d('Matches db or table:', $db_or_tbl); return uc($dds . ($obj ? " $obj" : '')), $db_or_tbl; } my @verbs = $query =~ m/\b($verbs)\b/gio; @verbs = do { my $last = ''; grep { my $pass = $_ ne $last; $last = $_; $pass } map { uc } @verbs; }; if ( ($verbs[0] || '') eq 'SELECT' && @verbs > 1 ) { PTDEBUG && _d("False-positive verbs after SELECT:", @verbs[1..$#verbs]); my $union = grep { $_ eq 'UNION' } @verbs; @verbs = $union ? qw(SELECT UNION) : qw(SELECT); } my $verb_str = join(q{ }, @verbs); return $verb_str; } sub __distill_tables { my ( $self, $query, $table, %args ) = @_; my $qp = $args{QueryParser} || $self->{QueryParser}; die "I need a QueryParser argument" unless $qp; my @tables = map { $_ =~ s/`//g; $_ =~ s/(_?)[0-9]+/$1?/g; $_; } grep { defined $_ } $qp->get_tables($query); push @tables, $table if $table; @tables = do { my $last = ''; grep { my $pass = $_ ne $last; $last = $_; $pass } @tables; }; return @tables; } sub distill { my ( $self, $query, %args ) = @_; if ( $args{generic} ) { my ($cmd, $arg) = $query =~ m/^(\S+)\s+(\S+)/; return '' unless $cmd; $query = (uc $cmd) . ($arg ? " $arg" : ''); } else { my ($verbs, $table) = $self->distill_verbs($query, %args); if ( $verbs && $verbs =~ m/^SHOW/ ) { my %alias_for = qw( SCHEMA DATABASE KEYS INDEX INDEXES INDEX ); map { $verbs =~ s/$_/$alias_for{$_}/ } keys %alias_for; $query = $verbs; } else { my @tables = $self->__distill_tables($query, $table, %args); $query = join(q{ }, $verbs, @tables); } } if ( $args{trf} ) { $query = $args{trf}->($query, %args); } return $query; } sub convert_to_select { my ( $self, $query ) = @_; return unless $query; return if $query =~ m/=\s*\(\s*SELECT /i; $query =~ s{ \A.*? update(?:\s+(?:low_priority|ignore))?\s+(.*?) \s+set\b(.*?) (?:\s*where\b(.*?))? (limit\s*[0-9]+(?:\s*,\s*[0-9]+)?)? \Z } {__update_to_select($1, $2, $3, $4)}exsi || $query =~ s{ \A.*? (?:insert(?:\s+ignore)?|replace)\s+ .*?\binto\b(.*?)\(([^\)]+)\)\s* values?\s*(\(.*?\))\s* (?:\blimit\b|on\s+duplicate\s+key.*)?\s* \Z } {__insert_to_select($1, $2, $3)}exsi || $query =~ s{ \A.*? (?:insert(?:\s+ignore)?|replace)\s+ (?:.*?\binto)\b(.*?)\s* set\s+(.*?)\s* (?:\blimit\b|on\s+duplicate\s+key.*)?\s* \Z } {__insert_to_select_with_set($1, $2)}exsi || $query =~ s{ \A.*? delete\s+(.*?) \bfrom\b(.*) \Z } {__delete_to_select($1, $2)}exsi; $query =~ s/\s*on\s+duplicate\s+key\s+update.*\Z//si; $query =~ s/\A.*?(?=\bSELECT\s*\b)//ism; return $query; } sub convert_select_list { my ( $self, $query ) = @_; $query =~ s{ \A\s*select(.*?)\bfrom\b } {$1 =~ m/\*/ ? "select 1 from" : "select isnull(coalesce($1)) from"}exi; return $query; } sub __delete_to_select { my ( $delete, $join ) = @_; if ( $join =~ m/\bjoin\b/ ) { return "select 1 from $join"; } return "select * from $join"; } sub __insert_to_select { my ( $tbl, $cols, $vals ) = @_; PTDEBUG && _d('Args:', @_); my @cols = split(/,/, $cols); PTDEBUG && _d('Cols:', @cols); $vals =~ s/^\(|\)$//g; # Strip leading/trailing parens my @vals = $vals =~ m/($quote_re|[^,]*${bal}[^,]*|[^,]+)/g; PTDEBUG && _d('Vals:', @vals); if ( @cols == @vals ) { return "select * from $tbl where " . join(' and ', map { "$cols[$_]=$vals[$_]" } (0..$#cols)); } else { return "select * from $tbl limit 1"; } } sub __insert_to_select_with_set { my ( $from, $set ) = @_; $set =~ s/,/ and /g; return "select * from $from where $set "; } sub __update_to_select { my ( $from, $set, $where, $limit ) = @_; return "select $set from $from " . ( $where ? "where $where" : '' ) . ( $limit ? " $limit " : '' ); } sub wrap_in_derived { my ( $self, $query ) = @_; return unless $query; return $query =~ m/\A\s*select/i ? "select 1 from ($query) as x limit 1" : $query; } sub _d { my ($package, undef, $line) = caller 0; @_ = map { (my $temp = $_) =~ s/\n/\n# /g; $temp; } map { defined $_ ? $_ : 'undef' } @_; print STDERR "# $package:$line $PID ", join(' ', @_), "\n"; } 1; } # ########################################################################### # End QueryRewriter package # ########################################################################### # ########################################################################### # This is a combination of modules and programs in one -- a runnable module. # http://www.perl.com/pub/a/2006/07/13/lightning-articles.html?page=last # Or, look it up in the Camel book on pages 642 and 643 in the 3rd edition. # # Check at the end of this package for the call to main() which actually runs # the program. # ########################################################################### package pt_kill; use English qw(-no_match_vars); use POSIX qw(setsid); use List::Util qw(max); use Data::Dumper; $Data::Dumper::Indent = 1; $Data::Dumper::Sortkeys = 1; $Data::Dumper::Quotekeys = 0; Transformers->import(qw(ts)); use constant PTDEBUG => $ENV{PTDEBUG} || 0; $OUTPUT_AUTOFLUSH = 1; my $o; # ######################################################################## # Configuration info. # ######################################################################## sub main { @ARGV = @_; # set global ARGV for this package my $ms = new MasterSlave(); my $pl = new Processlist(MasterSlave => $ms); my $vp = new VersionParser(); my $qr = new QueryRewriter(); # ######################################################################## # Get configuration information. # ######################################################################## $o = new OptionParser(); $o->get_specs(); $o->get_opts(); my $dp = $o->DSNParser(); $dp->prop('set-vars', $o->get('set-vars')); if ( !$o->got('busy-time') ) { $o->set('interval', 30) unless $o->got('interval'); } else { $o->set('interval', max(1, $o->get('busy-time') / 2)) unless $o->got('interval'); } # Disable opts that don't make sense when reading proclist # from a file (or STDIN). if ( $o->get('test-matching') ) { $o->set('run-time', 0); $o->set('interval', 0); $o->set('ignore-self', 0); } # TODO: parse valid values from POD my $victims = lc $o->get('victims'); if ( !grep { $victims eq $_ } qw(oldest all all-but-oldest) ) { $o->save_error("Invalid value for --victims: $victims"); } $o->usage_or_errors(); # ######################################################################## # First things first: if --stop was given, create the sentinel file. # ######################################################################## if ( $o->get('stop') ) { my $sentinel = $o->get('sentinel'); PTDEBUG && _d('Creating sentinel file', $sentinel); open my $fh, '>', $sentinel or die "Cannot open $sentinel: $OS_ERROR\n"; print $fh "Remove this file to permit pt-kill to run.\n" or die "Cannot write to $sentinel: $OS_ERROR\n"; close $fh or die "Cannot close $sentinel: $OS_ERROR\n"; print "Successfully created file $sentinel\n"; return 0; } # ######################################################################## # Create the --filter sub. # ######################################################################## my $filter_sub; if ( my $filter = $o->get('filter') ) { if ( -f $filter && -r $filter ) { PTDEBUG && _d('Reading file', $filter, 'for --filter code'); open my $fh, "<", $filter or die "Cannot open $filter: $OS_ERROR"; $filter = do { local $/ = undef; <$fh> }; close $fh; } else { $filter = "( $filter )"; # issue 565 } my $code = 'sub { my ( $event ) = @_; ' . "$filter && return \$event; };"; PTDEBUG && _d('--filter code:', $code); $filter_sub = eval $code or die "Error compiling --filter code: $code\n$EVAL_ERROR"; } # ######################################################################## # Make input sub that will either get processlist from MySQL or a file. # ######################################################################## my $dsn; my $dbh; my $kill_sth; my $get_proclist; my $files; if ( $files = $o->get('test-matching') ) { PTDEBUG && _d('Getting processlist from files:', @$files); my $trp = new TextResultSetParser(); my $fh; $get_proclist = sub { if ( !$fh ) { my $file = shift @$files; die 'No more files' unless $file; if ( $file eq '-' ) { $fh = *STDIN; } else { if ( !open $fh, '<', $file ) { warn "Cannot open $file: $OS_ERROR"; $fh = undef; return; } } } if ( $fh ) { local $INPUT_RECORD_SEPARATOR = ''; my $proclist_text = <$fh>; if ( $proclist_text ) { return $trp->parse($proclist_text); } else { # No more proclists in this file. $fh = undef; } } return; }; } else { PTDEBUG && _d('Getting processlist from MySQL'); $dsn = $dp->parse_options($o); $dbh = get_cxn($dp, $dsn, 1); $kill_sth = $o->get('kill-query') ? $dbh->prepare('KILL QUERY ?') : $dbh->prepare('KILL ?'); my $proc_sth = $dbh->prepare('SHOW FULL PROCESSLIST'); $get_proclist = sub { $proc_sth->execute(); return $proc_sth->fetchall_arrayref({}); }; } # ######################################################################## # Daemonize only after (potentially) asking for passwords for --ask-pass. # ######################################################################## my $daemon; if ( $o->get('daemonize') ) { $daemon = new Daemon(o=>$o); $daemon->daemonize(); PTDEBUG && _d('I am a daemon now'); } elsif ( $o->get('pid') ) { # We're not daemoninzing, it just handles PID stuff. $daemon = new Daemon(o=>$o); $daemon->make_PID_file(); } # ######################################################################## # Start working. # ######################################################################## msg("$PROGRAM_NAME starting"); msg($dbh ? "Connected to host " . $dp->as_string($dsn) : "Test matching files @$files"); # Class-based match criteria. my $query_count = $o->get('query-count'); my $each_busy_time = $o->get('each-busy-time'); my $any_busy_time = $o->get('any-busy-time'); my $group_by = $o->get('group-by'); if ( $group_by && $group_by =~ m/id|user|host|db|command|time|state|info/i ) { # Processlist.pm is case-sensitive. It matches Id, Host, db, etc. # So we'll do the same because if we set NAME_lc on the dbh then # we'll break our Processlist obj. $group_by = lc $group_by; $group_by = ucfirst $group_by unless $group_by eq 'db'; } # Per-class match criteria. my %find_spec = ( busy_time => $o->get('busy-time'), idle_time => $o->get('idle-time'), all => $o->get('match-all'), replication_threads => $o->get('replication-threads'), ignore => { Command => $o->get('ignore-command'), db => $o->get('ignore-db'), Host => $o->get('ignore-host'), Id => $o->get('ignore-self') ? $dbh->{mysql_thread_id} : undef, Info => $o->get('ignore-info'), State => $o->get('ignore-state'), User => $o->get('ignore-user'), }, match => { Command => $o->get('match-command'), db => $o->get('match-db'), Host => $o->get('match-host'), Info => $o->get('match-info'), State => $o->get('match-state'), User => $o->get('match-user'), }, ); msg("Find spec: " . Dumper(\%find_spec)); my $sentinel = $o->get('sentinel'); my $run_time = $o->get('run-time') || 0; my $start = time(); my $end = $start + $run_time; # When we should exit my $now = $start; if ( $dbh ) { msg("Run-time: " . ($run_time ? "$run_time seconds" : "forever") . " at " . ($o->get('interval') || 0) . " second intervals"); } # We don't care about the executed command, and we don't want # to wait for it, so we ignore dead children so we don't have # to reap them and they won't become zombies. # https://bugs.launchpad.net/percona-toolkit/+bug/919819 if ( $o->get('execute-command') ) { $SIG{CHLD} = 'IGNORE'; } while ( (!$run_time || $now < $end) && !-f $sentinel ) { msg('Checking processlist'); my $proclist; eval { $proclist = $get_proclist->(); }; if ( $EVAL_ERROR ) { last if $EVAL_ERROR =~ m/No more files/; die "Error getting SHOW PROCESSLIST: $EVAL_ERROR"; } # Apply --filter to the processlist events. my $filtered_proclist; if ( $filter_sub && $proclist && @$proclist ) { foreach my $proc ( @$proclist ) { push @$filtered_proclist, $proc if $filter_sub->($proc); } } else { $filtered_proclist = $proclist; } my @queries; if ( $proclist ) { # ################################################################## # Group queries into classes. If --group-by wasn't specified # then all queries will be put in the "default" class. # ################################################################## my $query_classes = group_queries( proclist => $proclist, group_by => $group_by, strip_comments => $o->get('strip-comments'), QueryRewriter => $qr, ); # ################################################################## # Find matching queries in each class. # ################################################################## CLASS: foreach my $class ( keys %$query_classes ) { PTDEBUG && _d('Finding matching queries in class', $class); my @matches = $pl->find($query_classes->{$class}, %find_spec); PTDEBUG && _d(scalar @matches, 'queries in class', $class); next CLASS unless scalar @matches; # ############################################################### # Apply class-based filters. # ############################################################### if ( $query_count && @matches < $query_count ) { PTDEBUG && _d('Not enough queries in class', $class, '; has', scalar @matches, 'but needs at least', $query_count); next CLASS; } if ( $each_busy_time ) { foreach my $proc ( @matches ) { if ( ($proc->{Time} || 0) <= $each_busy_time ) { PTDEBUG && _d('This query in class', $class, 'hasn\'t been running long enough:', Dumper($proc)); next CLASS; } } } elsif ( $any_busy_time ) { my $busy_enough = 0; foreach my $proc ( @matches ) { if ( ($proc->{Time} || 0) > $any_busy_time ) { $busy_enough = 1; last; } } if ( !$busy_enough ) { PTDEBUG && _d('No query is busy enough in class', $class); next CLASS; } } # ############################################################### # Select the victims (which of the matching queries to kill). # ############################################################### @matches = reverse sort { ($a->{Time} || 0) <=> ($b->{Time} || 0) } @matches; if ( $victims eq 'oldest' ) { @matches = ($matches[0]); } elsif ( $victims eq 'all-but-oldest' ) { shift @matches; # remove fist/oldest query } elsif ( $victims eq 'all' ) { # Don't do anything. } else { # Shouldn't happen. Option val should be verified earlier. die "I don't know how to kill $victims"; } # ############################################################### # Save matching queries in this class. # ############################################################### PTDEBUG && _d(scalar @matches, 'queries to kill in class', $class); push @queries, @matches; } # CLASS msg('Matched ' . scalar @queries . ' queries'); MATCHING_QUERY: foreach my $query ( @queries ) { if ( $o->get('print') ) { printf "# %s %s %d (%s %d sec) %s\n", ts(time), $o->get('kill-query') ? 'KILL QUERY' : 'KILL', $query->{Id}, ($query->{Command} || 'NULL'), $query->{Time}, ($query->{Info} || 'NULL'); } if ( $o->get('execute-command') ) { exec_cmd($o->get('execute-command')); msg('Executed ' . $o->get('execute-command')); } if ( $o->get('kill') || $o->get('kill-query') ) { if ( $o->get('wait-before-kill') ) { msg("Sleeping " . $o->get('wait-before-kill') . " seconds before kill"); sleep $o->get('wait-before-kill'); } eval { $kill_sth->execute($query->{Id}); }; if ( $EVAL_ERROR ) { msg("Error killing $query->{Id}: $EVAL_ERROR"); } else { msg("Killed $query->{Id}"); } } } } else { msg('Processlist returned no queries'); } if ( $dbh ) { if ( @queries && $o->get('wait-after-kill') ) { msg("Sleeping " . $o->get('wait-after-kill') . " seconds after killing queries"); sleep $o->get('wait-after-kill'); } else { msg("Sleeping " . $o->get('interval') . " seconds after normal interval"); sleep $o->get('interval'); } } $now = time(); } msg("Sentinel file $sentinel exists") if $sentinel && -f $sentinel; msg("$PROGRAM_NAME ending"); return 0; } # ############################################################################ # Subroutines. # ############################################################################ sub get_cxn { my ( $dp, $dsn, $ac ) = @_; if ( $o->get('ask-pass') ) { $dsn->{p} = OptionParser::prompt_noecho("Enter password: "); } my $dbh = $dp->get_dbh($dp->get_cxn_params($dsn), {AutoCommit => $ac}); $dbh->{InactiveDestroy} = 1; # Because of forking. return $dbh; } # Forks and detaches from parent to execute the given command; # does not block parent. sub exec_cmd { my ( $cmd ) = @_; PTDEBUG && _d('exec cmd:', $cmd); return unless $cmd; my $pid = fork(); if ( $pid ) { # parent PTDEBUG && _d('child pid:', $pid); return $pid; } # child POSIX::setsid() or die "Cannot start a new session: $OS_ERROR"; my $retval = system($cmd); $retval = $retval >> 8; PTDEBUG && _d('child exit status:', $retval); exit $retval; } sub msg { my ( $msg ) = @_; print '# ', ts(time), " $msg\n" if $o->get('verbose'); PTDEBUG && _d($msg); return; } sub group_queries { my ( %args ) = @_; my ($proclist, $group_by, $qr) = @args{qw(proclist group_by QueryRewriter)}; PTDEBUG && _d("Grouping queries by", $group_by); # If there's proclist then there's nothing to do. If there's no group by # then all the procs in the list are in the same class. return $proclist unless $proclist; return { 'default' => $proclist } unless $group_by; my $query_classes = {}; foreach my $proc ( @$proclist ) { if ( $args{strip_comments} && $proc->{Info} ) { $proc->{Info} = $qr->strip_comments($proc->{Info}); } my $class; if ( $group_by eq 'fingerprint' ) { $class = $proc->{Info} ? $qr->fingerprint($proc->{Info}) : 'NULL'; } else { $class = $proc->{$group_by} ? $proc->{$group_by} : 'NULL'; } push @{$query_classes->{$class}}, $proc; } return $query_classes; } sub _d { my ($package, undef, $line) = caller 0; @_ = map { (my $temp = $_) =~ s/\n/\n# /g; $temp; } map { defined $_ ? $_ : 'undef' } @_; print STDERR "# $package:$line $PID ", join(' ', @_), "\n"; } # ############################################################################ # Run the program. # ############################################################################ if ( !caller ) { exit main(@ARGV); } 1; # Because this is a module as well as a script. # ############################################################################ # Documentation # ############################################################################ =pod =head1 NAME pt-kill - Kill MySQL queries that match certain criteria. =head1 SYNOPSIS Usage: pt-kill [OPTIONS] pt-kill kills MySQL connections. pt-kill connects to MySQL and gets queries from SHOW PROCESSLIST if no FILE is given. Else, it reads queries from one or more FILE which contains the output of SHOW PROCESSLIST. If FILE is -, pt-kill reads from STDIN. Kill queries running longer than 60s: pt-kill --busy-time 60 --kill Print, do not kill, queries running longer than 60s: pt-kill --busy-time 60 --print Check for sleeping processes and kill them all every 10s: pt-kill --match-command Sleep --kill --victims all --interval 10 Print all login processes: pt-kill --match-state login --print --victims all See which queries in the processlist right now would match: mysql -e "SHOW PROCESSLIST" > proclist.txt pt-kill --test-matching proclist.txt --busy-time 60 --print =head1 RISKS The following section is included to inform users about the potential risks, whether known or unknown, of using this tool. The two main categories of risks are those created by the nature of the tool (e.g. read-only tools vs. read-write tools) and those created by bugs. pt-kill kills queries if you use the L<"--kill"> option, so it can disrupt your database's users, of course. You should test with the <"--print"> option, which is safe, if you're unsure what the tool will do. At the time of this release, we know of no bugs that could cause serious harm to users. The authoritative source for updated information is always the online issue tracking system. Issues that affect this tool will be marked as such. You can see a list of such issues at the following URL: L. See also L<"BUGS"> for more information on filing bugs and getting help. =head1 DESCRIPTION pt-kill captures queries from SHOW PROCESSLIST, filters them, and then either kills or prints them. This is also known as a "slow query sniper" in some circles. The idea is to watch for queries that might be consuming too many resources, and kill them. For brevity, we talk about killing queries, but they may just be printed (or some other future action) depending on what options are given. Normally pt-kill connects to MySQL to get queries from SHOW PROCESSLIST. Alternatively, it can read SHOW PROCESSLIST output from files. In this case, pt-kill does not connect to MySQL and L<"--kill"> has no effect. You should use L<"--print"> instead when reading files. The ability to read a file with L<"--test-matching"> allows you to capture SHOW PROCESSLIST and test it later with pt-kill to make sure that your matches kill the proper queries. There are a lot of special rules to follow, such as "don't kill replication threads," so be careful not to kill something important! Two important options to know are L<"--busy-time"> and L<"--victims">. First, whereas most match/filter options match their corresponding value from SHOW PROCESSLIST (e.g. L<"--match-command"> matches a query's Command value), the Time value is matched by L<"--busy-time">. See also L<"--interval">. Second, L<"--victims"> controls which matching queries from each class are killed. By default, the matching query with the highest Time value is killed (the oldest query). See the next section, L<"GROUP, MATCH AND KILL">, for more details. Usually you need to specify at least one C<--match> option, else no queries will match. Or, you can specify L<"--match-all"> to match all queries that aren't ignored by an C<--ignore> option. =head1 GROUP, MATCH AND KILL Queries pass through several steps to determine which exactly will be killed (or printed--whatever action is specified). Understanding these steps will help you match precisely the queries you want. The first step is grouping queries into classes. The L<"--group-by"> option controls grouping. By default, this option has no value so all queries are grouped into one default class. All types of matching and filtering (the next step) are applied per-class. Therefore, you may need to group queries in order to match/filter some classes but not others. The second step is matching. Matching implies filtering since if a query doesn't match some criteria, it is removed from its class. Matching happens for each class. First, queries are filtered from their class by the various C options like L<"--match-user">. Then, entire classes are filtered by the various C options like L<"--query-count">. The third step is victim selection, that is, which matching queries in each class to kill. This is controlled by the L<"--victims"> option. Although many queries in a class may match, you may only want to kill the oldest query, or all queries, etc. The forth and final step is to take some action on all matching queries from all classes. The C options specify which actions will be taken. At this step, there are no more classes, just a single list of queries to kill, print, etc. =head1 OUTPUT If only L<"--kill"> is given, then there is no output. If only L<"--print"> is given, then a timestamped KILL statement if printed for every query that would have been killed, like: # 2009-07-15T15:04:01 KILL 8 (Query 42 sec) SELECT * FROM huge_table The line shows a timestamp, the query's Id (8), its Time (42 sec) and its Info (usually the query SQL). If both L<"--kill"> and L<"--print"> are given, then matching queries are killed and a line for each like the one above is printed. Any command executed by L<"--execute-command"> is responsible for its own output and logging. After being executed, pt-kill has no control or interaction with the command. =head1 OPTIONS Specify at least one of L<"--kill">, L<"--kill-query">, L<"--print">, L<"--execute-command"> or L<"--stop">. L<"--any-busy-time"> and L<"--each-busy-time"> are mutually exclusive. L<"--kill"> and L<"--kill-query"> are mutually exclusive. L<"--daemonize"> and L<"--test-matching"> are mutually exclusive. =over =item --ask-pass Prompt for a password when connecting to MySQL. =item --charset short form: -A; type: string Default character set. If the value is utf8, sets Perl's binmode on STDOUT to utf8, passes the mysql_enable_utf8 option to DBD::mysql, and runs SET NAMES UTF8 after connecting to MySQL. Any other value sets binmode on STDOUT without the utf8 layer, and runs SET NAMES after connecting to MySQL. =item --config type: Array Read this comma-separated list of config files; if specified, this must be the first option on the command line. =item --daemonize Fork to the background and detach from the shell. POSIX operating systems only. =item --defaults-file short form: -F; type: string Only read mysql options from the given file. You must give an absolute pathname. =item --filter type: string Discard events for which this Perl code doesn't return true. This option is a string of Perl code or a file containing Perl code that gets compiled into a subroutine with one argument: $event. This is a hashref. If the given value is a readable file, then pt-kill reads the entire file and uses its contents as the code. The file should not contain a shebang (#!/usr/bin/perl) line. If the code returns true, the chain of callbacks continues; otherwise it ends. The code is the last statement in the subroutine other than C. The subroutine template is: sub { $event = shift; filter && return $event; } Filters given on the command line are wrapped inside parentheses like like C<( filter )>. For complex, multi-line filters, you must put the code inside a file so it will not be wrapped inside parentheses. Either way, the filter must produce syntactically valid code given the template. For example, an if-else branch given on the command line would not be valid: --filter 'if () { } else { }' # WRONG Since it's given on the command line, the if-else branch would be wrapped inside parentheses which is not syntactically valid. So to accomplish something more complex like this would require putting the code in a file, for example filter.txt: my $event_ok; if (...) { $event_ok=1; } else { $event_ok=0; } $event_ok Then specify C<--filter filter.txt> to read the code from filter.txt. If the filter code won't compile, pt-kill will die with an error. If the filter code does compile, an error may still occur at runtime if the code tries to do something wrong (like pattern match an undefined value). pt-kill does not provide any safeguards so code carefully! It is permissible for the code to have side effects (to alter C<$event>). =item --group-by type: string Apply matches to each class of queries grouped by this SHOW PROCESSLIST column. In addition to the basic columns of SHOW PROCESSLIST (user, host, command, state, etc.), queries can be matched by C which abstracts the SQL query in the C column. By default, queries are not grouped, so matches and actions apply to all queries. Grouping allows matches and actions to apply to classes of similar queries, if any queries in the class match. For example, detecting cache stampedes (see C under L<"--victims"> for an explanation of that term) requires that queries are grouped by the C attribute. This creates classes of identical queries (stripped of comments). So queries C<"SELECT c FROM t WHERE id=1"> and C<"SELECT c FROM t WHERE id=1"> are grouped into the same class, but query c<"SELECT c FROM t WHERE id=3"> is not identical to the first two queries so it is grouped into another class. Then when L<"--victims"> C is specified, all but the oldest query in each class is killed for each class of queries that matches the match criteria. =item --help Show help and exit. =item --host short form: -h; type: string; default: localhost Connect to host. =item --interval type: time How often to check for queries to kill. If L<"--busy-time"> is not given, then the default interval is 30 seconds. Else the default is half as often as L<"--busy-time">. If both L<"--interval"> and L<"--busy-time"> are given, then the explicit L<"--interval"> value is used. See also L<"--run-time">. =item --log type: string Print all output to this file when daemonized. =item --password short form: -p; type: string Password to use when connecting. =item --pid type: string Create the given PID file when daemonized. The file contains the process ID of the daemonized instance. The PID file is removed when the daemonized instance exits. The program checks for the existence of the PID file when starting; if it exists and the process with the matching PID exists, the program exits. =item --port short form: -P; type: int Port number to use for connection. =item --run-time type: time How long to run before exiting. By default pt-kill runs forever, or until its process is killed or stopped by the creation of a L<"--sentinel"> file. If this option is specified, pt-kill runs for the specified amount of time and sleeps L<"--interval"> seconds between each check of the PROCESSLIST. =item --sentinel type: string; default: /tmp/pt-kill-sentinel Exit if this file exists. The presence of the file specified by L<"--sentinel"> will cause all running instances of pt-kill to exit. You might find this handy to stop cron jobs gracefully if necessary. See also L<"--stop">. =item --set-vars type: string; default: wait_timeout=10000 Set these MySQL variables. Immediately after connecting to MySQL, this string will be appended to SET and executed. =item --socket short form: -S; type: string Socket file to use for connection. =item --stop Stop running instances by creating the L<"--sentinel"> file. Causes pt-kill to create the sentinel file specified by L<"--sentinel"> and exit. This should have the effect of stopping all running instances which are watching the same sentinel file. =item --[no]strip-comments default: yes Remove SQL comments from queries in the Info column of the PROCESSLIST. =item --user short form: -u; type: string User for login if not current user. =item --version Show version and exit. =item --victims type: string; default: oldest Which of the matching queries in each class will be killed. After classes have been matched/filtered, this option specifies which of the matching queries in each class will be killed (or printed, etc.). The following values are possible: =over =item oldest Only kill the single oldest query. This is to prevent killing queries that aren't really long-running, they're just long-waiting. This sorts matching queries by Time and kills the one with the highest Time value. =item all Kill all queries in the class. =item all-but-oldest Kill all but the oldest query. This is the inverse of the C value. This value can be used to prevent "cache stampedes", the condition where several identical queries are executed and create a backlog while the first query attempts to finish. Since all queries are identical, all but the first query are killed so that it can complete and populate the cache. =back =item --wait-after-kill type: time Wait after killing a query, before looking for more to kill. The purpose of this is to give blocked queries a chance to execute, so we don't kill a query that's blocking a bunch of others, and then kill the others immediately afterwards. =item --wait-before-kill type: time Wait before killing a query. The purpose of this is to give L<"--execute-command"> a chance to see the matching query and gather other MySQL or system information before it's killed. =back =head2 QUERY MATCHES These options filter queries from their classes. If a query does not match, it is removed from its class. The C<--ignore> options take precedence. The matches for command, db, host, etc. correspond to the columns returned by SHOW PROCESSLIST: Command, db, Host, etc. All pattern matches are case-sensitive by default, but they can be made case-insensitive by specifying a regex pattern like C<(?i-xsm:select)>. See also L<"GROUP, MATCH AND KILL">. =over =item --busy-time type: time; group: Query Matches Match queries that have been running for longer than this time. The queries must be in Command=Query status. This matches a query's Time value as reported by SHOW PROCESSLIST. =item --idle-time type: time; group: Query Matches Match queries that have been idle/sleeping for longer than this time. The queries must be in Command=Sleep status. This matches a query's Time value as reported by SHOW PROCESSLIST. =item --ignore-command type: string; group: Query Matches Ignore queries whose Command matches this Perl regex. See L<"--match-command">. =item --ignore-db type: string; group: Query Matches Ignore queries whose db (database) matches this Perl regex. See L<"--match-db">. =item --ignore-host type: string; group: Query Matches Ignore queries whose Host matches this Perl regex. See L<"--match-host">. =item --ignore-info type: string; group: Query Matches Ignore queries whose Info (query) matches this Perl regex. See L<"--match-info">. =item --[no]ignore-self default: yes; group: Query Matches Don't kill pt-kill's own connection. =item --ignore-state type: string; group: Query Matches; default: Locked Ignore queries whose State matches this Perl regex. The default is to keep threads from being killed if they are locked waiting for another thread. See L<"--match-state">. =item --ignore-user type: string; group: Query Matches Ignore queries whose user matches this Perl regex. See L<"--match-user">. =item --match-all group: Query Matches Match all queries that are not ignored. If no ignore options are specified, then every query matches (except replication threads, unless L<"--replication-threads"> is also specified). This option allows you to specify negative matches, i.e. "match every query I..." where the exceptions are defined by specifying various C<--ignore> options. This option is I the same as L<"--victims"> C. This option matches all queries within a class, whereas L<"--victims"> C specifies that all matching queries in a class (however they matched) will be killed. Normally, however, the two are used together because if, for example, you specify L<"--victims"> C, then although all queries may match, only the oldest will be killed. =item --match-command type: string; group: Query Matches Match only queries whose Command matches this Perl regex. Common Command values are: Query Sleep Binlog Dump Connect Delayed insert Execute Fetch Init DB Kill Prepare Processlist Quit Reset stmt Table Dump See L for a full list and description of Command values. =item --match-db type: string; group: Query Matches Match only queries whose db (database) matches this Perl regex. =item --match-host type: string; group: Query Matches Match only queries whose Host matches this Perl regex. The Host value often time includes the port like "host:port". =item --match-info type: string; group: Query Matches Match only queries whose Info (query) matches this Perl regex. The Info column of the processlist shows the query that is being executed or NULL if no query is being executed. =item --match-state type: string; group: Query Matches Match only queries whose State matches this Perl regex. Common State values are: Locked login copy to tmp table Copying to tmp table Copying to tmp table on disk Creating tmp table executing Reading from net Sending data Sorting for order Sorting result Table lock Updating See L for a full list and description of State values. =item --match-user type: string; group: Query Matches Match only queries whose User matches this Perl regex. =item --replication-threads group: Query Matches Allow matching and killing replication threads. By default, matches do not apply to replication threads; i.e. replication threads are completely ignored. Specifying this option allows matches to match (and potentially kill) replication threads on masters and slaves. =item --test-matching type: array; group: Query Matches Files with processlist snapshots to test matching options against. Since the matching options can be complex, you can save snapshots of processlist in files, then test matching options against queries in those files. This option disables L<"--run-time">, L<"--interval">, and L<"--[no]ignore-self">. =back =head2 CLASS MATCHES These matches apply to entire query classes. Classes are created by specifying the L<"--group-by"> option, else all queries are members of a single, default class. See also L<"GROUP, MATCH AND KILL">. =over =item --any-busy-time type: time; group: Class Matches Match query class if any query has been running for longer than this time. "Longer than" means that if you specify C<10>, for example, the class will only match if there's at least one query that has been running for greater than 10 seconds. See L<"--each-busy-time"> for more details. =item --each-busy-time type: time; group: Class Matches Match query class if each query has been running for longer than this time. "Longer than" means that if you specify C<10>, for example, the class will only match if each and every query has been running for greater than 10 seconds. See also L<"--any-busy-time"> (to match a class if ANY query has been running longer than the specified time) and L<"--busy-time">. =item --query-count type: int; group: Class Matches Match query class if it has at least this many queries. When queries are grouped into classes by specifying L<"--group-by">, this option causes matches to apply only to classes with at least this many queries. If L<"--group-by"> is not specified then this option causes matches to apply only if there are at least this many queries in the entire SHOW PROCESSLIST. =item --verbose short form: -v Print information to STDOUT about what is being done. =back =head2 ACTIONS These actions are taken for every matching query from all classes. The actions are taken in this order: L<"--print">, L<"--execute-command">, L<"--kill">/L<"--kill-query">. This order allows L<"--execute-command"> to see the output of L<"--print"> and the query before L<"--kill">/L<"--kill-query">. This may be helpful because pt-kill does not pass any information to L<"--execute-command">. See also L<"GROUP, MATCH AND KILL">. =over =item --execute-command type: string; group: Actions Execute this command when a query matches. After the command is executed, pt-kill has no control over it, so the command is responsible for its own info gathering, logging, interval, etc. The command is executed each time a query matches, so be careful that the command behaves well when multiple instances are ran. No information from pt-kill is passed to the command. See also L<"--wait-before-kill">. =item --kill group: Actions Kill the connection for matching queries. This option makes pt-kill kill the connections (a.k.a. processes, threads) that have matching queries. Use L<"--kill-query"> if you only want to kill individual queries and not their connections. Unless L<"--print"> is also given, no other information is printed that shows that pt-kill matched and killed a query. See also L<"--wait-before-kill"> and L<"--wait-after-kill">. =item --kill-query group: Actions Kill matching queries. This option makes pt-kill kill matching queries. This requires MySQL 5.0 or newer. Unlike L<"--kill"> which kills the connection for matching queries, this option only kills the query, not its connection. =item --print group: Actions Print a KILL statement for matching queries; does not actually kill queries. If you just want to see which queries match and would be killed without actually killing them, specify L<"--print">. To both kill and print matching queries, specify both L<"--kill"> and L<"--print">. =back =head1 DSN OPTIONS These DSN options are used to create a DSN. Each option is given like C. The options are case-sensitive, so P and p are not the same option. There cannot be whitespace before or after the C<=> and if the value contains whitespace it must be quoted. DSN options are comma-separated. See the L manpage for full details. =over =item * A dsn: charset; copy: yes Default character set. =item * D dsn: database; copy: yes Default database. =item * F dsn: mysql_read_default_file; copy: yes Only read default options from the given file =item * h dsn: host; copy: yes Connect to host. =item * p dsn: password; copy: yes Password to use when connecting. =item * P dsn: port; copy: yes Port number to use for connection. =item * S dsn: mysql_socket; copy: yes Socket file to use for connection. =item * u dsn: user; copy: yes User for login if not current user. =back =head1 ENVIRONMENT The environment variable C enables verbose debugging output to STDERR. To enable debugging and capture all output to a file, run the tool like: PTDEBUG=1 pt-kill ... > FILE 2>&1 Be careful: debugging output is voluminous and can generate several megabytes of output. =head1 SYSTEM REQUIREMENTS You need Perl, DBI, DBD::mysql, and some core packages that ought to be installed in any reasonably new version of Perl. =head1 BUGS For a list of known bugs, see L. Please report bugs at L. Include the following information in your bug report: =over =item * Complete command-line used to run the tool =item * Tool L<"--version"> =item * MySQL version of all servers involved =item * Output from the tool including STDERR =item * Input files (log/dump/config files, etc.) =back If possible, include debugging output by running the tool with C; see L<"ENVIRONMENT">. =head1 DOWNLOADING Visit L to download the latest release of Percona Toolkit. Or, get the latest release from the command line: wget percona.com/get/percona-toolkit.tar.gz wget percona.com/get/percona-toolkit.rpm wget percona.com/get/percona-toolkit.deb You can also get individual tools from the latest release: wget percona.com/get/TOOL Replace C with the name of any tool. =head1 AUTHORS Baron Schwartz and Daniel Nichter =head1 ABOUT PERCONA TOOLKIT This tool is part of Percona Toolkit, a collection of advanced command-line tools developed by Percona for MySQL support and consulting. Percona Toolkit was forked from two projects in June, 2011: Maatkit and Aspersa. Those projects were created by Baron Schwartz and developed primarily by him and Daniel Nichter, both of whom are employed by Percona. Visit L for more software developed by Percona. =head1 COPYRIGHT, LICENSE, AND WARRANTY This program is copyright 2009-2011 Baron Schwartz, 2011-2012 Percona Inc. Feedback and improvements are welcome. THIS PROGRAM IS PROVIDED "AS IS" AND WITHOUT ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, WITHOUT LIMITATION, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE. This program is free software; you can redistribute it and/or modify it under the terms of the GNU General Public License as published by the Free Software Foundation, version 2; OR the Perl Artistic License. On UNIX and similar systems, you can issue `man perlgpl' or `man perlartistic' to read these licenses. You should have received a copy of the GNU General Public License along with this program; if not, write to the Free Software Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA. =head1 VERSION pt-kill 2.1.1 =cut