Merge lp:~gryp/percona-toolkit/ptosc-lagwaiter-ptheartbeat into lp:percona-toolkit/2.2

[Kenny Gryp]

Hi,

This merge request adds 2 features to 2 tools.

First of all, I have added --plugin to pt-table-checksum.
The necessary tests (similar to pt-osc) are added.

Then I have added a new --plugin call for pt-osc named 'get_slave_lag'.

Reason for this development is that a customer (issue #28725) wants to use the official percona toolkit packaged tools while using pt-heartbeat or tungsten to check slavelag. Both pt-osc and pt-tc need to have pt-heartbeat functionality which can be achieved with the --plugin and by overriding the $get_lag subroutine.

I have build an example plugin for both pt-osc and pt-tc in the following github branch: https://github.com/grypyrg/percona-toolkit-plugin-heartbeat/blob/master/pt-plugin-heartbeat.pm

I tried to do it as compatible as possible.
Implementing pt-heartbeat in the tools itself looked a bit more overkill and I think the plugin is a simple codechange which gives a lot of flexibility. Of course the flexibility means that this is not for novice users.

The changes that Daniel requested were implemented.

Kenny

[Daniel Nichter (daniel-nichter)]

This will make a good addition to the tool. Thanks for testing it. Before we merge, let's make the following changes:

1. Rename plugin hook to get-slave-lag

2. Rewrite code like:

my $get_lag;
if ( $plugin && $plugin->can('get-slave-lag') ) {
    $get_lag = $plugin->override_slavelag_check(oktorun => \$oktorun);
}
else {
    $get_lag = sub {...}
}

3. Break the code comments "# The plugin is able"... on/before 80 cols.

review: Needs Resubmitting

Unmerged revisions

=== modified file 'bin/pt-online-schema-change'
--- bin/pt-online-schema-change	2014-02-20 08:10:16 +0000
+++ bin/pt-online-schema-change	2014-04-23 11:30:39 +0000
@@ -8105,21 +8105,29 @@
          return;
       };
 
-      my $get_lag = sub {
-         my ($cxn) = @_;
-         my $dbh = $cxn->dbh();
-         if ( !$dbh || !$dbh->ping() ) {
-            eval { $dbh = $cxn->connect() };  # connect or die trying
-            if ( $EVAL_ERROR ) {
-               $oktorun = 0;  # flag for cleanup tasks
-               chomp $EVAL_ERROR;
-               die "Lost connection to replica " . $cxn->name()
-                  . " while attempting to get its lag ($EVAL_ERROR)\n";
+      my $get_lag;
+      # The plugin is able to override the slavelag check so tools like 
+      # pt-heartbeat or other replicators (Tungsten...) can be used to 
+      # measure replication lag
+      if ( $plugin && $plugin->can('get_slave_lag') ) {
+         $get_lag = $plugin->get_slave_lag(oktorun => \$oktorun);
+      } else {      
+         $get_lag = sub {
+            my ($cxn) = @_;
+            my $dbh = $cxn->dbh();
+            if ( !$dbh || !$dbh->ping() ) {
+               eval { $dbh = $cxn->connect() };  # connect or die trying
+               if ( $EVAL_ERROR ) {
+                  $oktorun = 0;  # flag for cleanup tasks
+                  chomp $EVAL_ERROR;
+                  die "Lost connection to replica " . $cxn->name()
+                     . " while attempting to get its lag ($EVAL_ERROR)\n";
+               }
             }
-         }
-         return $ms->get_slave_lag($dbh);
-      };
-
+            return $ms->get_slave_lag($dbh);
+         };
+      }
+      
       $replica_lag = new ReplicaLagWaiter(
          slaves   => $slave_lag_cxns,
          max_lag  => $o->get('max-lag'),
@@ -11345,6 +11353,7 @@
    after_drop_old_table
    before_drop_triggers
    before_exit
+   get_slave_lag
 
 Each hook is passed different arguments.  To see which arguments are passed
 to a hook, search for the hook's name in the tool's source code, like:
 
=== modified file 'bin/pt-table-checksum'
--- bin/pt-table-checksum	2014-02-20 08:10:16 +0000
+++ bin/pt-table-checksum	2014-04-23 11:30:39 +0000
@@ -9223,6 +9223,30 @@
    my $slaves = [];    # all slaves (that we can find)
    my $slave_lag_cxns; # slaves whose lag we'll check
 
+   # ########################################################################
+   # Create --plugin.
+   # ########################################################################   
+   my $plugin;
+   if ( my $file = $o->get('plugin') ) {
+      die "--plugin file $file does not exist\n" unless -f $file;
+      eval {
+         require $file;
+      };
+      die "Error loading --plugin $file: $EVAL_ERROR" if $EVAL_ERROR;
+      eval {
+         $plugin = pt_table_checksum_plugin->new(
+            master_cxn  => $master_cxn,
+            explain     => $o->get('explain'),
+            quiet       => $o->get('quiet'),
+            resume      => $o->get('resume'),
+            Quoter      => $q,
+            TableParser => $tp,
+         );
+      };
+      die "Error creating --plugin: $EVAL_ERROR" if $EVAL_ERROR;
+      print "Created plugin from $file.\n";
+   }
+
    my $replica_lag;    # ReplicaLagWaiter object
    my $replica_lag_pr; # Progress for ReplicaLagWaiter
    my $sys_load;       # MySQLStatusWaiter object
@@ -9447,6 +9471,11 @@
       # #####################################################################
       if ( $o->get('replicate-check') && $o->get('replicate-check-only') ) {
          PTDEBUG && _d('Will --replicate-check and exit');
+      
+         # --plugin hook
+         if ( $plugin && $plugin->can('before_replicate_check') ) {
+            $plugin->before_replicate_check();
+         }
 
          foreach my $slave ( @$slaves ) {
             my $diffs = $rc->find_replication_differences(
@@ -9467,6 +9496,11 @@
             }
          }
 
+         # --plugin hook
+         if ( $plugin && $plugin->can('after_replicate_check') ) {
+            $plugin->after_replicate_check();
+         }
+
          PTDEBUG && _d('Exit status', $exit_status, 'oktorun', $oktorun);
          return $exit_status;
       }
@@ -9545,23 +9579,31 @@
          return;
       };
 
-      my $get_lag = sub {
-         my ($cxn) = @_;
-         my $dbh = $cxn->dbh();
-         if ( !$dbh || !$dbh->ping() ) {
-            PTDEBUG && _d('Lost connection to slave', $cxn->name(),
-               'while waiting for slave lag');
-            eval { $dbh = $cxn->connect() };  # connect or die trying
-            if ( $EVAL_ERROR ) {
-               $oktorun = 0;  # Fatal error
-               chomp $EVAL_ERROR;
-               die "Lost connection to replica " . $cxn->name()
-                  . " while attempting to get its lag ($EVAL_ERROR)";
+      my $get_lag;
+      # The plugin is able to override the slavelag check so tools like 
+      # pt-heartbeat or other replicators (Tungsten...) can be used to 
+      # measure replication lag
+      if ( $plugin && $plugin->can('get_slave_lag') ) {
+         $get_lag = $plugin->get_slave_lag(oktorun => \$oktorun);
+      } else {
+         $get_lag = sub {
+            my ($cxn) = @_;
+            my $dbh = $cxn->dbh();
+            if ( !$dbh || !$dbh->ping() ) {
+               PTDEBUG && _d('Lost connection to slave', $cxn->name(),
+                  'while waiting for slave lag');
+               eval { $dbh = $cxn->connect() };  # connect or die trying
+               if ( $EVAL_ERROR ) {
+                  $oktorun = 0;  # Fatal error
+                  chomp $EVAL_ERROR;
+                  die "Lost connection to replica " . $cxn->name()
+                     . " while attempting to get its lag ($EVAL_ERROR)";
+               }
             }
-         }
-         return $ms->get_slave_lag($dbh);
-      };
-
+            return $ms->get_slave_lag($dbh);
+         };
+      }
+      
       $replica_lag = new ReplicaLagWaiter(
          slaves   => $slave_lag_cxns,
          max_lag  => $o->get('max-lag'),
@@ -10169,6 +10211,19 @@
    };
 
    # ########################################################################
+   # Init the --plugin.
+   # ########################################################################   
+
+   # --plugin hook
+   if ( $plugin && $plugin->can('init') ) {
+      $plugin->init(
+         slaves         => $slaves,
+         slave_lag_cxns => $slave_lag_cxns,
+         repl_table     => $repl_table,
+      );
+   }
+
+   # ########################################################################
    # Checksum each table.
    # ########################################################################
 
@@ -10272,6 +10327,12 @@
                              @$all_cols;
             $tbl->{checksum_cols} = \@cols;
 
+            # --plugin hook
+            if ( $plugin && $plugin->can('before_checksum_table') ) {
+               $plugin->before_checksum_table(
+                  tbl   => $tbl);
+            }
+
             # Finally, checksum the table.
             # The "1 while" loop is necessary because we're executing REPLACE
             # statements which don't return rows and NibbleIterator only
@@ -10280,6 +10341,11 @@
             # from the done callback, uses this start time.
             $tbl->{checksum_results}->{start_time} = time;
             1 while $nibble_iter->next();
+            
+            # --plugin hook
+            if ( $plugin && $plugin->can('after_checksum_table') ) {
+               $plugin->after_checksum_table();
+            }
          }
       };
       if ( $EVAL_ERROR ) {
@@ -12054,6 +12120,18 @@
 tool will overwrite the PID file with the current PID.  The PID file is
 removed automatically when the tool exits.
 
+=item --plugin
+
+type: string
+
+Perl module file that defines a C<pt_table_checksum_plugin> class.
+A plugin allows you to write a Perl module that can hook into many parts
+of pt-table-checksum.  This requires a good knowledge of Perl and
+Percona Toolkit conventions, which are beyond this scope of this
+documentation.  Please contact Percona if you have questions or need help.
+
+See L<"PLUGIN"> for more information.
+
 =item --port
 
 short form: -P; type: int; group: Connection
@@ -12402,6 +12480,39 @@
 
 =back
 
+=head1 PLUGIN
+
+The file specified by L<"--plugin"> must define a class (i.e. a package)
+called C<pt_table_checksum_plugin> with a C<new()> subroutine.
+The tool will create an instance of this class and call any hooks that
+it defines.  No hooks are required, but a plugin isn't very useful without
+them.
+
+These hooks, in this order, are called if defined:
+
+   init
+   before_replicate_check
+   after_replicate_check
+   get_slave_lag
+   before_checksum_table
+   after_checksum_table
+
+Each hook is passed different arguments.  To see which arguments are passed
+to a hook, search for the hook's name in the tool's source code, like:
+
+   # --plugin hook
+   if ( $plugin && $plugin->can('init') ) {
+      $plugin->init(
+         slaves         => $slaves,
+         slave_lag_cxns => $slave_lag_cxns,
+         repl_table     => $repl_table,
+      );
+   }
+
+The comment C<# --plugin hook> precedes every hook call.
+
+Please contact Percona if you have questions or need help.
+
 =head1 DSN OPTIONS
 
 These DSN options are used to create a DSN.  Each option is given like
 
=== modified file 't/pt-online-schema-change/plugin.t'
--- t/pt-online-schema-change/plugin.t	2013-03-01 00:27:39 +0000
+++ t/pt-online-schema-change/plugin.t	2014-04-23 11:30:39 +0000
@@ -56,6 +56,7 @@
 is_deeply(
    \@called,
    [
+      'PLUGIN get_slave_lag',
       'PLUGIN init',
       'PLUGIN before_create_new_table',
       'PLUGIN after_create_new_table',
 
=== modified file 't/pt-online-schema-change/samples/plugins/all_hooks.pm'
--- t/pt-online-schema-change/samples/plugins/all_hooks.pm	2013-03-01 00:27:39 +0000
+++ t/pt-online-schema-change/samples/plugins/all_hooks.pm	2014-04-23 11:30:39 +0000
@@ -96,4 +96,11 @@
    print "PLUGIN before_exit\n";
 }
 
+sub get_slave_lag {
+   my ($self, %args) = @_;
+   print "PLUGIN get_slave_lag\n";
+
+   return sub { return 0; };
+}
+
 1;
 
=== added file 't/pt-table-checksum/plugin.t'
--- t/pt-table-checksum/plugin.t	1970-01-01 00:00:00 +0000
+++ t/pt-table-checksum/plugin.t	2014-04-23 11:30:39 +0000
@@ -0,0 +1,98 @@
+#!/usr/bin/env perl
+
+BEGIN {
+   die "The PERCONA_TOOLKIT_BRANCH environment variable is not set.\n"
+      unless $ENV{PERCONA_TOOLKIT_BRANCH} && -d $ENV{PERCONA_TOOLKIT_BRANCH};
+   unshift @INC, "$ENV{PERCONA_TOOLKIT_BRANCH}/lib";
+};
+
+use strict;
+use warnings FATAL => 'all';
+use English qw(-no_match_vars);
+use Test::More;
+
+use PerconaTest;
+use Sandbox;
+require "$trunk/bin/pt-table-checksum";
+
+use Data::Dumper;
+$Data::Dumper::Indent    = 1;
+$Data::Dumper::Sortkeys  = 1;
+$Data::Dumper::Quotekeys = 0;
+
+my $dp         = new DSNParser(opts=>$dsn_opts);
+my $sb         = new Sandbox(basedir => '/tmp', DSNParser => $dp);
+my $master_dbh = $sb->get_dbh_for('master');
+
+if ( !$master_dbh ) {
+   plan skip_all => 'Cannot connect to sandbox master';
+}
+
+my $output;
+my $master_dsn = $sb->dsn_for('master');
+my $sample     = "t/pt-table-checksum/samples";
+my $plugin     = "$trunk/$sample/plugins";
+my $exit;
+my $rows;
+
+$master_dbh->prepare("drop database if exists percona")->execute();
+$master_dbh->prepare("create database percona")->execute();
+$master_dbh->prepare("create table if not exists percona.t ( a int primary key);")->execute();
+$master_dbh->prepare("insert into percona.t values (0),(1),(2),(3),(4),(5),(6),(7),(8),(9)")->execute();
+$master_dbh->prepare("analyze table percona.t;")->execute();
+
+# #############################################################################
+# all_hooks.pm
+# #############################################################################
+
+($output) = full_output(
+   sub { pt_table_checksum::main(
+      "$master_dsn",
+      '--databases', 'percona',
+      '--plugin', "$plugin/all_hooks.pm",
+   )},
+   stderr => 1,
+);
+
+my @called = $output =~ m/^PLUGIN \S+$/gm;
+
+is_deeply(
+   \@called,
+   [
+      'PLUGIN get_slave_lag',
+      'PLUGIN init',
+      'PLUGIN before_checksum_table',
+      'PLUGIN after_checksum_table',
+   ],
+   "Called all plugins on basic run"
+) or diag(Dumper($output));
+
+
+($output) = full_output(
+   sub { pt_table_checksum::main(
+      "$master_dsn",
+      '--replicate-check', '--replicate-check-only',
+      '--databases', 'percona',
+      '--plugin', "$plugin/all_hooks.pm",
+   )},
+   stderr => 1,
+);
+
+@called = $output =~ m/^PLUGIN \S+$/gm;
+
+is_deeply(
+   \@called,
+   [
+      'PLUGIN before_replicate_check',
+      'PLUGIN after_replicate_check',
+   ],
+   "Called all plugins on replicate-check run"
+) or diag(Dumper($output));
+
+
+# #############################################################################
+# Done.
+# #############################################################################
+$sb->wipe_clean($master_dbh);
+ok($sb->ok(), "Sandbox servers") or BAIL_OUT(__FILE__ . " broke the sandbox");
+done_testing;
 
=== added directory 't/pt-table-checksum/samples/plugins'
=== added file 't/pt-table-checksum/samples/plugins/all_hooks.pm'
--- t/pt-table-checksum/samples/plugins/all_hooks.pm	1970-01-01 00:00:00 +0000
+++ t/pt-table-checksum/samples/plugins/all_hooks.pm	2014-04-23 11:30:39 +0000
@@ -0,0 +1,45 @@
+package pt_table_checksum_plugin;
+
+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 };
+   return bless $self, $class;
+}
+
+sub init {
+   my ($self, %args) = @_;
+   print "PLUGIN init\n";
+}
+ 
+sub before_replicate_check {
+   my ($self, %args) = @_;
+   print "PLUGIN before_replicate_check\n";
+}
+ 
+sub after_replicate_check {
+   my ($self, %args) = @_;
+   print "PLUGIN after_replicate_check\n";
+}
+ 
+sub get_slave_lag {
+   my ($self, %args) = @_;
+   print "PLUGIN get_slave_lag\n";
+   return sub { return 0; };
+}
+ 
+sub before_checksum_table {
+   my ($self, %args) = @_;
+   print "PLUGIN before_checksum_table\n";
+}
+
+sub after_checksum_table {
+   my ($self, %args) = @_;
+   print "PLUGIN after_checksum_table\n";
+}
+
+1;