Changed replica_parallel_workers in configuration for 8.0 and 8.4 to o, because wiat_for_replicas is buggy for multi-threaded slaves

from t/pt-table-checksum/pt-1616.t

lib/Sandbox.pm

@@ -465,6 +465,7 @@ sub is_cluster_node {
 sub can_load_data {
     my ($self, $server) = @_;
     my $output = $self->use($server, q{-e "SELECT * FROM percona_test.load_data"});
+    #die(0 =~ /1/);
     return ($output || '') =~ /1/;
 }
 

sandbox/servers/8.0/my.sandbox.cnf

@@ -40,3 +40,6 @@ binlog_format              = STATEMENT
 #performance-schema-instrument='wait/lock/metadata/sql/mdl=ON'
 #performance-schema-instrument='transaction=ON'
 secure-file-priv           =
+
+# wait_for_replica buggy on multi-threaded replica
+slave-parallel-workers=0

sandbox/servers/8.4/my.sandbox.cnf

@@ -40,3 +40,6 @@ binlog_format              = STATEMENT
 #performance-schema-instrument='wait/lock/metadata/sql/mdl=ON'
 #performance-schema-instrument='transaction=ON'
 secure-file-priv           =
+
+# wait_for_replica buggy on multi-threaded replica
+replica-parallel-workers=0

t/pt-archiver/replica_lag.t

@@ -251,6 +251,7 @@ like(
 # Done.
 # #############################################################################
 $replica1_dbh->do("STOP ${replica_name}");
+$replica1_dbh->do("CHANGE ${source_change} TO ${source_name}_DELAY=0");
 $source_dbh->do("RESET ${source_reset}");
 $replica1_dbh->do("RESET ${replica_name}");
 $replica1_dbh->do("START ${replica_name}");

t/pt-online-schema-change/pt-1717-resume.t

@@ -398,6 +398,7 @@ $replica_dbh2 = $sb->get_dbh_for('replica2');
 diag("Setting replica delay to 0 seconds");
 $replica_dbh1->do("STOP ${replica_name}");
 $replica_dbh2->do("STOP ${replica_name}");
+$replica_dbh1->do("CHANGE ${source_change} TO ${source_name}_DELAY=0");
 $source_dbh->do("RESET ${source_reset}");
 $replica_dbh1->do("RESET ${source_reset}");
 $replica_dbh1->do("RESET ${replica_name}");

t/pt-online-schema-change/pt-2168-cancel.t

@@ -162,6 +162,7 @@ $replica_dbh2 = $sb->get_dbh_for('replica2');
 diag("Setting replica delay to 0 seconds");
 $replica_dbh1->do("STOP ${replica_name}");
 $replica_dbh2->do("STOP ${replica_name}");
+$replica_dbh1->do("CHANGE ${source_change} TO ${source_name}_DELAY=0");
 $source_dbh->do("RESET ${source_reset}");
 $replica_dbh1->do("RESET ${replica_name}");
 $replica_dbh2->do("RESET ${replica_name}");

t/pt-online-schema-change/pt-2168.t

@@ -328,6 +328,7 @@ $replica_dbh2 = $sb->get_dbh_for('replica2');
 diag("Setting replica delay to 0 seconds");
 $replica_dbh1->do("STOP ${replica_name}");
 $replica_dbh2->do("STOP ${replica_name}");
+$replica_dbh1->do("CHANGE ${source_change} TO ${source_name}_DELAY=0");
 $source_dbh->do("RESET ${source_reset}");
 $replica_dbh1->do("RESET ${replica_name}");
 $replica_dbh2->do("RESET ${replica_name}");

t/pt-online-schema-change/pt-2241.t

@@ -98,6 +98,8 @@ unlike(
 diag("Setting replica delay to 0 seconds");
 $replica1_dbh->do("STOP ${replica_name}");
 $replica2_dbh->do("STOP ${replica_name}");
+$replica1_dbh->do("CHANGE ${source_change} TO ${source_name}_DELAY=0");
+$replica2_dbh->do("CHANGE ${source_change} TO ${source_name}_DELAY=0");
 $source_dbh->do("RESET ${source_reset}");
 $replica1_dbh->do("RESET ${replica_name}");
 $replica1_dbh->do("START ${replica_name}");

t/pt-online-schema-change/slave_lag.t

@@ -354,6 +354,7 @@ $sb->do_as_root("source", q/FLUSH TABLES/);
 
 diag("Setting replica delay to 0 seconds");
 $replica_dbh->do("STOP ${replica_name}");
+$replica_dbh->do("CHANGE ${source_change} TO ${source_name}_DELAY=0");
 $source_dbh->do("RESET ${source_reset}");
 $replica_dbh->do("RESET ${replica_name}");
 $replica_dbh->do("START ${replica_name}");

t/pt-table-checksum/chunk_size.t

@@ -104,7 +104,7 @@ unlike(
    $output,
    qr/Cannot checksum table/,
    "Very small --chunk-time doesn't cause zero --chunk-size"
-);
+) or diag($output);
 }
 # #############################################################################
 # Bug 921700: pt-table-checksum doesn't add --where to chunk-oversize test

t/pt-table-checksum/pt-1616.t

@@ -12,7 +12,7 @@ use English qw(-no_match_vars);
 use Test::More;
 
 #if ( !$ENV{SLOW_TESTS} ) { 
-#   plan skip_all => "pt-table-checksum/replication_filters.t is one of the top slowest files; set SLOW_TESTS=1 to enable it.";
+#   plan skip_all => "pt-table-checksum/pt-1616.t is one of the top slowest files; set SLOW_TESTS=1 to enable it.";
 #}
 
 use PerconaTest;
@@ -23,6 +23,8 @@ require "$trunk/bin/pt-table-checksum";
 my $dp  = new DSNParser(opts=>$dsn_opts);
 my $sb  = new Sandbox(basedir => '/tmp', DSNParser => $dp);
 my $dbh = $sb->get_dbh_for('source');
+my $replica1_dbh = $sb->get_dbh_for('replica1');
+my $replica2_dbh = $sb->get_dbh_for('replica2');
 
 if ( !$dbh ) {
     plan skip_all => 'Cannot connect to sandbox source';
@@ -104,7 +106,18 @@ unlike(
 # #############################################################################
 # Done.
 # #############################################################################
-$sb->wipe_clean($dbh);
+diag("Resetting replicas, because this test sporadically fails");
+$replica1_dbh->do("STOP ${replica_name}");
+$replica2_dbh->do("STOP ${replica_name}");
+$dbh->do("RESET ${source_reset}");
+$replica1_dbh->do("RESET ${replica_name}");
+$replica1_dbh->do("START ${replica_name}");
+$replica2_dbh->do("RESET ${replica_name}");
+$replica2_dbh->do("START ${replica_name}");
+
+diag("Replicas reset, syncing");
 $sb->wait_for_replicas();
+diag("Cleaning up");
+$sb->wipe_clean($dbh);
 ok($sb->ok(), "Sandbox servers") or BAIL_OUT(__FILE__ . " broke the sandbox");
 exit;

t/pt-table-checksum/resume.t

@@ -127,7 +127,7 @@ is_deeply(
       [qw( sakila city          )],
    ],
    "Checksum results for 1/4 of sakila singles"
-);
+) or diag(Dumper($row));
 
 $output = output(
    sub { pt_table_checksum::main(@args, qw(-d sakila --resume --chunk-size 10000)) },

t/pt-table-sync/char_chunking.t

@@ -49,7 +49,7 @@ like(
    $output,
    qr/#\s+0\s+4\s+0\s+0\s+Chunk\s+/,
    "Chunks char col"
-);
+) or diag($output);
 like(
    $output,
    qr/FORCE INDEX \(`c`\)/,

t/pt-table-sync/pt-1194.t

@@ -10,6 +10,7 @@ use strict;
 use warnings FATAL => 'all';
 use English qw(-no_match_vars);
 use Test::More;
+use POSIX ":sys_wait_h";
 
 use PerconaTest;
 use Sandbox;
@@ -38,26 +39,36 @@ else {
    plan tests => 3;
 }
 
+diag("OK\n");
+
 $sb->load_file('source', "t/pt-table-sync/samples/pt-1205.sql");
+diag("OK 2\n");
 $sb->wait_for_replicas();
 
+diag("OK 3\n");
+
 # Setting up tunnels
 my $pid1 = fork();
 
 if ( !$pid1 ) {
    setpgrp;
-   system('ncat -k -l localhost 3333 --sh-exec "ncat 127.0.0.1 12345"');
+   system('ncat -k -l localhost 33333 --sh-exec "ncat 127.0.0.1 12345"');
+diag("OK 4\n");
    exit;
 }
 
+diag("OK 5\n");
+
 my $pid2 = fork();
 
 if ( !$pid2 ) {
    setpgrp;
-   system('ncat -k -l localhost 3334 --sh-exec "ncat 127.0.0.1 12346"');
+   system('ncat -k -l localhost 33334 --sh-exec "ncat 127.0.0.1 12346"');
+diag("OK 6\n");
    exit;
 }
 
+diag("OK 7\n");
 my $o = new OptionParser();
 my $q = new Quoter();
 my $ms = new MasterSlave(
@@ -67,28 +78,35 @@ my $ms = new MasterSlave(
             );
 my $ss = $ms->get_replica_status($replica1_dbh);
 
+diag("OK 8\n");
 $replica1_dbh->do("STOP ${replica_name}");
-$replica1_dbh->do("CHANGE ${source_change} TO ${source_name}_PORT=3333, ${source_name}_LOG_POS=" . $ss->{"exec_${source_name}_log_pos"});
+$replica1_dbh->do("CHANGE ${source_change} TO ${source_name}_PORT=33333, ${source_name}_LOG_POS=" . $ss->{"exec_${source_name}_log_pos"});
 $replica1_dbh->do("START ${replica_name}");
 
-my $output = `$trunk/bin/pt-table-sync h=127.0.0.1,P=3334,u=msandbox,p=msandbox --database=test --table=t1 --sync-to-source --execute --verbose 2>&1`;
+diag("OK 9\n");
+my $output = `$trunk/bin/pt-table-sync h=127.0.0.1,P=33334,u=msandbox,p=msandbox --database=test --table=t1 --sync-to-source --execute --verbose 2>&1`;
 
+diag("OK 10\n");
 unlike(
    $output,
    qr/The replica is connected to \d+ but the source's port is/,
    'No error for redirected replica'
 ) or diag($output);
 
+diag("OK 11\n");
 kill -1, getpgrp($pid1);
 kill -1, getpgrp($pid2);
 
+diag("OK 12\n");
 $replica1_dbh->do("STOP ${replica_name}");
 $ss = $ms->get_replica_status($replica1_dbh);
 $replica1_dbh->do("CHANGE ${source_change} TO ${source_name}_PORT=12347, ${source_name}_LOG_POS=" . $ss->{"exec_${source_name}_log_pos"});
 $replica1_dbh->do("START ${replica_name} SQL_THREAD");
 
+diag("OK 13\n");
 $output = `$trunk/bin/pt-table-sync h=127.0.0.1,P=12346,u=msandbox,p=msandbox --database=test --table=t1 --sync-to-source --execute --verbose 2>&1`;
 
+diag("OK 14\n");
 like(
    $output,
    qr/The server specified as a source has no connected replicas/,
@@ -98,9 +116,12 @@ like(
 $replica1_dbh->do("STOP ${replica_name}");
 $replica1_dbh->do("CHANGE ${source_change} TO ${source_name}_PORT=12345, ${source_name}_LOG_POS=" . $ss->{"exec_${source_name}_log_pos"});
 $replica1_dbh->do("START ${replica_name}");
+diag("OK 15\n");
 # #############################################################################
 # Done.
 # #############################################################################
 $sb->wipe_clean($source_dbh);
+diag("OK 16\n");
 ok($sb->ok(), "Sandbox servers") or BAIL_OUT(__FILE__ . " broke the sandbox");
+diag("OK 17\n");
 exit;

t/pt-table-sync/safety_checks.t

@@ -146,9 +146,8 @@ like(
 
 $replica1_dbh->do("STOP ${replica_name}");
 $replica1_dbh->do("CHANGE ${source_change} TO ${source_name}_port=12345, ${source_name}_user='msandbox'");
-$replica1_dbh->do("START ${replica_name}");
-$replica1_dbh->do("STOP ${replica_name}");
-$replica1_dbh->do("CHANGE ${source_change} TO ${source_name}_port=12345, ${source_name}_user='msandbox'");
+$source_dbh->do("RESET ${source_reset}");
+$replica1_dbh->do("RESET ${replica_name}");
 $replica1_dbh->do("START ${replica_name}");
 
 # #############################################################################