MDEV-15792 Fix mtr to be able to wait for >1 exited mysqld

If a mtr test case has started two mysqld processes (replication tests),
then kills the first one and kills the second one before starting the
first (so at some point there are two mysqlds down), then the ./mtr
waiting process bricks and forgets to monitor the "expect" file of the
first mysqld, so it never gets started again, even when its contents is
changed to "restart".

A victim of this deficiency is at least galera.galera_gcache_recover.

The fix is to keep a list of all mysqlds we should wait to start, not
just one (the last one killed).

1 files changed, 103 insertions, 96 deletions

diff --git a/mysql-test/mysql-test-run.pl b/mysql-test/mysql-test-run.pl
index 9df96f636d3..4801cb77a4c 100755
--- a/mysql-test/mysql-test-run.pl
+++ b/mysql-test/mysql-test-run.pl
@@ -3980,14 +3980,14 @@ sub run_testcase ($$) {
   }
 
   my $test= $tinfo->{suite}->start_test($tinfo);
-  # Set only when we have to keep waiting after expectedly died server
-  my $keep_waiting_proc = 0;
+  # Set to a list of processes we have to keep waiting (expectedly died servers)
+  my %keep_waiting_proc = ();
   my $print_timeout= start_timer($print_freq * 60);
 
   while (1)
   {
-    my $proc;
-    if ($keep_waiting_proc)
+    my $proc = 0;
+    if (scalar(keys(%keep_waiting_proc)) > 0)
     {
       # Any other process exited?
       $proc = My::SafeProcess->check_any();
@@ -3997,48 +3997,34 @@ sub run_testcase ($$) {
       }
       else
       {
-	$proc = $keep_waiting_proc;
 	# Also check if timer has expired, if so cancel waiting
 	if ( has_expired($test_timeout) )
 	{
-	  $keep_waiting_proc = 0;
+	  %keep_waiting_proc = ();
 	}
       }
     }
-    if (! $keep_waiting_proc)
+    if (scalar(keys(%keep_waiting_proc)) == 0 && !$proc)
     {
-      if($test_timeout > $print_timeout)
+      if ($test_timeout > $print_timeout)
       {
-         $proc= My::SafeProcess->wait_any_timeout($print_timeout);
-         if ( $proc->{timeout} )
-         {
-            #print out that the test is still on
-            mtr_print("Test still running: $tinfo->{name}");
-            #reset the timer
-            $print_timeout= start_timer($print_freq * 60);
-            next;
-         }
+        $proc= My::SafeProcess->wait_any_timeout($print_timeout);
+        if ($proc->{timeout})
+        {
+          #print out that the test is still on
+          mtr_print("Test still running: $tinfo->{name}");
+          #reset the timer
+          $print_timeout= start_timer($print_freq * 60);
+          next;
+        }
       }
       else
       {
-         $proc= My::SafeProcess->wait_any_timeout($test_timeout);
+        $proc= My::SafeProcess->wait_any_timeout($test_timeout);
       }
     }
 
-    # Will be restored if we need to keep waiting
-    $keep_waiting_proc = 0;
-
-    unless ( defined $proc )
-    {
-      mtr_error("wait_any failed");
-    }
-    mtr_verbose("Got $proc");
-
-    mark_time_used('test');
-    # ----------------------------------------------------
-    # Was it the test program that exited
-    # ----------------------------------------------------
-    if ($proc eq $test)
+    if ($proc eq $test) # mysqltest itself exited
     {
       my $res= $test->exit_status();
 
@@ -4053,12 +4039,12 @@ sub run_testcase ($$) {
 
       if ( $res == 0 )
       {
-	my $check_res;
-	if ( $opt_check_testcases and
-	     $check_res= check_testcase($tinfo, "after"))
-	{
-	  if ($check_res == 1) {
-	    # Test case had sideeffects, not fatal error, just continue
+        my $check_res;
+        if ( $opt_check_testcases and
+             $check_res= check_testcase($tinfo, "after"))
+        {
+          if ($check_res == 1) {
+            # Test case had sideeffects, not fatal error, just continue
             if ($opt_warnings) {
               # Checking error logs for warnings, so need to stop server
               # gracefully so that memory leaks etc. can be properly detected.
@@ -4069,90 +4055,111 @@ sub run_testcase ($$) {
               # test.
             } else {
               # Not checking warnings, so can do a hard shutdown.
-	      stop_all_servers($opt_shutdown_timeout);
+              stop_all_servers($opt_shutdown_timeout);
             }
-	    mtr_report("Resuming tests...\n");
-	    resfile_output($tinfo->{'check'}) if $opt_resfile;
-	  }
-	  else {
-	    # Test case check failed fatally, probably a server crashed
-	    report_failure_and_restart($tinfo);
-	    return 1;
-	  }
-	}
-	mtr_report_test_passed($tinfo);
+            mtr_report("Resuming tests...\n");
+            resfile_output($tinfo->{'check'}) if $opt_resfile;
+          }
+          else {
+            # Test case check failed fatally, probably a server crashed
+            report_failure_and_restart($tinfo);
+            return 1;
+          }
+        }
+        mtr_report_test_passed($tinfo);
       }
       elsif ( $res == 62 )
       {
-	# Testcase itself tell us to skip this one
-	$tinfo->{skip_detected_by_test}= 1;
-	# Try to get reason from test log file
-	find_testcase_skipped_reason($tinfo);
-	mtr_report_test_skipped($tinfo);
-	# Restart if skipped due to missing perl, it may have had side effects
-	if ( $tinfo->{'comment'} =~ /^perl not found/ )
-	{
-	  stop_all_servers($opt_shutdown_timeout);
-	}
+        # Testcase itself tell us to skip this one
+        $tinfo->{skip_detected_by_test}= 1;
+        # Try to get reason from test log file
+        find_testcase_skipped_reason($tinfo);
+        mtr_report_test_skipped($tinfo);
+        # Restart if skipped due to missing perl, it may have had side effects
+        if ( $tinfo->{'comment'} =~ /^perl not found/ )
+        {
+          stop_all_servers($opt_shutdown_timeout);
+        }
       }
       elsif ( $res == 65 )
       {
-	# Testprogram killed by signal
-	$tinfo->{comment}=
-	  "testprogram crashed(returned code $res)";
-	report_failure_and_restart($tinfo);
+        # Testprogram killed by signal
+        $tinfo->{comment}=
+          "testprogram crashed(returned code $res)";
+        report_failure_and_restart($tinfo);
       }
       elsif ( $res == 1 )
       {
-	# Check if the test tool requests that
-	# an analyze script should be run
-	my $analyze= find_analyze_request();
-	if ($analyze){
-	  run_on_all($tinfo, "analyze-$analyze");
-	}
+        # Check if the test tool requests that
+        # an analyze script should be run
+        my $analyze= find_analyze_request();
+        if ($analyze){
+          run_on_all($tinfo, "analyze-$analyze");
+        }
 
-	# Wait a bit and see if a server died, if so report that instead
-	mtr_milli_sleep(100);
-	my $srvproc= My::SafeProcess::check_any();
-	if ($srvproc && grep($srvproc eq $_, started(all_servers()))) {
-	  $proc= $srvproc;
-	  goto SRVDIED;
-	}
+        # Wait a bit and see if a server died, if so report that instead
+        mtr_milli_sleep(100);
+        my $srvproc= My::SafeProcess::check_any();
+        if ($srvproc && grep($srvproc eq $_, started(all_servers()))) {
+          $proc= $srvproc;
+          goto SRVDIED;
+        }
 
-	# Test case failure reported by mysqltest
-	report_failure_and_restart($tinfo);
+        # Test case failure reported by mysqltest
+        report_failure_and_restart($tinfo);
       }
       else
       {
-	# mysqltest failed, probably crashed
-	$tinfo->{comment}=
-	  "mysqltest failed with unexpected return code $res\n";
-	report_failure_and_restart($tinfo);
+        # mysqltest failed, probably crashed
+        $tinfo->{comment}=
+          "mysqltest failed with unexpected return code $res\n";
+        report_failure_and_restart($tinfo);
       }
 
       # Save info from this testcase run to mysqltest.log
       if( -f $path_current_testlog)
       {
-	if ($opt_resfile && $res && $res != 62) {
-	  resfile_output_file($path_current_testlog);
-	}
-	mtr_appendfile_to_file($path_current_testlog, $path_testlog);
-	unlink($path_current_testlog);
+        if ($opt_resfile && $res && $res != 62) {
+          resfile_output_file($path_current_testlog);
+        }
+        mtr_appendfile_to_file($path_current_testlog, $path_testlog);
+        unlink($path_current_testlog);
       }
 
       return ($res == 62) ? 0 : $res;
-
     }
 
-    # ----------------------------------------------------
-    # Check if it was an expected crash
-    # ----------------------------------------------------
-    my $check_crash = check_expected_crash_and_restart($proc);
-    if ($check_crash)
+    if ($proc)
     {
-      # Keep waiting if it returned 2, if 1 don't wait or stop waiting.
-      $keep_waiting_proc = 0 if $check_crash == 1;
-      $keep_waiting_proc = $proc if $check_crash == 2;
+      # It was not mysqltest that exited, add to a wait-to-be-started-again list.
+      $keep_waiting_proc{$proc} = 1;
+    }
+
+    mtr_verbose("Got " . join(",", keys(%keep_waiting_proc)));
+
+    mark_time_used('test');
+    my $expected_exit = 1;
+    foreach my $wait_for_proc (keys(%keep_waiting_proc)) {
+      # ----------------------------------------------------
+      # Check if it was an expected crash
+      # ----------------------------------------------------
+      my $check_crash = check_expected_crash_and_restart($wait_for_proc);
+      if ($check_crash == 0) # unexpected exit/crash of $wait_for_proc
+      {
+        $expected_exit = 0;
+        last;
+      }
+      elsif ($check_crash == 1) # $wait_for_proc was started again by check_expected_crash_and_restart()
+      {
+        delete $keep_waiting_proc{$wait_for_proc};
+      }
+      elsif ($check_crash == 2) # we must keep waiting
+      {
+        # do nothing
+      }
+    }
+
+    if ($expected_exit) {
       next;
     }