/**
* @dataProvider realPointsProvider
*/
public function testHeuristicWithRealPoints($jobid, $metric_x, $metric_y, $num_clusters, $num_noise = 0)
{
$points = $this->getRealPoints($jobid, $metric_x, $metric_y);
$dbscan = new DBSCAN();
list($cluster, $noise) = $dbscan->execute($points);
$assert_error = "";
$assert_error .= "Check the chart here:\n";
$assert_error .= " http://127.0.0.1:8080/dbscan?jobid={$jobid}\n";
$assert_error .= " Metric X: \"" . DBUtils::$TASK_METRICS[$metric_x] . "\"\n";
$assert_error .= " Metric Y: \"" . DBUtils::$TASK_METRICS[$metric_y] . "\"";
$this->assertCount($num_clusters, $cluster, $assert_error);
$this->assertCount($num_noise, $noise, $assert_error);
}
public function dbscanexecsDataAction()
{
//ini_set('memory_limit', '384M');
$db = $this->container->getDBUtils();
$this->buildFilters();
$whereClause = $this->filters->getWhereClause(array('execs' => 'e', 'clusters' => 'c'));
$table_name = "e";
$jobid = Utils::get_GET_string("jobid");
$metric_x = Utils::get_GET_int("metric_x") !== null ? Utils::get_GET_int("metric_x") : 0;
$metric_y = Utils::get_GET_int("metric_y") !== null ? Utils::get_GET_int("metric_y") : 1;
$task_type = $db->get_task_type(Utils::get_GET_string("task_type"));
list($bench, $job_offset, $id_exec) = $db->get_jobid_info($jobid);
// Calc pending dbscanexecs (if any)
$pending = $db->get_dbscanexecs_pending($bench, $job_offset, $metric_x, $metric_y, $task_type, $whereClause);
if (count($pending) > 0) {
$db->get_dbscan($pending[0]['jobid'], $metric_x, $metric_y, $task_type);
}
// Retrieve calculated dbscanexecs from database
$task_type_select = $db->get_task_type_query($task_type, $filter_null = true);
$query = "\n SELECT\n d.`id_exec`,\n d.`centroid_x`,\n d.`centroid_y`\n FROM aloja2.JOB_dbscan d, aloja2.execs e\n JOIN aloja2.clusters c USING (id_cluster)\n LEFT JOIN aloja_ml.predictions p USING (id_exec)\n WHERE\n d.`id_exec` = e.`id_exec` AND\n d.`bench` = :bench AND\n d.`job_offset` = :job_offset AND\n d.`metric_x` = :metric_x AND\n d.`metric_y` = :metric_y\n " . $task_type_select('d') . "\n {$whereClause}\n ;";
$query_params = array(":bench" => $bench, ":job_offset" => $job_offset, ":metric_x" => $metric_x, ":metric_y" => $metric_y);
// Since we are calculating new results, we have to bypass the cache
$_GET['NO_CACHE'] = 1;
$rows = $db->get_rows($query, $query_params);
$points = new Cluster();
// Used instead of a simple array to calc x/y min/max
foreach ($rows as $row) {
$points[] = new Point($row['centroid_x'], $row['centroid_y'], array('id_exec' => $row['id_exec']));
}
$dbscan = new DBSCAN();
list($clusters, $noise) = $dbscan->execute((array) $points);
$seriesData = array();
foreach ($clusters as $cluster) {
$data = array();
foreach ($cluster as $point) {
$data[] = array('x' => $point->x, 'y' => $point->y, 'id_exec' => $point->info['id_exec']);
}
if ($data) {
$seriesData[] = array('points' => $data, 'size' => $cluster->count(), 'x_min' => $cluster->getXMin(), 'x_max' => $cluster->getXMax(), 'y_min' => $cluster->getYMin(), 'y_max' => $cluster->getYMax());
}
}
$noiseData = array();
foreach ($noise as $point) {
$noiseData[] = array('x' => $point->x, 'y' => $point->y, 'id_exec' => $point->info['id_exec']);
}
$result = ['seriesData' => $seriesData, 'noiseData' => $noiseData, 'pending' => max(0, count($pending) - 1)];
header('Content-Type: application/json');
ob_start('ob_gzhandler');
echo json_encode($result, JSON_NUMERIC_CHECK);
}
public function dbscanexecsDataAction()
{
//ini_set('memory_limit', '384M');
$db = $this->container->getDBUtils();
$where_configs = '';
$table_name = "e";
$datefrom = Utils::read_params('datefrom', $where_configs, false, $table_name);
$dateto = Utils::read_params('dateto', $where_configs, false, $table_name);
$benchs = Utils::read_params('benchs', $where_configs, false, $table_name);
$nets = Utils::read_params('nets', $where_configs, false, $table_name);
$disks = Utils::read_params('disks', $where_configs, false, $table_name);
$blk_sizes = Utils::read_params('blk_sizes', $where_configs, false, $table_name);
$comps = Utils::read_params('comps', $where_configs, false, $table_name);
$id_clusters = Utils::read_params('id_clusters', $where_configs, false, $table_name);
$mapss = Utils::read_params('mapss', $where_configs, false, $table_name);
$replications = Utils::read_params('replications', $where_configs, false, $table_name);
$iosfs = Utils::read_params('iosfs', $where_configs, false, $table_name);
$iofilebufs = Utils::read_params('iofilebufs', $where_configs, false, $table_name);
$money = Utils::read_params('money', $where_configs, false, $table_name);
$datanodes = Utils::read_params('datanodess', $where_configs, false, $table_name);
$benchtype = Utils::read_params('bench_types', $where_configs, false, $table_name);
$vm_sizes = Utils::read_params('vm_sizes', $where_configs, false, $table_name);
$vm_coress = Utils::read_params('vm_coress', $where_configs, false, $table_name);
$vm_RAMs = Utils::read_params('vm_RAMs', $where_configs, false, $table_name);
$hadoop_versions = Utils::read_params('hadoop_versions', $where_configs, false, $table_name);
$types = Utils::read_params('types', $where_configs, false, $table_name);
$filters = Utils::read_params('filters', $where_configs, false, $table_name);
$allunchecked = isset($_GET['allunchecked']) ? $_GET['allunchecked'] : '';
$minexetime = Utils::read_params('minexetime', $where_configs, false, $table_name);
$maxexetime = Utils::read_params('maxexetime', $where_configs, false, $table_name);
$provider = Utils::read_params('providers', $where_configs, false, $table_name);
$jobid = Utils::get_GET_string("jobid");
$metric_x = Utils::get_GET_int("metric_x") !== null ? Utils::get_GET_int("metric_x") : 0;
$metric_y = Utils::get_GET_int("metric_y") !== null ? Utils::get_GET_int("metric_y") : 1;
$task_type = $db->get_task_type(Utils::get_GET_string("task_type"));
list($bench, $job_offset, $id_exec) = $db->get_jobid_info($jobid);
// Calc pending dbscanexecs (if any)
$pending = $db->get_dbscanexecs_pending($bench, $job_offset, $metric_x, $metric_y, $task_type, $where_configs);
if (count($pending) > 0) {
$db->get_dbscan($pending[0]['jobid'], $metric_x, $metric_y, $task_type);
}
// Retrieve calculated dbscanexecs from database
$task_type_select = $db->get_task_type_query($task_type, $filter_null = true);
$query = "\n SELECT\n d.`id_exec`,\n d.`centroid_x`,\n d.`centroid_y`\n FROM `JOB_dbscan` d, `execs` e\n WHERE\n d.`id_exec` = e.`id_exec` AND\n d.`bench` = :bench AND\n d.`job_offset` = :job_offset AND\n d.`metric_x` = :metric_x AND\n d.`metric_y` = :metric_y\n " . $task_type_select('d') . "\n {$where_configs}\n ;";
$query_params = array(":bench" => $bench, ":job_offset" => $job_offset, ":metric_x" => $metric_x, ":metric_y" => $metric_y);
// Since we are calculating new results, we have to bypass the cache
$_GET['NO_CACHE'] = 1;
$rows = $db->get_rows($query, $query_params);
$points = new Cluster();
// Used instead of a simple array to calc x/y min/max
foreach ($rows as $row) {
$points[] = new Point($row['centroid_x'], $row['centroid_y'], array('id_exec' => $row['id_exec']));
}
$dbscan = new DBSCAN();
list($clusters, $noise) = $dbscan->execute((array) $points);
$seriesData = array();
foreach ($clusters as $cluster) {
$data = array();
foreach ($cluster as $point) {
$data[] = array('x' => $point->x, 'y' => $point->y, 'id_exec' => $point->info['id_exec']);
}
if ($data) {
$seriesData[] = array('points' => $data, 'size' => $cluster->count(), 'x_min' => $cluster->getXMin(), 'x_max' => $cluster->getXMax(), 'y_min' => $cluster->getYMin(), 'y_max' => $cluster->getYMax());
}
}
$noiseData = array();
foreach ($noise as $point) {
$noiseData[] = array('x' => $point->x, 'y' => $point->y, 'id_exec' => $point->info['id_exec']);
}
$result = ['seriesData' => $seriesData, 'noiseData' => $noiseData, 'pending' => max(0, count($pending) - 1)];
header('Content-Type: application/json');
ob_start('ob_gzhandler');
echo json_encode($result, JSON_NUMERIC_CHECK);
}
/**
* Calculates the DBSCAN.
*/
public function get_dbscan($jobid, $metric_x, $metric_y, $task_type, $eps = null, $minPoints = null)
{
$query_select1 = $this->get_task_metric_query($this::$TASK_METRICS[$metric_x]);
$query_select2 = $this->get_task_metric_query($this::$TASK_METRICS[$metric_y]);
$task_type_select = $this->get_task_type_query($task_type);
$query = "\n SELECT\n t.`TASKID` as TASK_ID,\n " . $query_select1('t') . " as TASK_VALUE_X,\n " . $query_select2('t') . " as TASK_VALUE_Y\n FROM `JOB_tasks` t\n WHERE t.`JOBID` = :jobid\n " . $task_type_select('t') . "\n ORDER BY t.`TASKID`\n ;";
$query_params = array(":jobid" => $jobid);
$rows = $this->get_rows($query, $query_params);
$points = new Cluster();
// Used instead of a simple array to calc x/y min/max
foreach ($rows as $row) {
$task_id = $row['TASK_ID'];
$task_value_x = $row['TASK_VALUE_X'] ?: 0;
$task_value_y = $row['TASK_VALUE_Y'] ?: 0;
// Show only task id (not the whole string)
$task_id = substr($task_id, 23);
$points[] = new Point($task_value_x, $task_value_y, array('task_id' => $task_id));
}
$dbscan = new DBSCAN($eps, $minPoints);
$dbscan->execute((array) $points);
// If heuristic was used, cache results in database
if ($eps === null && $minPoints === null) {
$this->add_dbscan($jobid, $metric_x, $metric_y, $task_type, $dbscan->getClusters());
}
return $dbscan;
}
