I use Moodle 4.1

Need to run script with php cli which fills csv file with nearly million rows of data.

<?php
define('CLI_SCRIPT', true);

require "config.php";

use core_reportbuildertablecustom_report_table_view;

$table = custom_report_table_view::create(264);
$table->setup();
$table->download = 'excel';
$columns = (new table_dataformat_export_format($table, $table->download))->format_data($table->headers);
[$sql, $params] = $table->getPaginatedDataSQL();

$filePath = '/var/excelreport/tmp_264_' . date("d-m-Y_H:i:s") . '.csv';

try {
    $file = fopen($filePath, 'w');
    fprintf($file, chr(0xEF).chr(0xBB).chr(0xBF));
    echo "File '$filePath' successfully created.";
} catch (Throwable $throwable) {
    echo $throwable->getMessage();
    echo "Failed to create '$filePath'.";
}


$limit = 10000;
$rowCount = $table->getRowsCount();
fputcsv($file, $columns,';');
for ($offset = 0; $offset < $rowCount; $offset += $limit) {
    $chunk = $DB->get_records_sql($sql. " LIMIT $limit OFFSET $offset", $params); // here $DB is global variable which comes from config.php
    foreach ($chunk as $row) {
        $row = array_values($table->format_row($row));
        fputcsv($file, $row,';');
    }
    echo "n offset $offset memory: ". (memory_get_usage() / 1024)."KBn" ;
    unset($chunk);
    gc_collect_cycles();
}
$table->close_recordset();
fclose($file);

Simply I fetch data chunks with size of 10000 rows from moodle’s db and append to file in each iteration with fputcsv. After each iteration I echo memory usage which increases as script runs.

enter image description here

Did I miss something to free or php somehow increases its memory footprint in long running tasks? AS a solution should I execute each chunk as seperate process with exec?