Steps to reproduce * Execute a query that produces 50-60-70Mb log file per helper. It should crash. * Download the log from one or more helpers * Try to start a new query.