The only way I can think of would be if some other process writes to the same disk. It could be the storage node logfile that keeps growing. The way this works is that the storage node will stop accepting new files around 500MB free space. It checks the disk space. Let’s assume there would be another process that fills these 500MB. That would mean the storage node will wait for some delete messages to free up another 500MB. The other process can fill that again. The longer this goes on the less used space your node will have.