@digitalocean We migrated the critical part of our infrastructure to a different droplet that was stable which solved our problem.