About a month ago, we started seeing issues with our @ChainflowSol Firedancer validator staying synced with the network. After extensive troubleshooting, we decided to migrate back to the Anza/Jito client, until we were more confident about the root cause. About a week ago we deployed a server with a more powerful processor, after conversations with the Firedancer team and other operators, which led us to believe this might resolve the issue we had been experiencing. We ran a non-voting node on the new server first in performance, then in revenue mode, with the suggested tile configuration. Both stayed synced for at least a few days without issue. Today we migrated to the new server running Firedancer in performance mode with the suggested tile configuration. Unfortunately the new server started falling behind after about two hours. We're now migrating back to the server running the Anza/Jito client. While discouraging, we're not deterred and will continue troubleshooting until we're able to successfully run a stable Firedancer deployment.