I’m facing a networking issue with my MMORPG server setup. The server infrastructure is as follows:
- Host: Windows Server
- Server Software and Services: Hosted inside a Virtual Machine running Debian
- Networking: The VM is bridged/NATed (specify as needed)
For some accounts, certain packets (e.g., packet 0x09a1, 2 bytes in length) are sent from the client (confirmed by local Wireshark and an external library hooked into the client), but never arrive at the server—Wireshark running on the Debian VM never sees them. For other accounts, everything works fine and packets are received as expected.
Technical Details
The problematic packets are clearly seen as sent in local Wireshark captures.
On the server (Debian VM), those same packets are completely missing from the capture when using a problematic account.
When comparing local vs server captures, I sometimes notice 4 extra bytes added at the end of the packet (e.g., f7 38 dc d7), for working accounts.
If I change to a working account ID, the packets are received correctly, and the extra 4 bytes are not present.
On the application side, the relevant breakpoint is never hit, confirming the packet never reaches the server logic.
Other TCP traffic between client and server works normally. Only certain packets with specific account IDs “disappear.”
No special firewall rules or network restrictions are configured for specific accounts or packets.
Additional Analysis
Cloned VM Works: I made an identical copy of the Debian VM (same configuration, same server software, same data) and ran it locally on my machine. In this setup, everything works perfectly—even for the problematic accounts, both locally and for external users connecting from outside.
SSH Forwarding Works: I also tested connecting via SSH port forwarding to the live server. When forwarding the emulator ports through SSH, the connection works fine for those accounts—the missing packets arrive as they should.
Suspect: The issue seems isolated to the networking path between the client and the Debian VM when running on the production Windows Server host. No OS/application logs report dropped packets or errors.
Has anyone experienced a scenario where TCP packets sent from the client never arrive at the server, but only for specific account IDs—and only on a particular Windows Server + Debian VM setup? Could this be caused by deep packet inspection, host firewall, virtualization networking settings, or certain offloading features (TCP Offloading, checksum offload, etc.) on the Windows Server host?
Any suggestions or diagnostic steps would be highly appreciated!
