Router OS inside GNS3 - TCP resets between host and guests.

RouterOS Virtualization
1

Did someone test Router OS inside GNS3 (latest version 2.28) ?

I have disconnects between Winbox on the Host and virtual QEMU Router OS VM inside GNS3. For example if i try to upload a file through Winbox on the host machine (Windows 7), i cannot get more than about 30 Ko transferred and Winbox is disconnected. I tried with a FTP client, same problem.

GNS3 VM is running inside VMWare. Cloud running in the GNS3 VM is used for connectivity between host and GNS3 Router OS instances, GNS3 VM is using the bridged VMWare driver on Eth0. Same results with host only.

Windows 7x64 Host, same problem with Router OS 6 or 7.

If i try to run a CHR VM directly inside VMware, there is no problem. No problem with EVE-NG neither. The problem seems related to GNS3 network layer, or QEMU version used inside GNS3 (4.2.1).

Seems like a buffering problem between GNS3VM and QEMU VMs causing TCP resets.

Transferring files is working from Winbox in MAC mode. IP mode only reveal the problem.

2

I have the exact same issue. Was going to report it to the GNS3 team as I believe the issue is on their end but never got around to it.

3

EVE-NG is using an older QEMU 2.12 version. GNS3 is using version 4.2.1. This could explain why NGS3 is not working here.

But, i did test inside EVE-NG with QEMU 4.1.0 and it’s working. Eventually QEMU 4.2.1 could still be the culprit but there is a bit more probability that the NGS3 VM is responsible of the problem.

Sadly GNS3 does not give the possibility to change the QEMU version for each node. At least not easily. EVE-NG do allow QEMU versions from 2.1.2 to 4.1.0.


Could be eventually a problem with bridging, i suspect duplicated packets. Not sure about that, more Wiresharking needed.