VMware Cloud Community
TimR26
Enthusiast
Enthusiast

Server 2019 and vSphere 6.7U3 - Large file transfers cause intermittent disconnects

I'm running about a dozen Server 2019 servers all using VMXNET3 NICs on a vSAN 6.7 U3 cluster, using 4 x 10GB uplinks with MTU 9000. When I attempt to transfer large files (roughly 100MB or more) via RDP with folder redirection with copy/paste, Windows Admin Center file transfer, or setting up a filezilla server on the OS can used a client to upload files, the files will transfer for a second or two, then drops the connection completely. For RDP I observer a frozen screen then reattempts to connect and after a few seconds it usually reconnects with an incomplete file transfer. On Windows Admin Center, I get an error (it uses Powershell to transfer the file and gets errors related), and in FileZilla, observing the mgmt interface, the transfer stalls during the disconnect. during the FTP scenario, I'll use VMRC to connect to the guest OS and monitor the mgmt interface of FTP Server and see delays in the file transfer. I also run a ping from my workstation to the FTP server and will observe random packet loss, specifically 4-5 successful packets followed by 4-5 packets lost and this repeats.

In one sparticular instance, vCenter reported the VM NIC was disconnected from the vDS switch and couldn't reconnect it. I had to remove the vNIC and create a new one for it to connect to the vDS. The file transfer issue persists. I did observe this large file transfer issue on other Server 2019 guest OSes but i'm having a hard time trying to discover the root cause.

I did attempt to upgrade VMtools on my FTP server to 11.0.5 to see if it was a undiscovered issue, but the issue remains. I'm thinking there may be an issue with the Server OS itself, but i'd like to do my due diligence on the VMware side of things before moving on to the OS.

Anyone experience this issue, or can anyone recommend where to start first for adv troubleshooting...a specific log, etc?

0 Kudos
5 Replies
Alex_Romeo
Leadership
Leadership

HI,

Normally the MTU setting of the networks is set identical to the MTU setting of the switch port. try to check.

ARomeo

Blog: https://www.aleadmin.it/
0 Kudos
TimR26
Enthusiast
Enthusiast

Physical switches have jumbo frames enabled, all physical ports have MTU configured to MTU 9000, vDS port groups are configured with MTU 9000. NICs in guest OS are default.

To add, this only happens on Server 2019. Server 2016 and older have no issues.

0 Kudos
Alex_Romeo
Leadership
Leadership

Hi,

Well! excellent explanation ...

at the same time you should open a post in the Microsoft Technet community.

I try to understand what happens and I rewrite you.

ARomeo

Blog: https://www.aleadmin.it/
0 Kudos
veldthui
Enthusiast
Enthusiast

What is your physical NIC? There is a driver issue with the ConnectX-3 NIC that causes symptoms like that. I had issues with mine and changed to an Intel brand NIC and no more issues.

0 Kudos
TimR26
Enthusiast
Enthusiast

We are using DL380 Gen10 servers, using 1 x HPE Ethernet 10/25Gb 2-port 640FLR-SFP28 and 1 x HPE Ethernet 10/25Gb 2-port 640SFP28 Adapter. They are connected to a pair of Cisco Nexus 5400 series (not sure the exact model) switches.

0 Kudos