Dear All,
Our infrastructure is a Vmware vsphere 6 host OS which serves 3 Windows Server 2008 r2 as guest OSes.
The users connect through RDP from Linux thin clients to the three Windows Server 2008 r2 by means of Terminal Server
We have an issue with those Windows Server 2008 R2. During network congestion, all users from the three Terminal Server receive unresponsive status on their sessions.
During such unresponsiveness, on taskmgr CPU appears at 6%, the RAM memory is used at 60% and the disk activity appears at 1 Mbyte/sec. The network activity during it vary between 4Mbps and 100Mbps from one of the Windows Server 2008 r2 -probably similar on the rest of two Windows Server
There are 17 users per Terminal Server. When one of the users from one Terminal Server experiences the unresponsiveness, all the users from the three Terminal Servers experience it as well.
I suspect this is something related to the windows network configuration or the vmxnet3 network adapter configuration.
I tried enabling Receive Side Scale on the vmxnet3 properties, but the unresponsiveness periods became longer in time, so I disabled it again.
I show you the Windows network configuration:
C:\Users\juan.magraner>netsh int tcp show global
Querying active state...
TCP Global Parameters
----------------------------------------------
Receive-Side Scaling State : enabled
Chimney Offload State : automatic
NetDMA State : enabled
Direct Cache Acess (DCA) : disabled
Receive Window Auto-Tuning Level : normal
Add-On Congestion Control Provider : ctcp
ECN Capability : disabled
RFC 1323 Timestamps : disabled
C:\Users\juan.magraner>
And the vmxnet3 configuration is as follows:
Enable adaptive rx ring sizing: Enabled
Interrupt moderation: Enabled
IPv4 Checksum Offload: Rx & Tx Enabled
IPv4 TSO Offload: Enabled
Jumbo Packet: Standard 1500
Large Rx Buffers: Not present
Large Send Offload V2 (IPv4): Enabled
Large Send Offload V2 (IPv6): Enabled
MAC address: Not present
Max Tx Queues: Not present
Maximum number of RSS Processors: Not present
Offload IP options: Enabled
Offload tagged traffic: Enabled
Offload TCP options: Enabled
Priority / VLAN tag: Priority & VLAN Enabled
Receive Side Scaling: Disabled
Receive Throttle: Not present
RSS Base Processor Number: Not present
Rx Ring #1 Size: Not present
Rx Ring #2 Size: Not present
Small Rx Buffers: Not present
Speed & Duplex: Auto Negotiation
TCP Checksum Offload (IPv4): Rx & Tx Enabled
TCP Checksum Offload (IPv6): Rx & Tx Enabled
Tx Ring Size: Not present
UDP Checksum Offload (IPv4): Rx & Tx Enabled
UDP Checksum Offload (IPv6): Rx & Tx Enabled
VLAN ID: Not present
Wake on magic packet: Enabled
Wake on pattern match: Enabled
Wake-on-LAN: Enabled
Reading the above configuration, could you recommend me if I need to change anything? I've read different opinions searching on Google.
Thank you for your consideration.
Kind Regards,
videoclocknet
I've had the exact same issues you have described and have been trying to figure out exactly how best to set up Receive Side Scaling.
However, I have found this to work to at least kill the intermittent (yet predictable) bottlenecks in traffic:
Enable adaptive rx ring sizing: Enabled
Interrupt moderation: Enabled
IPv4 Checksum Offload: Rx & Tx Disabled
IPv4 TSO Offload: Enabled
Jumbo Packet: Standard 1500
Large Rx Buffers: Not present
Large Send Offload V2 (IPv4): Disabled
Large Send Offload V2 (IPv6): Disabled
MAC address: Not present
Max Tx Queues: Not present
Maximum number of RSS Processors: Not present
Offload IP options: Enabled
Offload tagged traffic: Enabled
Offload TCP options: Enabled
Priority / VLAN tag: Priority & VLAN Enabled
Receive Side Scaling: Disabled
Receive Throttle: Not present
Recv Segment Coalescing (IPv4): Disabled
Recv Segment Coalescing (IPv6): Disabled
RSS Base Processor Number: Not present
Rx Ring #1 Size: Not present
Rx Ring #2 Size: Not present
Small Rx Buffers: Not present
Speed & Duplex: Auto Negotiation
TCP Checksum Offload (IPv4): Rx & Tx Disabled
TCP Checksum Offload (IPv6): Rx & Tx Disabled
Tx Ring Size: Not present
UDP Checksum Offload (IPv4): Rx & Tx Disabled
UDP Checksum Offload (IPv6): Rx & Tx Disabled
VLAN ID: Not present
Wake on magic packet: Enabled
Wake on pattern match: Enabled
Wake-on-LAN: Enabled
I have encountered the same problem on most new deployments of Windows Server on VMware with ESXi 5 and up. These settings invariably solve the problem.
Dear WavePoint2014 user,
Thank you for your recommendations.
I will try them on a test environment before and will see the results.
Kind Regards