Larg Recieve Packets Dropped on virtual machines

[DavoudTeimouri]

Hi,

I asked this question before but no body could help me.

The issue is happening again. All our virtual machines are using VMXNET 3 as network card. The issue is happening just on some virtual machines that they are not hosting by same ESXi.

More than 4000 packets dropping in 5 or 10 minutes. I've read about ring size or buffer size in NICs but I think, it's not root cause because it's happening on virtual machines randomly with any specific pattern.

It's affecting my users sounds.

My environment specification:

1. Physical Server: HP DL585 G7
2. Hypervisor: ESXi 5.1 (Customized by HP)
3. vCenter 5.1
4. Client OS: Windows XP 64-bit SP2
5. NIC Type: VMXNET 3
6. vCPU per VM: 2
7. RAM per VM: 2GB
8. vSwitch with 2 physical NICs

Ask me if you need to more information, I should clear this issue from my environment permanently.

Thanks

[tomtom901]

Are you using the VSS (Standard Switch) or DVS (Distributed vSwitch)

[DavoudTeimouri]

Using standard switches.

Two switches with two NICs.

[tomtom901]

Any hints on the physical switches? Packet drops there?

[DavoudTeimouri]

As I mentioned before, this problem was happened and our uplinks were root cause.

So I asked to our Network team to check that, they told me that this issue is not related to uplinks and switches.

Also it's happening on some virtual machines randomly.

[OscarDavey]

To resolve this issue, slowly increase both values in the guest OS.

To reduce burst traffic drops in Windows 2008 R2 Buffer Settings:

1. Click Start > Control Panel > Device Manager.
2. Right-click vmxnet3 and click Properties.
3. Click the Advanced tab.
4. Click Small Rx Buffers and increase the value. The default value is 1024 and the maximum is 4096.
5. Click Rx Ring #1 Size and increase the value. The default value is 512 and the maximum is 8192.

Notes: 

• It is important to increase the value of Small Rx Buffers and Rx Ring #1 gradually to avoid drastically increasing the memory overhead on the host and possibly causing performance issues if resources are close to capacity.
• If this issue occurs on only 2-3 virtual machines, set the value of Small Rx Buffers and Rx Ring #1 to the maximum value. Monitor virtual machine performance to see if this resolves the issue.
• The Small Rx Buffers and Rx Ring #1 variables affect non-jumbo frame traffic only on the adapter.

Hope this can help

Best regards

Yours, Oscar

[DavoudTeimouri]

I've read this article but it's Windows XP and the article talking about Windows 2008 R2.

But I'll test that.

[akkayyakapisett]

It may be because of driver issue.

As you are using HP customized drivers update the ESXi with latest patches - it should fix the issue.

Venkat

http://www.peeradmin.com

[DavoudTeimouri]

I'm using HP customized CD for all our HP servers.

I'll send some screen-shots.

I don't know, it's vCenter bug or It's happening, because if virtual machines have 8000 packets lost, why their users can't feel anything?