Dropped packets in vSphere 6

[leo_gorn]

Hi all!

I have installed vSphere 6 (with last updates) on two hosts (Dell R730),  create some VMs, some VMs have been migrated from vSphere 4 (remove from inventory/ add to inventory).

I see a large number of dropped packets (only ‘Receive packets dropped’) for every VMs in vSphere client.

Esxtop as well as physical switches have no dropped packets.

What’s the matter?

Thanks.

[DavoudTeimouri]

Hi,

Change the virtual machines' NIC to VMXNET3 and update VMware Tools on them.

[ssumichrast]

‌im actually having the exact same issue in vSphere 6 (have been for a month or two now). I have a support case open but GS seems to be unable to resolve the issue. They're calling for driver updates.  I don't expect that to resolve my issue, but that is the game we have to play.

The  are affecting VMs -- we have some appliances that are application level failover. They're losing HA packets  causes them to fail over unnecessarily whenever heavier traffic (such as a backup) is going off. It is quite frustrating especially that I don't have a resolution on this. It makes no sense to me.

[de2rfg]

Same here, I didn't see any issue, only the rx drops in the VM's perf charts.  According to support this is a known regression in ESXi 6.0.x.

[leo_gorn]

All VMs have VMXNET3 NICs. Vmware Tools update does not resolve the problem

[Bono_]

There is issue with vspher6, this is log from Debian 7.2, I'm planning to update vSphere to U1 to see if that will help:

[302634.584663] e1000 0000:02:00.0 eth0: Detected Tx Unit Hang

  Tx Queue             <0>

  TDH                  <c>

  TDT                  <e>

  next_to_use          <e>

  next_to_clean        <c>

buffer_info[next_to_clean]

  time_stamp           <10480da66>

  next_to_watch        <d>

  jiffies              <10480dd6c>

  next_to_watch.status <0>

[302636.433553] e1000 0000:02:00.0 eth0: Detected Tx Unit Hang

  Tx Queue             <0>

  TDH                  <c>

  TDT                  <e>

  next_to_use          <e>

  next_to_clean        <c>

buffer_info[next_to_clean]

  time_stamp           <10480da66>

  next_to_watch        <d>

  jiffies              <10480df3a>

  next_to_watch.status <0>

[302637.091424] e1000 0000:02:00.0 eth0: Detected Tx Unit Hang

  Tx Queue             <0>

  TDH                  <c>

  TDT                  <e>

  next_to_use          <e>

  next_to_clean        <c>

buffer_info[next_to_clean]

  time_stamp           <10480da66>

  next_to_watch        <d>

  jiffies              <10480dfde>

  next_to_watch.status <0>

[302639.434786] e1000 0000:02:00.0 eth0: Detected Tx Unit Hang

  Tx Queue             <0>

  TDH                  <c>

  TDT                  <e>

  next_to_use          <e>

  next_to_clean        <c>

buffer_info[next_to_clean]

  time_stamp           <10480da66>

  next_to_watch        <d>

  jiffies              <10480e228>

  next_to_watch.status <0>

[302640.794140] e1000 0000:02:00.0 eth0: Reset adapter

[302640.854587] e1000: eth0 NIC Link is Up 1000 Mbps Full Duplex, Flow Control: None

[Lost_Steak]

Hi,

We have the same problem. Long story short, it is a known issues with vSphere

http://kb.vmware.com/kb/2052917

Turns out that it "is a cosmetic issue and does not indicate an actual network problem. This is a known issue affecting ESXi/ vCenter Server 6.0. Currently, there is no resolution."

Regards,

[Lost_Steak]

Just upgraded to version 6.2. this worked without a problem.

[HaniHammad]

would you check for rx dropped packets from the Guest OS ? if found check for the number of rx buffers or ring config for the guest adapter

[JamieGator32]

We are currently working through dropped packets issues.  So glad to see we aren't the only ones experiencing this. 

Thanks for sharing the link!