Why appear dropped packets on LACP but there are no dropped packet on VSS ?

[baber]

I have two type of switch in my environment vss and vds now there is strange problem when  reside the vm in a portgroup in vss I dont have any packet deop on it such as follow statistics :

Packet statistics for port 33554442

   Packets received: 185

   Packets sent: 138

   Bytes received: 22487

   Bytes sent: 20783

   Broadcast packets received: 37

   Broadcast packets sent: 59

   Multicast packets received: 90

   Multicast packets sent: 15

   Unicast packets received: 58

   Unicast packets sent: 64

   Receive packets dropped: 0

   Transmit packets dropped: 0

But when I change the portgroup from vss to VDS appears drop packets on it such as follow :

Packet statistics for port 50331725

   Packets received: 3513

   Packets sent: 1162

   Bytes received: 42368

   Bytes sent: 10622

   Broadcast packets received: 1435

   Broadcast packets sent: 87

   Multicast packets received: 1172

   Multicast packets sent: 21

   Unicast packets received: 906

   Unicast packets sent: 1054

   Receive packets dropped: 0

   Transmit packets dropped: 252

my standard switch topology is : one standard switch with 2 physical adapter and set explicit failover order for this vm

on the other hand my VDS topology is LACP with one Lag that contain two physical adapter

What is the issue ?

BR

[scott28tt]

Your port group on the vDS is set to route based on IP hash? (Which would match your LACP/LAG setup)

[baber]

Exactly, that port group on LACP is Route Base o IP hash.

[baber]

Can anyone help me about this issue?

[baber]

I found this :

Capture Dropped Packets

now I used follow command :

pktcap-uw --capture Drop --srcip 172.20.121.21 --outfile /vmfs/volumes/LUN-Low-2/drop/test.pcap

My vm Ip Address is 172.20.121.21

but it show follow output :

The session capture point is Drop.

The session filter source IP address is 172.20.121.21.

The output file is /vmfs/volumes/LUN-Low-2/drop/test.pcap.

No server port specifed, select 41365 as the port.

Local CID 2.

Listen on port 41365.

Accept...

Vsock connection from port 1029 cid 2.

What is the problem ?

BR

[ZibiM]

Are your VDS uses LACP or IP Hash ? These are 2 separate things

What are the port configurations on the switches - portchannel (vpc) or static etherchannel ?

If you have portchannels on the port switches, then you should create LAG (active or passive) on the VDS and migrate your portgroups to use them

If you have etherchannel then you need to use IP Hash

Unfortunately it's not possible to say more without knowing your switch topology and port configuration.

[baber]

My port configuration on switch is VPC.

I have create LACP on vds with a lag but select route base on ip hash for that port group in teaming

According to vMware document “vsphere network”.

BR

[ZibiM]

This document is bit misleading

When you setting up load balancing on the portgroup together with LAG, you can see small info sign

It should state that you should pick you LAG active, all other uplinks as the unused and that the load balancing policy does not really matter, b/s it is decided by the LAG policy.

In my environment I left load balancing policy on originating port id.

Did not see any issues with that later on.

[baber]

I have attached 2 pic from my LACP config which is Active also PG config that is IP Hash .

I think all my configuration are Ok according to vMware Practices

BR

[baber]

Also I was read this document for troubleshoot :

Capture Dropped Packets

and used from this command :

pktcap-uw --capture Drop --srcip 172.20.121.21 --outfile /vmfs/volumes/LUN-DS/drop/test20201006.pcap --count 20

Is that correct ?

There was a note in that document :

You can see the reason and the place where a packet is dropped only when you capture packets to the console output. The pktcap-uw utility saves only the content of packets to a .pcap or .pcapng file.

I have type it in dcui with SSH now what does console means here ?

BR