Comware

HI there;

I have 2 HP 2810-48 Switches in 2 racks, 1 switch in each rack.

Ports 47 and 48 on each switch are configured as a trunk and connected to each switch.

I have VLANS 1 through 8 defined with the VLAN ID for each VLAN matching its name (ie: VLAN1 has id 1, VLAN2 has id 2, etc). I have tagged all VLANs to the trunk ports on both switches.

Spanning tree is off (the factory default)

Connected to these swicthes are about 20 linux hosts. Each linux hosts is setup for ethernet 'bonding' in active/passive mode.

2 of the above linux hosts are setup for 802.1q vlan tagging. That is to say that on each of these 2 hosts there is a 'bond0' interface defined with no ip address and then a number of bond0.X interfaces with X being numbers 1-8 to bind to each of the vlans I have configured. The switch ports that these 2 linux hosts doing vlan tagging are setup to tag all 8 VLANs.


PROBLEM:

I seem to be experiencing network hiccups that last 1-2 seconds every 1-3 minutes. I am suspecting that there is some sort of address table reset or relearning going on. This has been occuring since I implemented this new config a few days ago.


Thoughts? I can draw a network diagram to clarify but I dont have any useful tools installed right now.

Fyi, here's a jpeg image that shows a view of my network architecture with another attempt to explain my vlan scheme. Note that I only display 2 of my 'standard' linux hosts, but as indicated in my OP I have 20+ with more coming all the time.

Hi,

You don't need spanning-tree running in that setup so I don't think there are any issues there.
When you say there is a network hiccup, where in the network are you seeing this (monitoring on the hosts or externally..) and through what mechanism (ping, service checking etc.)
Is there anything in the switch logs to suggest a port issue?

Reply to Mohammed Faiz:

The 'hiccup' is presenting itself as a momentary pause (descibed in the OP above as a 1-2, possibly as long as 4 second pause) while typing in a command shell (ie: ssh remote shell login) on the linux hosts.

However, it turns out that my local ISP provider has been having minor issues the past couple of days and the source of my 'hiccups' could easily be related to the local ISP issues.

I just implemented the production architecture illustrated above and I'm possibly being a bit oversensitive to any small oddities while interacting with said new architecture.

Thanks for your response Mohammed.

It might be worth getting some monitoring going on the servers in question. If you monitor something like Google.com at the same time you should be able to see if the blips are just your server network or your isp.