Ask Your Question
0

VM's across different compute nodes not able to ping with packet size more than 1500

asked 2014-09-30 02:51:30 -0500

prat gravatar image

I am running a two compute nodes devstack pod with VXLAN as an overlay protocol. When I am trying to ping VM's on same compute node as well as across different compute nodes with packet size 64 ( or anything less than 1400 ), I am able to successfully ping.

But when I am trying to ping with pakcet size above 1500 : 1. I am able to successfully ping between VM's on same compute nodes. 2. But My ping is unsuccessful when I am trying to ping across different compute nodes.

In my understanding, it could not be an guest OS issue, otherwise I won't be successful on even same compute nodes.

So, is there any specific flag for VXLAN which I have to set to ping across different compute nodes for larger packet sizes ( above 1500 ).

Thanks Prat

edit retag flag offensive close merge delete

1 answer

Sort by » oldest newest most voted
1

answered 2014-09-30 09:14:47 -0500

jbajin gravatar image

You have to add up the amount of overhead each of your protocols that you are using on your network and ensure that each underlying piece of your network can support that size or lower.

We ran into this problem as well. We had tenant VXLAN running which adds an overhead of 50 bytes. Then we underneath that we had VLAN that was moving data around. That was 4 bytes. So I had to add that in as well. So we had to move our VM's down to 1400 to cover all the bases.

edit flag offensive delete link more

Your Answer

Please start posting anonymously - your entry will be published after you log in or create a new account.

Add Answer

Get to know Ask OpenStack

Resources for moderators

Question Tools

1 follower

Stats

Asked: 2014-09-30 02:51:30 -0500

Seen: 267 times

Last updated: Sep 30 '14