[PVE-User] implementing openvswitch

Brian Hart brianhart at ou.edu
Tue Jan 27 19:30:06 CET 2015


Sten,

Do you see any other indications that anything is failing in the cluster?
Things such as 'pvecm status' and 'clustat' - how do they output?  When
looking in the web interface are you able to see and connect to all nodes?
Do they show a green or red status?

I ask because I recently went round and around with clustering and open
vSwitch and in the end it was a minor misconfiguration on my part.  But I
have been through all the possible troubleshooting.  It SHOULD work
regardless of how the other nodes are configured with regards to OVS and
linux bridges.  I had one node configured with OVS and the other without
and they were able to talk for me without issue.  My problem came down to
an MTU mismatch between the nodes.

Brian


On Tue, Jan 27, 2015 at 2:40 AM, Sten Aus <sten.aus at eenet.ee> wrote:

>  Hi
>
> I have a problem when implementing Open vSwitch. When I configure my one
> node to use ovs, all cluster work is heavily disturbed - GUI lacks,
> commands take forever to reply.
> As soon as I take proxmox interface down, everything normalizes.
>
> Switches support multicast and IGMP snooping is turned off. Multicast ping
> (*ssmpingd* on one node and *asmping* on second node) works fluently on
> both directions.
>
> *Should cluster work when all of the nodes are using Linux bridges and one
> is using ovs?* This node has been a cluster member and has used Linux
> bridges as well.
>
> *On all nodes:*
> - running kernel: 2.6.32-34-pve
> - proxmox-ve-2.6.32: 3.3-139 (running kernel: 2.6.32-34-pve)
> - pve-manager: 3.3-5 (running version: 3.3-5/bfebec03)
>
> *On this node:*
> openvswitch-switch:
>   Installed: 2.3.0-1
>
> *ovs-vsctl show* output
>
> *# ovs-vsctl show*
> dec7d5db-216b-42cf-bee7-60886d30c44e
>     Bridge "vmbr1234"
>         Port backnet
>             tag: 10
>             Interface backnet
>                 type: internal
>         Port "eth0"
>             Interface "eth0"
>         Port proxmox
>             tag: 15
>             Interface proxmox
>                 type: internal
>         Port "storage84"
>             tag: 84
>             Interface "storage84"
>                 type: internal
>         Port "storage88"
>             tag: 88
>             Interface "storage88"
>                 type: internal
>         Port "vmbr1234"
>             Interface "vmbr1234"
>                 type: internal
>     ovs_version: "2.3.0"
>
> OVS is configured properly, at least I think so, because I have
> connectivity between hosts and I can ping and ssh between nodes, ISCSI
> connection is up and running.
>
> Any directions or ideas are helpful!
>
> Thanks!
> Sten
>
>
> On 21.01.15 15:53, Sten Aus wrote:
>
> Hi
>
> Will my cluster stay OK, if I implement openvswitch one by one for each
> node?
>
> Node A is empty, I've migrated all VMs to other nodes and will install
> openvswitch-switch and make config changes in /etc/network/interfaces, but
> if I don't manage to do it for all nodes today, will everything be okay and
> can I use this node?
>
> All of my switches are supporting multicast and it's turned on.
>
> Thanks!
>
>
>
> _______________________________________________
> pve-user mailing listpve-user at pve.proxmox.comhttp://pve.proxmox.com/cgi-bin/mailman/listinfo/pve-user
>
>
>
> _______________________________________________
> pve-user mailing list
> pve-user at pve.proxmox.com
> http://pve.proxmox.com/cgi-bin/mailman/listinfo/pve-user
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://pve.proxmox.com/pipermail/pve-user/attachments/20150127/52323818/attachment-0015.html>


More information about the pve-user mailing list