[pve-devel] openvswitch: ifupdown.sh : add linux_ports to allow plug classic

[Alexandre DERUMIER]

> - I like to manage bond with classic linux kernel (don't have tested it yet
> openvswitch bonding) ...

just tested to create eth0 (bnx2drver) with openvswitch

allow-vmbr0 eth0
iface eth0 inet manual
    ovs_bridge vmbr0
    ovs_type OVSPort


bnx2 is a hanging....I need to check why...

[  581.872983] bnx2 0000:05:00.0: irq 109 for MSI/MSI-X
[  584.138014] bnx2: fw sync timeout, reset code = 5030012
[  584.138110] bnx2 0000:05:00.0 eth0: <--- start MCP states dump --->
[  584.138219] bnx2 0000:05:00.0 eth0: DEBUG: MCP_STATE_P0[00000106] MCP_STATE_P1[f2fffff7]
[  584.138351] bnx2 0000:05:00.0 eth0: DEBUG: MCP mode[0000b800] state[80004000] evt_mask[00000500]
[  584.138492] bnx2 0000:05:00.0 eth0: DEBUG: pc[08006b88] pc[08006b88] instr[00401021]
[  584.138620] bnx2 0000:05:00.0 eth0: DEBUG: shmem states:
[  584.138711] bnx2 0000:05:00.0 eth0: DEBUG: drv_mb[05030012] fw_mb[00000011] link_status[0000f26f]
[  584.138851]  drv_pulse_mb[00000000]
[  584.138857] bnx2 0000:05:00.0 eth0: DEBUG: dev_info_signature[44564905] reset_type[01005254]
[  584.138995]  condition[00000106]
[  584.139014] bnx2 0000:05:00.0 eth0: DEBUG: 000001c0: 01005254 4253800a 00000106 fbffffff
[  584.139151] bnx2 0000:05:00.0 eth0: DEBUG: 000003cc: 44444444 44444444 44444444 00000a28
[  584.139284] bnx2 0000:05:00.0 eth0: DEBUG: 000003dc: 0004ffff 00000000 00000000 00000000
[  584.139416] bnx2 0000:05:00.0 eth0: DEBUG: 000003ec: 00000000 00000000 00000000 00ea37fd
[  584.139544] bnx2 0000:05:00.0 eth0: DEBUG: 0x3fc[0000ffff]
[  584.139635] bnx2 0000:05:00.0 eth0: <--- end MCP states dump --->

 


> currently I have this working config without any problem:
> 
> 
> -bond0.94---vmbr0 linux bridge
> -bond0---openvzswitch vmbr1-----tap
> 
> 
> But, I can't define bond0.94 if bond0 is create with openvswitch.

>>There is no way to setup that with OVS?

if bond0 is create with openswitch, I can't create bond0.94 with linux "classic" kernel.

Also a bond0 create with openswitch can be plugged to a linux bridge


Note,that this kind of config should be ok:

bond0-->openswitch (vmbr0)------>bridge (vmbr1)---->tap
                          ------>tap
                          ------>tap

like this, it's possible to use vmbr1 and vmbr0 on same bond0





----- Mail original ----- 

De: "Dietmar Maurer" <dietmar at proxmox.com> 
À: "Alexandre DERUMIER" <aderumier at odiso.com> 
Cc: pve-devel at pve.proxmox.com 
Envoyé: Mercredi 18 Décembre 2013 18:05:35 
Objet: RE: [pve-devel] openvswitch: ifupdown.sh : add linux_ports to allow plug classic 

> >>Mixing 2 different network stack make no sense for me. 
> 
> - I'm not sure openvz can plug on openswitch yet 
> - Maybe some users want to mix linux bridge with iptables filtering for some vm 
> and openvswitch for others vm. 

Users can create different types of bridges (OVS and linux). So there is no need to 
mix features? 

> - I'm not sure that nat/iptables can work with ip on openvswitch+iptables 

I someone want to use iptables, he can simply use a linux bridge. 

> - I like to manage bond with classic linux kernel (don't have tested it yet 
> openvswitch bonding) ... 

why? 

> currently I have this working config without any problem: 
> 
> 
> -bond0.94---vmbr0 linux bridge 
> -bond0---openvzswitch vmbr1-----tap 
> 
> 
> But, I can't define bond0.94 if bond0 is create with openvswitch. 

There is no way to setup that with OVS?