Re: [ovs-discuss] Bonding configuration for KVM guests

2023-11-07 Thread Xuo Guoto via discuss 
--- Original Message ---
On Friday, October 13th, 2023 at 6:49 PM, Ilya Maximets  
wrote:


> > Am I correct in assuming that ovs is sending LACP packets 
> > correctly while Ubuntu is not responding and not sending its own 
> > LACP packets?
> 
> 
> Yes, looks like it.

I took another stab at this on another set of VM and it's now working!

The difference I can see is that I added the interfaces in netplan yml ie:

network:
  ethernets:
ens1:
  addresses:
  - 192.168.200.148/24
  nameservers:
addresses:
- 8.8.8.8
- 8.8.4.4
  routes:
  - to: default
via: 192.168.200.1
enp2s11:
  dhcp4: no
enp2s12:
  dhcp4: no
enp2s13:
  dhcp4: no
enp2s14:
  dhcp4: no
enp2s15:
  dhcp4: no
enp2s16:
  dhcp4: no
enp2s17:
  dhcp4: no
enp2s18:
  dhcp4: no
  bonds:
bond0:
  addresses: [192.168.201.141/24]
  interfaces:
 - enp2s11
 - enp2s12
 - enp2s13
 - enp2s14
 - enp2s15
 - enp2s16
 - enp2s17
 - enp2s18
  parameters:
mode: 802.3ad
mii-monitor-interval: 100
  renderer: networkd
  version: 2

If you notice, the device enp2s11 is added twice, first under ethernets and the 
under bond:interfaces. This is the only change I can think of from previous 
configuration. 

At the OVS side:

sudo ovs-appctl bond/show bond0
 bond0 
bond_mode: active-backup
bond may use recirculation: no, Recirc-ID : -1
bond-hash-basis: 0
lb_output action: disabled, bond-id: -1
updelay: 0 ms
downdelay: 0 ms
lacp_status: negotiated
lacp_fallback_ab: false
active-backup primary: 
active member mac: fe:54:00:7c:71:79(vnet365)

member vnet365: enabled
  active member
  may_enable: true

member vnet366: enabled
  may_enable: true

member vnet367: enabled
  may_enable: true

member vnet368: enabled
  may_enable: true

member vnet369: enabled
  may_enable: true

member vnet370: enabled
  may_enable: true

member vnet371: enabled
  may_enable: true

member vnet372: enabled
  may_enable: true

I am able to ping between another host in the same network/switch and so I 
guess its working fine.

Thanks a lot to Ilya Maximets for patiently helping me.

X.
___
discuss mailing list
disc...@openvswitch.org
https://mail.openvswitch.org/mailman/listinfo/ovs-discuss

Re: [ovs-discuss] Bonding configuration for KVM guests

2023-10-13 Thread Ilya Maximets via discuss 
On 10/13/23 14:22, Xuo Guoto via discuss wrote:
> --- Original Message ---
> On Friday, October 13th, 2023 at 4:51 PM, Ilya Maximets  
> wrote:
> 
>> I'd suggest you disable the lacp-fallback-ab, so the bond doesn't
>> fall back to active-backup mode on failures. At least while you're
>> troubleshooting.
> 
> Executed 
> 
> # ovs-vsctl set port bond0 bond_mode=balance-tcp lacp=active 
> other-config:lacp-fallback-ab=false other-config:lacp-time=slow
> 
> Now all the interfaces are disabled

This is expected.  They were enabled before, because of ab fallback.

> 
>  bond0 
> bond_mode: balance-tcp
> bond may use recirculation: yes, Recirc-ID : 1
> bond-hash-basis: 0
> lb_output action: disabled, bond-id: -1
> updelay: 0 ms
> downdelay: 0 ms
> next rebalance: 9560 ms
> lacp_status: configured
> lacp_fallback_ab: false
> active-backup primary: 
> active member mac: 00:00:00:00:00:00(none)
> 
> member vnet74: disabled
>   may_enable: false
> 
> member vnet75: disabled
>   may_enable: false
> 
> member vnet76: disabled
>   may_enable: false
> 
> member vnet77: disabled
>   may_enable: false
> 
> member vnet78: disabled
>   may_enable: false
> 
> member vnet79: disabled
>   may_enable: false
> 
> 
>> Also, how does the flow table look like? i.e. what's in the output
>> of 'ovs-ofctl dump-flows ovsbr-lacp0' ? In case there are rules that
>> would drop LACP traffic.
> 
> # ovs-ofctl dump-flows ovsbr-lacp0
>  cookie=0x0, duration=103674.235s, table=0, n_packets=6798, n_bytes=2111314, 
> priority=0 actions=NORMAL
> 
> I guess this is fine.

Yep.  Looks fine.

> 
> I did some more digging and when running tcpdump on both the ovs and on 
> Ubuntu I get this packet repeatedly at both interfaces:
> 
> 12:23:17.081894 LACPv1, length 110
> Actor Information TLV (0x01), length 20
>   System a6:a0:01:43:3a:41, System Priority 65534, Key 15, Port 17, 
> Port Priority 65535
>   State Flags [Activity, Aggregation, Collecting, Distributing, 
> Default]
> Partner Information TLV (0x02), length 20
>   System 00:00:00:00:00:00, System Priority 0, Key 0, Port 0, Port 
> Priority 0
>   State Flags [none]
> Collector Information TLV (0x03), length 16
>   Max Delay 0
> Terminator TLV (0x00), length 0
> 
> I am assuming this is a packet originated from a6:a0:01:43:3a:41 which 
> happens to be the MAC of ovsbr-lacp0
> 
> # ovs-vsctl -f table --columns=name,mac_in_use list interface | grep 
> a6:a0:01:43:3a:41
> ovsbr-lacp0  "a6:a0:01:43:3a:41"
> 
> Am I correct in assuming that ovs is sending LACP packets correctly while 
> Ubuntu is not responding and not sending its own LACP packets? 

Yes, looks like it.  I'm not an expert in configuring native linux
bonds, so I'm not sure how to troubleshoot that.  Are the ports
including the bond port UP in the guest?

Best regards, Ilya Maximets.
___
discuss mailing list
disc...@openvswitch.org
https://mail.openvswitch.org/mailman/listinfo/ovs-discuss

Re: [ovs-discuss] Bonding configuration for KVM guests

2023-10-13 Thread Xuo Guoto via discuss 
--- Original Message ---
On Friday, October 13th, 2023 at 4:51 PM, Ilya Maximets  
wrote:

> I'd suggest you disable the lacp-fallback-ab, so the bond doesn't
> fall back to active-backup mode on failures. At least while you're
> troubleshooting.

Executed 

# ovs-vsctl set port bond0 bond_mode=balance-tcp lacp=active 
other-config:lacp-fallback-ab=false other-config:lacp-time=slow

Now all the interfaces are disabled

 bond0 
bond_mode: balance-tcp
bond may use recirculation: yes, Recirc-ID : 1
bond-hash-basis: 0
lb_output action: disabled, bond-id: -1
updelay: 0 ms
downdelay: 0 ms
next rebalance: 9560 ms
lacp_status: configured
lacp_fallback_ab: false
active-backup primary: 
active member mac: 00:00:00:00:00:00(none)

member vnet74: disabled
  may_enable: false

member vnet75: disabled
  may_enable: false

member vnet76: disabled
  may_enable: false

member vnet77: disabled
  may_enable: false

member vnet78: disabled
  may_enable: false

member vnet79: disabled
  may_enable: false


> Also, how does the flow table look like? i.e. what's in the output
> of 'ovs-ofctl dump-flows ovsbr-lacp0' ? In case there are rules that
> would drop LACP traffic.

# ovs-ofctl dump-flows ovsbr-lacp0
 cookie=0x0, duration=103674.235s, table=0, n_packets=6798, n_bytes=2111314, 
priority=0 actions=NORMAL

I guess this is fine.

I did some more digging and when running tcpdump on both the ovs and on Ubuntu 
I get this packet repeatedly at both interfaces:

12:23:17.081894 LACPv1, length 110
Actor Information TLV (0x01), length 20
  System a6:a0:01:43:3a:41, System Priority 65534, Key 15, Port 17, 
Port Priority 65535
  State Flags [Activity, Aggregation, Collecting, Distributing, Default]
Partner Information TLV (0x02), length 20
  System 00:00:00:00:00:00, System Priority 0, Key 0, Port 0, Port 
Priority 0
  State Flags [none]
Collector Information TLV (0x03), length 16
  Max Delay 0
Terminator TLV (0x00), length 0

I am assuming this is a packet originated from a6:a0:01:43:3a:41 which happens 
to be the MAC of ovsbr-lacp0

# ovs-vsctl -f table --columns=name,mac_in_use list interface | grep 
a6:a0:01:43:3a:41
ovsbr-lacp0  "a6:a0:01:43:3a:41"

Am I correct in assuming that ovs is sending LACP packets correctly while 
Ubuntu is not responding and not sending its own LACP packets? 
___
discuss mailing list
disc...@openvswitch.org
https://mail.openvswitch.org/mailman/listinfo/ovs-discuss

Re: [ovs-discuss] Bonding configuration for KVM guests

2023-10-13 Thread Ilya Maximets via discuss 
On 10/13/23 12:55, Xuo Guoto wrote:
> 
> --- Original Message ---
> On Friday, October 13th, 2023 at 3:51 PM, Ilya Maximets  
> wrote:
> 
> Thanks again for taking time to answer my questions!
> 
> 
>> Here is your problem. The bond is in active-backup mode. This mode doesn't
>> support LACP. You need to switch to the balanced mode, e.g. by adding
>> 'bond_mode=balance-tcp' to the bond creation command. Or you may set it after
>> the fact with:
>>
>> ovs-vsctl set port bond0 bond_mode=balance-tcp
> 
> After some trial and error I got this command from Redhat bugzilla BZ#1267291
> 
> ovs-vsctl set port bond0 bond_mode=balance-tcp lacp=active 
> other-config:lacp-fallback-ab=true other-config:lacp-time=fast
> 
> Now the output is 
> 
> ovs-appctl bond/show bond0
>  bond0 
> bond_mode: balance-tcp
> bond may use recirculation: no, Recirc-ID : -1
> bond-hash-basis: 0
> lb_output action: disabled, bond-id: -1
> updelay: 0 ms
> downdelay: 0 ms
> lacp_status: configured
> lacp_fallback_ab: true
> active-backup primary: 
> active member mac: fe:54:00:79:8e:a0(vnet77)
> 
> member vnet74: enabled
>   may_enable: true
> 
> member vnet75: enabled
>   may_enable: true
> 
> member vnet76: enabled
>   may_enable: true
> 
> member vnet77: enabled
>   active member
>   may_enable: true
> 
> member vnet78: enabled
>   may_enable: true
> 
> member vnet79: enabled
>   may_enable: true
> 
> I can see that all member interfaces are enabled now, and I get an active 
> member mac. But still there is no ping between two devices in the switch. 
> Also in linux I can see this error bond0: Warning: No 802.3ad response from 
> the link partner for any adapters in the bond
> 
> Could lb_output action: disabled, bond-id: -1 be the reason the bond0 is not 
> getting up?

No, it's just an optimization and kernel datapath doesn't support it.
So, should not affect the actual traffic.

You have an 'lacp_status: configured', meaning the negotiation with
the other side didn't work out.  It should be in 'negotiated' state
if everything is working.

I'd suggest you disable the lacp-fallback-ab, so the bond doesn't
fall back to active-backup mode on failures.  At least while you're
troubleshooting.

And you probably better use slow lacp-time as well.


Also, how does the flow table look like?  i.e. what's in the output
of 'ovs-ofctl dump-flows ovsbr-lacp0' ?  In case there are rules that
would drop LACP traffic.

Best regards, Ilya Maximets.
___
discuss mailing list
disc...@openvswitch.org
https://mail.openvswitch.org/mailman/listinfo/ovs-discuss

Re: [ovs-discuss] Bonding configuration for KVM guests

2023-10-13 Thread Xuo Guoto via discuss 


--- Original Message ---
On Friday, October 13th, 2023 at 3:51 PM, Ilya Maximets  
wrote:

Thanks again for taking time to answer my questions!


> Here is your problem. The bond is in active-backup mode. This mode doesn't
> support LACP. You need to switch to the balanced mode, e.g. by adding
> 'bond_mode=balance-tcp' to the bond creation command. Or you may set it after
> the fact with:
> 
> ovs-vsctl set port bond0 bond_mode=balance-tcp

After some trial and error I got this command from Redhat bugzilla BZ#1267291

ovs-vsctl set port bond0 bond_mode=balance-tcp lacp=active 
other-config:lacp-fallback-ab=true other-config:lacp-time=fast

Now the output is 

ovs-appctl bond/show bond0
 bond0 
bond_mode: balance-tcp
bond may use recirculation: no, Recirc-ID : -1
bond-hash-basis: 0
lb_output action: disabled, bond-id: -1
updelay: 0 ms
downdelay: 0 ms
lacp_status: configured
lacp_fallback_ab: true
active-backup primary: 
active member mac: fe:54:00:79:8e:a0(vnet77)

member vnet74: enabled
  may_enable: true

member vnet75: enabled
  may_enable: true

member vnet76: enabled
  may_enable: true

member vnet77: enabled
  active member
  may_enable: true

member vnet78: enabled
  may_enable: true

member vnet79: enabled
  may_enable: true

I can see that all member interfaces are enabled now, and I get an active 
member mac. But still there is no ping between two devices in the switch. Also 
in linux I can see this error bond0: Warning: No 802.3ad response from the link 
partner for any adapters in the bond

Could lb_output action: disabled, bond-id: -1 be the reason the bond0 is not 
getting up?

X.
___
discuss mailing list
disc...@openvswitch.org
https://mail.openvswitch.org/mailman/listinfo/ovs-discuss

Re: [ovs-discuss] Bonding configuration for KVM guests

2023-10-13 Thread Ilya Maximets via discuss 
On 10/13/23 07:22, Xuo Guoto via discuss wrote:
> --- Original Message ---
> On Thursday, October 12th, 2023 at 9:03 PM, Ilya Maximets 
>  wrote:
> 
> Thanks again for the detailed reply!
> 
>> If you want to preserve these, you'll need to re-add them manually.
>> Potentially as part of the same transaction, by adding more commands,
>> i.e.
>>
>> ovs-vsctl \
>> del-port vnet75 \
>> -- del-port vnet78 \
> 
> I had followed your advice and used the following command to create the bond
> 
> ovs-vsctl \
>   del-port vnet75 \
>-- del-port vnet78 \
>-- del-port vnet79 \
>-- del-port vnet77 \
>-- del-port vnet74 \
>-- del-port vnet76 \
>-- add-bond ovsbr-lacp0 bond0 vnet75 vnet78 vnet79 vnet77 vnet74 vnet76 
> lacp=active \
>-- set Interface vnet75 \
> external_ids='{attached-mac="52:54:00:0d:61:fc", iface-id="", 
> iface-status=active, vm-id=""}' \
>-- set Interface vnet78 \
> external_ids='{attached-mac="52:54:00:24:36:a7", iface-id="", 
> iface-status=active, vm-id=""}' \
>-- set Interface vnet79 \
> external_ids='{attached-mac="52:54:00:c7:90:3c", iface-id="", 
> iface-status=active, vm-id=""}' \
>-- set Interface vnet77 \
> external_ids='{attached-mac="52:54:00:79:8e:a0", iface-id="", 
> iface-status=active, vm-id=""}' \
>-- set Interface vnet74 \
> external_ids='{attached-mac="52:54:00:aa:fb:bb", iface-id="", 
> iface-status=active, vm-id=""}' \
>-- set Interface vnet76 \
> external_ids='{attached-mac="52:54:00:d4:cc:88", iface-id="", 
> iface-status=active, vm-id=""}'
> 
> Note: iface-id and vm-id values have been removed for brevity.
> 
> Now the command was successful and I could see the bond0 port with the 
> specified interfaces
> 
> Bridge ovsbr-lacp0
> Port bond0
> Interface vnet78
> Interface vnet75
> Interface vnet79
> Interface vnet74
> Interface vnet77
> Interface vnet76
> Port ovsbr-lacp0
> Interface ovsbr-lacp0
> type: internal
> Port vnet72
> Interface vnet72
> 
> But the bonding is not working correctly
> 
> $ sudo ovs-appctl bond/show bond0
>  bond0 
> bond_mode: active-backup

Here is your problem.  The bond is in active-backup mode.  This mode doesn't
support LACP.  You need to switch to the balanced mode, e.g. by adding
'bond_mode=balance-tcp' to the bond creation command.  Or you may set it after
the fact with:

  ovs-vsctl set port bond0 bond_mode=balance-tcp

Best regards, Ilya Maximets.

> bond may use recirculation: no, Recirc-ID : -1
> bond-hash-basis: 0
> lb_output action: disabled, bond-id: -1
> updelay: 0 ms
> downdelay: 0 ms
> lacp_status: configured
> lacp_fallback_ab: false
> active-backup primary: 
> active member mac: 00:00:00:00:00:00(none)
> 
> member vnet74: disabled
>   may_enable: false
> 
> member vnet75: disabled
>   may_enable: false
> 
> member vnet76: disabled
>   may_enable: false
> 
> member vnet77: disabled
>   may_enable: false
> 
> member vnet78: disabled
>   may_enable: false
> 
> member vnet79: disabled
>   may_enable: false
> 
> I have configured bonding on an Ubuntu 22.04 at the other end with following 
> configuration:
> 
>   bonds:
> bond0:
>   addresses: [192.168.201.141/24]
>   interfaces:
> - enp2s0
> - enp3s0
> - enp4s0
> - enp5s0
> - enp6s0
> - enp7s0
>   parameters:
> mode: 802.3ad
> mii-monitor-interval: 100
> 
> And I get this error in dmesg: bond0: Warning: No 802.3ad response from the 
> link partner for any adapters in the bond
> 
> In Ubuntu the bond mode is 802.3ad, while in ovs its shows as active-backup 
> could that be the reason for bond not getting online?
> 
> I am kind of stuck here!

You can see the 

___
discuss mailing list
disc...@openvswitch.org
https://mail.openvswitch.org/mailman/listinfo/ovs-discuss

Re: [ovs-discuss] Bonding configuration for KVM guests

2023-10-12 Thread Xuo Guoto via discuss 
--- Original Message ---
On Thursday, October 12th, 2023 at 9:03 PM, Ilya Maximets  
wrote:

Thanks again for the detailed reply!

> If you want to preserve these, you'll need to re-add them manually.
> Potentially as part of the same transaction, by adding more commands,
> i.e.
>
> ovs-vsctl \
> del-port vnet75 \
> -- del-port vnet78 \

I had followed your advice and used the following command to create the bond

ovs-vsctl \
  del-port vnet75 \
   -- del-port vnet78 \
   -- del-port vnet79 \
   -- del-port vnet77 \
   -- del-port vnet74 \
   -- del-port vnet76 \
   -- add-bond ovsbr-lacp0 bond0 vnet75 vnet78 vnet79 vnet77 vnet74 vnet76 
lacp=active \
   -- set Interface vnet75 \
external_ids='{attached-mac="52:54:00:0d:61:fc", iface-id="", 
iface-status=active, vm-id=""}' \
   -- set Interface vnet78 \
external_ids='{attached-mac="52:54:00:24:36:a7", iface-id="", 
iface-status=active, vm-id=""}' \
   -- set Interface vnet79 \
external_ids='{attached-mac="52:54:00:c7:90:3c", iface-id="", 
iface-status=active, vm-id=""}' \
   -- set Interface vnet77 \
external_ids='{attached-mac="52:54:00:79:8e:a0", iface-id="", 
iface-status=active, vm-id=""}' \
   -- set Interface vnet74 \
external_ids='{attached-mac="52:54:00:aa:fb:bb", iface-id="", 
iface-status=active, vm-id=""}' \
   -- set Interface vnet76 \
external_ids='{attached-mac="52:54:00:d4:cc:88", iface-id="", 
iface-status=active, vm-id=""}'

Note: iface-id and vm-id values have been removed for brevity.

Now the command was successful and I could see the bond0 port with the 
specified interfaces

Bridge ovsbr-lacp0
Port bond0
Interface vnet78
Interface vnet75
Interface vnet79
Interface vnet74
Interface vnet77
Interface vnet76
Port ovsbr-lacp0
Interface ovsbr-lacp0
type: internal
Port vnet72
Interface vnet72

But the bonding is not working correctly

$ sudo ovs-appctl bond/show bond0
 bond0 
bond_mode: active-backup
bond may use recirculation: no, Recirc-ID : -1
bond-hash-basis: 0
lb_output action: disabled, bond-id: -1
updelay: 0 ms
downdelay: 0 ms
lacp_status: configured
lacp_fallback_ab: false
active-backup primary: 
active member mac: 00:00:00:00:00:00(none)

member vnet74: disabled
  may_enable: false

member vnet75: disabled
  may_enable: false

member vnet76: disabled
  may_enable: false

member vnet77: disabled
  may_enable: false

member vnet78: disabled
  may_enable: false

member vnet79: disabled
  may_enable: false

I have configured bonding on an Ubuntu 22.04 at the other end with following 
configuration:

  bonds:
bond0:
  addresses: [192.168.201.141/24]
  interfaces:
- enp2s0
- enp3s0
- enp4s0
- enp5s0
- enp6s0
- enp7s0
  parameters:
mode: 802.3ad
mii-monitor-interval: 100

And I get this error in dmesg: bond0: Warning: No 802.3ad response from the 
link partner for any adapters in the bond

In Ubuntu the bond mode is 802.3ad, while in ovs its shows as active-backup 
could that be the reason for bond not getting online?

I am kind of stuck here!

X.
___
discuss mailing list
disc...@openvswitch.org
https://mail.openvswitch.org/mailman/listinfo/ovs-discuss

Re: [ovs-discuss] Bonding configuration for KVM guests

2023-10-12 Thread Ilya Maximets via discuss 
On 10/12/23 15:39, Xuo Guoto via discuss wrote:
> Thanks for your reply!
> 
> --- Original Message ---
> On Thursday, October 12th, 2023 at 6:48 PM, Ilya Maximets 
>  wrote:
> 
> 
>> You currently have port vnet75 with an interface vnet75, and the
>> port vnet78 with interface vnet78.
> 
> The differentiation between port and interface was new to me! Thanks!

'ovs-vsctl show' command makes the structure a little easier to see.

> 
>> But you want a port bond0 with interfaces
>> vnet75 and vnet78. In order to achieve
>> that you need to remove ports vnet75 and
>> vnet78 first, and then create a port
>> bond0.
> 
> How can I remove ports vnet75 and vnet78 without touching the underlying 
> interface? I have the VM1 running and is connected to the bridge, can the 
> ports be removed without disturbing the VM?

It might be possible by manually crafting OVSDB transactions, but
it's very prone to errors and I would not recommend that.

I'd recommend removing ports and creating a bond in a single
transaction, by chaining operations with '--':

 ovs-vsctl del-port vnet75 \
-- del-port vnet78 \
-- add-bond ovsbr-lacp0 bond0 vnet75 vnet78 lacp=active

The problem with this though is that you will loose configuration
applied to the interfaces while removing them anyway.  For example,
these interfaces will no longer have their external_ids.
If you want to preserve these, you'll need to re-add them manually.
Potentially as part of the same transaction, by adding more commands,
i.e.

 ovs-vsctl \
  del-port vnet75 \
   -- del-port vnet78 \
   -- add-bond ovsbr-lacp0 bond0 vnet75 vnet78 lacp=active \
   -- set Interface vnet75 \
external_ids='{attached-mac="52:54:00:0d:61:fc", iface-id="..", 
iface-status=active, vm-id=".."}' \
some_other_column= \
   -- set Interface vnet78
external_ids='{attached-mac="52:54:00:24:36:a7", iface-id="..", 
iface-status=active, vm-id=".."}' \
   ...

You may still have some slight network disruption though.
However, since you're enabling LACP you will have disruption anyway
until you bond these interfaces and enable LACP inside VM as well.

Best regards, Ilya Maximets.
___
discuss mailing list
disc...@openvswitch.org
https://mail.openvswitch.org/mailman/listinfo/ovs-discuss

Re: [ovs-discuss] Bonding configuration for KVM guests

2023-10-12 Thread Xuo Guoto via discuss 
Thanks for your reply!

--- Original Message ---
On Thursday, October 12th, 2023 at 6:48 PM, Ilya Maximets  
wrote:


> You currently have port vnet75 with an interface vnet75, and the
> port vnet78 with interface vnet78.

The differentiation between port and interface was new to me! Thanks!

> But you want a port bond0 with interfaces
> vnet75 and vnet78. In order to achieve
> that you need to remove ports vnet75 and
> vnet78 first, and then create a port
> bond0.

How can I remove ports vnet75 and vnet78 without touching the underlying 
interface? I have the VM1 running and is connected to the bridge, can the ports 
be removed without disturbing the VM?

X.

___
discuss mailing list
disc...@openvswitch.org
https://mail.openvswitch.org/mailman/listinfo/ovs-discuss

Re: [ovs-discuss] Bonding configuration for KVM guests

2023-10-12 Thread Ilya Maximets via discuss 
On 10/12/23 14:58, Xuo Guoto via discuss wrote:
> Hello all,
> 
> I am trying to bond two interfaces connecting my running VM with OVS.
> 
> I have a bridge ovsbr-lacp0 with 7 virtio interfaces connected to two VMs, 6 
> in VM1 and one in VM2.
> 
> # ovs-vsctl list-ifaces ovsbr-lacp0
> vnet72
> vnet74
> vnet75
> vnet76
> vnet77
> vnet78
> vnet79
> 
> # ovs-vsctl -f table --columns=name,external_id list interface | grep 
> 274ce169-b166-45dc-ba1d-7faa8fea3673
> vnet75       {attached-mac="52:54:00:0d:61:fc", iface-id="..", 
> iface-status=active, vm-id=".."}
> vnet78       {attached-mac="52:54:00:24:36:a7", iface-id="..", 
> iface-status=active, vm-id=".."}
> vnet79       {attached-mac="52:54:00:c7:90:3c", iface-id="..", 
> iface-status=active, vm-id=".."}
> vnet77       {attached-mac="52:54:00:79:8e:a0", iface-id="..", 
> iface-status=active, vm-id=".."}
> vnet74       {attached-mac="52:54:00:aa:fb:bb", iface-id="..", 
> iface-status=active, vm-id=".."}
> vnet76       {attached-mac="52:54:00:d4:cc:88", iface-id="..", 
> iface-status=active, vm-id=".."}
> 
> Where 274ce169-b166-45dc-ba1d-7faa8fea3673 is the UUID of VM1
> 
> # sudo ovs-vsctl -f table --columns=name,external_id list interface | grep 
> 2075bff5-ee89-492e-b50c-36bd3b92b662
> vnet72       {attached-mac="52:54:00:fd:fe:f6", iface-id="..", 
> iface-status=active, vm-id=".."}
> 
> and 2075bff5-ee89-492e-b50c-36bd3b92b662 is the UUID of VM2. I have removed 
> the iface-id and vm-id so that the output is easy to read.
> 
> Now I want to configure bonding for all the interface in VM1.
> 
> # ovs-vsctl add-bond  ovsbr-lacp0 bond0 vnet75 vnet78 lacp=active
> ovs-vsctl: cannot create an interface named vnet75 because a port named 
> vnet75 already exists on bridge ovsbr-lacp0
> 
> # ovs-vsctl --may-exist add-bond  ovsbr-lacp0 bond0 vnet75 vnet78 lacp=active
> ovs-vsctl: cannot create an interface named vnet75 because a port named 
> vnet75 already exists on bridge ovsbr-lacp0
> 
> I am guessing I am doing some thing wrong, but not able to figure out what. 
> Any help to get this working will be much appreciated.

'add-bond' command creates a 'port' with two or more 'interfaces'.
These interfaces are created as part of the port creation.  So,
"add-bond  ovsbr-lacp0 bond0 vnet75 vnet78" creates a 'port' named
bond0, creates two interfaces vnet75 and vnet78, and adds them into
a port.  But interfaces vnet75 and vnet78 already exist, so the
command fails.  The 'may-exist' flag only check for existence of
the port, not interfaces in it.

You currently have port vnet75 with an interface vnet75, and the
port vnet78 with interface vnet78.  But you want a port bond0
with interfaces vnet75 and vnet78.  In order to achieve that you
need to remove ports vnet75 and vnet78 first, and then create
a port bond0.

HTH
Best regards, Ilya Maximets.
___
discuss mailing list
disc...@openvswitch.org
https://mail.openvswitch.org/mailman/listinfo/ovs-discuss

[ovs-discuss] Bonding configuration for KVM guests

2023-10-12 Thread Xuo Guoto via discuss 
Hello all,

I am trying to bond two interfaces connecting my running VM with OVS.

I have a bridge ovsbr-lacp0 with 7 virtio interfaces connected to two VMs, 6 in 
VM1 and one in VM2.

# ovs-vsctl list-ifaces ovsbr-lacp0
vnet72
vnet74
vnet75
vnet76
vnet77
vnet78
vnet79

# ovs-vsctl -f table --columns=name,external_id list interface | grep 
274ce169-b166-45dc-ba1d-7faa8fea3673
vnet75 {attached-mac="52:54:00:0d:61:fc", iface-id="..", iface-status=active, 
vm-id=".."}
vnet78 {attached-mac="52:54:00:24:36:a7", iface-id="..", iface-status=active, 
vm-id=".."}
vnet79 {attached-mac="52:54:00:c7:90:3c", iface-id="..", iface-status=active, 
vm-id=".."}
vnet77 {attached-mac="52:54:00:79:8e:a0", iface-id="..", iface-status=active, 
vm-id=".."}
vnet74 {attached-mac="52:54:00:aa:fb:bb", iface-id="..", iface-status=active, 
vm-id=".."}
vnet76 {attached-mac="52:54:00:d4:cc:88", iface-id="..", iface-status=active, 
vm-id=".."}

Where 274ce169-b166-45dc-ba1d-7faa8fea3673 is the UUID of VM1

# sudo ovs-vsctl -f table --columns=name,external_id list interface | grep 
2075bff5-ee89-492e-b50c-36bd3b92b662
vnet72 {attached-mac="52:54:00:fd:fe:f6", iface-id="..", iface-status=active, 
vm-id=".."}

and 2075bff5-ee89-492e-b50c-36bd3b92b662 is the UUID of VM2. I have removed the 
iface-id and vm-id so that the output is easy to read.

Now I want to configure bonding for all the interface in VM1.

# ovs-vsctl add-bond ovsbr-lacp0 bond0 vnet75 vnet78 lacp=active
ovs-vsctl: cannot create an interface named vnet75 because a port named vnet75 
already exists on bridge ovsbr-lacp0

# ovs-vsctl --may-exist add-bond ovsbr-lacp0 bond0 vnet75 vnet78 lacp=active
ovs-vsctl: cannot create an interface named vnet75 because a port named vnet75 
already exists on bridge ovsbr-lacp0

I am guessing I am doing some thing wrong, but not able to figure out what. Any 
help to get this working will be much appreciated.

X.___
discuss mailing list
disc...@openvswitch.org
https://mail.openvswitch.org/mailman/listinfo/ovs-discuss