Re: [ovs-dev] [PATCH 2/2] bond: Sending learning packets on active-backup.
I am testing OVS bonding in active-backup mode (eth1 and eth2). However I am
seeing packets being TX from both eth1 and eth2. That is not what I would
expect. Any help would be appreciated.
[root@kvm1 network-scripts]# ovs-vswitchd --version
ovs-vswitchd (Open vSwitch) 1.9.3
Compiled Sep 14 2014 22:35:27
OpenFlow versions 0x1:0x1
[root@kvm1 network-scripts]#
[root@kvm1 network-scripts]# ovs-vsctl show
c7487bd2-b382-4d08-bba7-6077117294d7
Bridge "br0"
Port "br0"
Interface "br0"
type: internal
Port "vlan101"
tag: 101
Interface "vlan101"
type: internal
Port "vlan100"
tag: 100
Interface "vlan100"
type: internal
Port "bond0"
Interface "eth1"
Interface "eth2"
ovs_version: "1.9.3"
[root@kvm1 network-scripts]# ovs-vsctl list port bond0
_uuid : e44904c0-4a9f-413f-989a-e67fedc93b26
bond_downdelay : 0
bond_fake_iface : false
bond_mode : active-backup
bond_updelay: 0
external_ids: {}
fake_bridge : false
interfaces : [2735efa7-600d-4de5-bf00-d2459b8e9615,
759f88ef-dc2c-44bd-a2b2-0d991eb9fd05]
lacp: []
mac : []
name: "bond0"
other_config: {}
qos : []
statistics : {}
status : {}
tag : []
trunks : []
vlan_mode : []
[root@kvm1 network-scripts]# tcpdump -i eth1
tcpdump: WARNING: eth1: no IPv4 address assigned
tcpdump: verbose output suppressed, use -v or -vv for full protocol decode
listening on eth1, link-type EN10MB (Ethernet), capture size 65535 bytes
21:42:31.142112 IP 100.100.100.100 > 100.100.100.1: ICMP echo request, id 5182,
seq 1, length 64
21:42:31.142394 IP 100.100.100.1 > 100.100.100.100: ICMP echo reply, id 5182,
seq 1, length 64
21:42:32.542722 IP 192.168.116.1.17500 > 192.168.116.255.17500: UDP, length 146
^C
3 packets captured
3 packets received by filter
0 packets dropped by kernel
[root@kvm1 network-scripts]# tcpdump -i eth2
tcpdump: WARNING: eth2: no IPv4 address assigned
tcpdump: verbose output suppressed, use -v or -vv for full protocol decode
listening on eth2, link-type EN10MB (Ethernet), capture size 65535 bytes
21:42:36.141611 ARP, Request who-has 100.100.100.1 tell 100.100.100.100, length
28
21:42:36.142585 ARP, Reply 100.100.100.1 is-at 00:0c:29:5a:8f:82 (oui Unknown),
length 46
21:42:36.741367 IP 100.100.100.100 > 100.100.100.1: ICMP echo request, id 5950,
seq 1, length 64
21:42:36.741889 IP 100.100.100.1 > 100.100.100.100: ICMP echo reply, id 5950,
seq 1, length 64
^C
4 packets captured
4 packets received by filter
0 packets dropped by kernel
[root@kvm1 network-scripts]#
___
dev mailing list
dev@openvswitch.org
http://openvswitch.org/mailman/listinfo/dev
Re: [ovs-dev] [PATCH 2/2] bond: Sending learning packets on active-backup.
The test traffic was sent to bond0. Traffic appears to be received from active eth1 and backup eth2. This means the backup eth2 responded to arp which is not expected in the active-backup mode. Sent from my iPhone > On Sep 15, 2014, at 10:04 PM, Ben Pfaff wrote: > > Why are you sending your question as a reply to a thread from 2012? > >> On Mon, Sep 15, 2014 at 09:50:42PM -0500, Yi Zhang wrote: >> I am testing OVS bonding in active-backup mode (eth1 and eth2). >> However I am seeing packets being TX from both eth1 and eth2. That is >> not what I would expect. Any help would be appreciated. > > I suspect that your tcpdumps are showing a packet going out from eth1 > and then coming back, through the first-hop switch, on eth2 (or vice > versa). I don't know how to tell whether this is happening from tcpdump > output. ___ dev mailing list dev@openvswitch.org http://openvswitch.org/mailman/listinfo/dev
Re: [ovs-dev] [PATCH 2/2] bond: Sending learning packets on active-backup.
Here is my configuration: ovs-vsctl add-br br0 ovs-vsctl add-port br0 vlan100 tag=100 -- set interface vlan100 type=internal ovs-vsctl add-port br0 vlan101 tag=101 -- set interface vlan101 type=internal ovs-vsctl add-bond br0 bond0 eth1 eth2 Sent from my iPhone > On Sep 15, 2014, at 10:04 PM, Ben Pfaff wrote: > > Why are you sending your question as a reply to a thread from 2012? > >> On Mon, Sep 15, 2014 at 09:50:42PM -0500, Yi Zhang wrote: >> I am testing OVS bonding in active-backup mode (eth1 and eth2). >> However I am seeing packets being TX from both eth1 and eth2. That is >> not what I would expect. Any help would be appreciated. > > I suspect that your tcpdumps are showing a packet going out from eth1 > and then coming back, through the first-hop switch, on eth2 (or vice > versa). I don't know how to tell whether this is happening from tcpdump > output. ___ dev mailing list dev@openvswitch.org http://openvswitch.org/mailman/listinfo/dev
Re: [ovs-dev] [PATCH 2/2] bond: Sending learning packets on active-backup.
I added a bond interface consisted of eth1 and eth2 in active-backup mode. Test traffic was sent to bond. Sent from my iPhone > On Sep 15, 2014, at 10:39 PM, Ben Pfaff wrote: > > Please explain the scenario better. > > Is the test traffic received on eth1 or eth2? Or is received on some > other interface and being sent out bond0? > >> On Mon, Sep 15, 2014 at 10:36:43PM -0500, Yi Zhang wrote: >> The test traffic was sent to bond0. Traffic appears to be received >> from active eth1 and backup eth2. This means the backup eth2 >> responded to arp which is not expected in the active-backup mode. >> >> Sent from my iPhone >> >>> On Sep 15, 2014, at 10:04 PM, Ben Pfaff wrote: >>> >>> Why are you sending your question as a reply to a thread from 2012? >>> >>>> On Mon, Sep 15, 2014 at 09:50:42PM -0500, Yi Zhang wrote: >>>> I am testing OVS bonding in active-backup mode (eth1 and eth2). >>>> However I am seeing packets being TX from both eth1 and eth2. That is >>>> not what I would expect. Any help would be appreciated. >>> >>> I suspect that your tcpdumps are showing a packet going out from eth1 >>> and then coming back, through the first-hop switch, on eth2 (or vice >>> versa). I don't know how to tell whether this is happening from tcpdump >>> output. ___ dev mailing list dev@openvswitch.org http://openvswitch.org/mailman/listinfo/dev
Re: [ovs-dev] [PATCH 2/2] bond: Sending learning packets on active-backup.
Host (eth1 and eth2) <=== external-switch <--- test.traffic Sent from my iPhone > On Sep 15, 2014, at 10:58 PM, Ben Pfaff wrote: > > Where does the traffic forwarded to bond0 come from, and how is it > forwarded to the bond? > >> On Mon, Sep 15, 2014 at 10:44:50PM -0500, Yi Zhang wrote: >> I added a bond interface consisted of eth1 and eth2 in active-backup mode. >> >> Test traffic was sent to bond. >> >> Sent from my iPhone >> >>> On Sep 15, 2014, at 10:39 PM, Ben Pfaff wrote: >>> >>> Please explain the scenario better. >>> >>> Is the test traffic received on eth1 or eth2? Or is received on some >>> other interface and being sent out bond0? >>> >>>> On Mon, Sep 15, 2014 at 10:36:43PM -0500, Yi Zhang wrote: >>>> The test traffic was sent to bond0. Traffic appears to be received >>>> from active eth1 and backup eth2. This means the backup eth2 >>>> responded to arp which is not expected in the active-backup mode. >>>> >>>> Sent from my iPhone >>>> >>>>> On Sep 15, 2014, at 10:04 PM, Ben Pfaff wrote: >>>>> >>>>> Why are you sending your question as a reply to a thread from 2012? >>>>> >>>>>> On Mon, Sep 15, 2014 at 09:50:42PM -0500, Yi Zhang wrote: >>>>>> I am testing OVS bonding in active-backup mode (eth1 and eth2). >>>>>> However I am seeing packets being TX from both eth1 and eth2. That is >>>>>> not what I would expect. Any help would be appreciated. >>>>> >>>>> I suspect that your tcpdumps are showing a packet going out from eth1 >>>>> and then coming back, through the first-hop switch, on eth2 (or vice >>>>> versa). I don't know how to tell whether this is happening from tcpdump >>>>> output. ___ dev mailing list dev@openvswitch.org http://openvswitch.org/mailman/listinfo/dev
[ovs-dev] Valid use case?
I have a 2x NIC (eth0 & eth1) KVM topology that utilize OVS. We need OVS bridge to use active-standby bonding mode for the VLANs to the VMs. And on top of that we want to use dot1q vlan interfaces (like eth0.10 and eth1.11) for iSCSI storage traffic. I know this setup is supported in the legacy vmware environment, is this supported in the OVS case? Thanks! Yi ___ dev mailing list dev@openvswitch.org http://openvswitch.org/mailman/listinfo/dev
