Some more observations.
If I kill ovsdb-server first and then vswitchd, then things are working
fine.But, if vswitchd is killed first and then ovsdb-server, then bond's
members are coming up as disabled.
Is there something that is being carried over from previous instance or not
cleaned up properly?
-Thanks
Sent from Yahoo Mail. Get the app
From: "my_ovs_disc...@yahoo.com" <my_ovs_disc...@yahoo.com>
To: Ben Pfaff <b...@ovn.org>
Cc: "discuss@openvswitch.org" <discuss@openvswitch.org>
Sent: Thursday, September 15, 2016 5:54 PM
Subject: Re: [ovs-discuss] vswitchd restart and bond interfaces
Hi Ben, Thanks for the response.
I am able to reproduce this problem almost every time, but I had noticed that
one or two times it worked fine.I am not that worried about the MAC address
stuff right now.
port2 has valid mac address:
port2 Link encap:Ethernet HWaddr 00:00:46:FD:5E:59
UP BROADCAST RUNNING PROMISC MTU:1500 Metric:1
RX packets:0 errors:0 dropped:0 overruns:0 frame:0
TX packets:0 errors:0 dropped:0 overruns:0 carrier:0
collisions:0 txqueuelen:1000
RX bytes:0 (0 Bytes) TX bytes:0 (0 Bytes)
-Thanks
Sent from Yahoo Mail. Get the app
From: Ben Pfaff <b...@ovn.org>
To: my_ovs_disc...@yahoo.com
Cc: "discuss@openvswitch.org" <discuss@openvswitch.org>
Sent: Thursday, September 15, 2016 5:47 PM
Subject: Re: [ovs-discuss] vswitchd restart and bond interfaces
On Fri, Sep 16, 2016 at 12:02:57AM +0000, my_ovs_disc...@yahoo.com wrote:
> Hi,
> I am seeing this strange behavior of slaves getting disabled on bond
> interfaces upon restart of vswitchd.This bond is static LAG, no LACP.
>
> openvswitch-2.5.0 on Centos-6.2 based Linux
>
> This is the sequence I tried:
>
> - kill (vswitchd's pid)
> - kill (ovsdb-server's pid)
>
> - rm -f /usr/local/etc/openvswitch/conf.db
> - rm -f /usr/local/var/run/openvswitch/db.sock
> - ovsdb-tool create /usr/local/etc/openvswitch/conf.db
>/etc/vswitch.ovsschema
> - ovsdb-server --remote=punix:/usr/local/var/run/openvswitch/db.sock
>--remote=db:Open_vSwitch,Open_vSwitch,manager_options --pidfile
> - ovs-vsctl --no-wait emer-reset
> - ovs-vsctl --no-wait init
> - ovs-vswitchd --pidfile
> - ovs-vsctl add-br br0
> - ovs-vsctl set bridge br0 datapath_type=netdev
> - ovs-vsctl set Bridge br0 mcast_snooping_enable=true
> - ovs-vsctl set Bridge br0 other_config:mcast-snooping-table-size=8192
> - ovs-vsctl --may-exist add-bond br0 bond0 port1 port2
>bond_mode=balance-slb -- set port bond0 vlan_mode=trunk
>
> In case of regular bootup, only steps 1 and 2 won't be there.
> For regular fresh bootup case, I see
> ovs-appctl bond/show
> ---- bond0 ----
> bond_mode: balance-slb
> bond may use recirculation: no, Recirc-ID : -1
> bond-hash-basis: 0
> updelay: 0 ms
> downdelay: 0 ms
> next rebalance: 8524 ms
> lacp_status: off
> active slave mac: 00:00:00:00:00:00(port1)
>
> slave port1: enabled
> active slave
> may_enable: true
>
> slave port2: enabled
> may_enable: true
>
> But, if I follow steps 1-14, then I see that bond members are in disabled
> state:
> ovs-appctl bond/show
> ---- bond0 ----
> bond_mode: balance-slb
> bond may use recirculation: no, Recirc-ID : -1
> bond-hash-basis: 0
> updelay: 0 ms
> downdelay: 0 ms
> next rebalance: 6582 ms
> lacp_status: off
> active slave mac: 00:00:00:00:00:00(port2)
>
> slave port1: disabled
> may_enable: false
>
> slave port2: disabled
> may_enable: false
>
> Is there something that I am missing during restart?
I can't reproduce this problem in my own testing, from a VM, just now.
It's really weird that, in the case where there is an active slave, it
shows its MAC as all-zeros. Do port2 actually have an all-zeros MAC? I
assume not. There might be something weird even in the "working" case.
_______________________________________________
discuss mailing list
discuss@openvswitch.org
http://openvswitch.org/mailman/listinfo/discuss
