Skip site navigation (1)Skip section navigation (2)
Date:      Fri, 4 Apr 2014 11:51:28 +0200
From:      mxb <mxb@alumni.chalmers.se>
To:        freebsd-net@freebsd.org
Cc:        freebsd-pf@freebsd.org
Subject:   Re: LACP lagg and CARP - ENETDOWN (was: FreeBSD 10-STABLE and CARP states)
Message-ID:  <F8BC8FA8-B028-453C-97EA-F99593EADAA7@alumni.chalmers.se>
In-Reply-To: <1E20234E-4F81-4BA1-BA57-FADD80F782F4@alumni.chalmers.se>
References:  <EDCF2D3E-D785-4DCA-8033-09D547BDCC93@alumni.chalmers.se> <4A818132-757F-4BAD-8137-CDB1F6F0681C@alumni.chalmers.se> <B01BB972-3B67-4859-A96F-6CD2E14CEE97@alumni.chalmers.se> <1E20234E-4F81-4BA1-BA57-FADD80F782F4@alumni.chalmers.se>

next in thread | previous in thread | raw e-mail | index | archive | help

According my own research around this problem
ip_output() at line 839 of ip_carp.c returns ENETDOWN then lagg is =
configured in LACP mode.

On 2 apr 2014, at 15:13, mxb <mxb@alumni.chalmers.se> wrote:

>=20
> OK, thanks everyone whom replayed. E.g. NONE.
>=20
> The problem seems to be related to LACP trunking.
> Disabling LACP and configuring trunk in =91loadbalance=92 mode puts =
all in desired state (even after reboot).
>=20
> lagg0: flags=3D8943<UP,BROADCAST,RUNNING,PROMISC,SIMPLEX,MULTICAST> =
metric 0 mtu 9000
> 	=
options=3D8407bb<RXCSUM,TXCSUM,VLAN_MTU,VLAN_HWTAGGING,JUMBO_MTU,VLAN_HWCS=
UM,TSO4,TSO6,LRO,VLAN_HWTSO>
> 	ether 00:25:90:e3:71:f2
> 	inet 172.16.0.234 netmask 0xfffff800 broadcast 172.16.7.255
> 	inet6 fe80::225:90ff:fee3:71f2%lagg0 prefixlen 64 scopeid 0x5
> 	inet 172.16.0.231 netmask 0xfffff800 broadcast 172.16.7.255 vhid =
201
> 	inet 172.16.0.233 netmask 0xfffff800 broadcast 172.16.7.255 vhid =
202
> 	nd6 options=3D29<PERFORMNUD,IFDISABLED,AUTO_LINKLOCAL>
> 	media: Ethernet autoselect
> 	status: active
> 	carp: MASTER vhid 201 advbase 1 advskew 1
> 	carp: BACKUP vhid 202 advbase 5 advskew 100
> 	laggproto loadbalance lagghash l2,l3,l4
> 	laggport: ix1 flags=3D4<ACTIVE>
> 	laggport: ix0 flags=3D4<ACTIVE>
> vlan2: flags=3D8943<UP,BROADCAST,RUNNING,PROMISC,SIMPLEX,MULTICAST> =
metric 0 mtu 9000
> 	options=3D303<RXCSUM,TXCSUM,TSO4,TSO6>
> 	ether 00:25:90:e3:71:f2
> 	inet 10.11.11.201 netmask 0xffffff00 broadcast 10.11.11.255
> 	inet6 fe80::225:90ff:fee3:71f2%vlan2 prefixlen 64 scopeid 0x6
> 	inet 10.11.12.203 netmask 0xffffff00 broadcast 10.11.12.255 vhid =
12
> 	nd6 options=3D29<PERFORMNUD,IFDISABLED,AUTO_LINKLOCAL>
> 	media: Ethernet autoselect
> 	status: active
> 	vlan: 2 parent interface: lagg0
> 	carp: BACKUP vhid 12 advbase 1 advskew 100
>=20
> //mxb
>=20
> On 2 apr 2014, at 09:35, mxb <mxb@alumni.chalmers.se> wrote:
>=20
>>=20
>> Moving this to freebsd-pf.
>>=20
>> On 31 mar 2014, at 22:21, mxb <mxb@alumni.chalmers.se> wrote:
>>=20
>>>=20
>>> Manually setting net.inet.carp.demotion brought BOTH VHIDs in =
desired state.
>>> pfsync bulk update seems to not put everything back as it should.
>>>=20
>>> lagg0: flags=3D8943<UP,BROADCAST,RUNNING,PROMISC,SIMPLEX,MULTICAST> =
metric 0 mtu 9000
>>> 	=
options=3D8407bb<RXCSUM,TXCSUM,VLAN_MTU,VLAN_HWTAGGING,JUMBO_MTU,VLAN_HWCS=
UM,TSO4,TSO6,LRO,VLAN_HWTSO>
>>> 	ether 00:25:90:e3:71:f2
>>> 	inet 172.16.0.234 netmask 0xfffff800 broadcast 172.16.7.255
>>> 	inet6 fe80::225:90ff:fee3:71f2%lagg0 prefixlen 64 scopeid 0x5
>>> 	inet 172.16.0.231 netmask 0xfffff800 broadcast 172.16.7.255 vhid =
201
>>> 	inet 172.16.0.233 netmask 0xfffff800 broadcast 172.16.7.255 vhid =
202
>>> 	nd6 options=3D29<PERFORMNUD,IFDISABLED,AUTO_LINKLOCAL>
>>> 	media: Ethernet autoselect
>>> 	status: active
>>> 	carp: MASTER vhid 201 advbase 1 advskew 1
>>> 	carp: BACKUP vhid 202 advbase 5 advskew 100
>>> 	laggproto lacp lagghash l2,l3,l4
>>> 	laggport: ix1 flags=3D1c<ACTIVE,COLLECTING,DISTRIBUTING>
>>> 	laggport: ix0 flags=3D1c<ACTIVE,COLLECTING,DISTRIBUTING>
>>>=20
>>>=20
>>> On 31 mar 2014, at 20:42, mxb <mxb@alumni.chalmers.se> wrote:
>>>=20
>>>>=20
>>>> Hi list,
>>>>=20
>>>> hopefully this is the right place to have my question regarding =
CARP on 10-STABLE.
>>>>=20
>>>> I have two nodes with following setup(node1):
>>>>=20
>>>> lagg0: flags=3D8943<UP,BROADCAST,RUNNING,PROMISC,SIMPLEX,MULTICAST> =
metric 0 mtu 9000
>>>> 	=
options=3D8407bb<RXCSUM,TXCSUM,VLAN_MTU,VLAN_HWTAGGING,JUMBO_MTU,VLAN_HWCS=
UM,TSO4,TSO6,LRO,VLAN_HWTSO>
>>>> 	ether 00:25:90:e3:71:f2
>>>> 	inet 172.16.0.234 netmask 0xfffff800 broadcast 172.16.7.255
>>>> 	inet6 fe80::225:90ff:fee3:71f2%lagg0 prefixlen 64 scopeid 0x5
>>>> 	inet 172.16.0.231 netmask 0xfffff800 broadcast 172.16.7.255 vhid =
201
>>>> 	inet 172.16.0.233 netmask 0xfffff800 broadcast 172.16.7.255 vhid =
202
>>>> 	nd6 options=3D29<PERFORMNUD,IFDISABLED,AUTO_LINKLOCAL>
>>>> 	media: Ethernet autoselect
>>>> 	status: active
>>>> 	carp: BACKUP vhid 201 advbase 1 advskew 1
>>>> 	carp: BACKUP vhid 202 advbase 5 advskew 100
>>>> 	laggproto lacp lagghash l2,l3,l4
>>>> 	laggport: ix1 flags=3D1c<ACTIVE,COLLECTING,DISTRIBUTING>
>>>> 	laggport: ix0 flags=3D1c<ACTIVE,COLLECTING,DISTRIBUTING>
>>>>=20
>>>> net.inet.carp.preempt=3D1 on both nodes. as well as PSYNC as this:
>>>>=20
>>>> pfsync0: flags=3D41<UP,RUNNING> metric 0 mtu 1500
>>>> 	pfsync: syncdev: vlan22 syncpeer: 10.22.22.2 maxupd: 128 defer: =
off
>>>>=20
>>>> The problem is (if it is not clear from the ifconfig-output for the =
lagg0) the state of VHID 201.
>>>> Node2 with advskew of 100 is currently MASTER, but it SHOULD NOT as =
of setup.
>>>>=20
>>>> Am I hitting a bug or doing something wrong?
>>>>=20
>>>> I also have noted that after the pfsync bulk update the demotion =
counter never setts to 0, but stays on 480,
>>>> thus preventing node1 become a MASTER 201(?). Or is this a normal =
behavior?
>>>>=20
>>>> Regards,
>>>> mxb
>>>>=20
>>>>=20
>>>=20
>>=20
>=20




Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?F8BC8FA8-B028-453C-97EA-F99593EADAA7>