You are viewing a plain text version of this content. The canonical link for it is here.
Posted to users@cloudstack.apache.org by Milamber <mi...@apache.org> on 2013/04/11 19:19:28 UTC
CS4.1 Lost ping between VM and virtual router - arp issue?
Hello,
My setup : CS 4.1 (snapshot 2013/04/11)
Advanced network, 3 physical networks (public, guest, mngt/storage)
Guest network interfaces are connected on a switch L2, with trunk
enabled on vlan tagged 200-300 (and 199 for untagged packets)
Works fine:
Node 1 : VM + Virtual Router
Node 2 : nothing
I can ping VM since VR (on private network),
I can ping VR since VM (on private network),
I can ping VM on public IP.
Works with >80% ping loses
Node 1 : Virtual Router
Node 2 : VM
I can ping VM since VR (on private network) with alot of lost packets,
I can ping VR since VM (on private network) with a lot of lost packets,
I can ping VM on public IP with a lot of lost packets.
On VM, when I make some "arp -n" concurrent with ping VR, I found that
the arp address of VR changes!
ARP address 1 matches with p5p1 / p5p2 / bond2 / guestbr2 (on Node 1)
(note I use IP Bonding, but currently p5p2 isn't cable connected)
ARP address 2 matches with vnet4 (on Node 1)
(note this address is the mac address of VR/eth0)
Can you help me? this is a bug or misconfiguration?
Need add some kernel parameters on hosts related with arp?
Thanks.
Milamber
======
# brctl show
bridge name bridge id STP enabled interfaces
brbond2-234 8000.6805ca1182b2 no bond2.234
vnet4
cloud0 8000.fe00a9fe0333 no vnet0
vnet5
cloudbr0 8000.d4ae52e84796 no bond0
vnet2
vnet6
cloudbr1 8000.6805ca118352 no bond1
vnet1
vnet3
guestbr2 8000.6805ca1182b2 no bond2
virbr0 8000.5254008c282f yes virbr0-nic
=====
# ip a | grep -B1 82
8: p5p1: <BROADCAST,MULTICAST,SLAVE,UP,LOWER_UP> mtu 1500 qdisc mq
master bond2 state UP qlen 1000
link/ether 68:05:ca:11:82:b2 brd ff:ff:ff:ff:ff:ff
9: p5p2: <NO-CARRIER,BROADCAST,MULTICAST,SLAVE,UP> mtu 1500 qdisc mq
master bond2 state DOWN qlen 1000
link/ether 68:05:ca:11:82:b3 brd ff:ff:ff:ff:ff:ff
--
18: bond2: <BROADCAST,MULTICAST,MASTER,UP,LOWER_UP> mtu 1500 qdisc
noqueue state UP
link/ether 68:05:ca:11:82:b2 brd ff:ff:ff:ff:ff:ff
inet6 fe80::6a05:caff:fe11:82b2/64 scope link
--
19: guestbr2: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc noqueue
state UNKNOWN
link/ether 68:05:ca:11:82:b2 brd ff:ff:ff:ff:ff:ff
inet 172.17.7.13/24 brd 172.17.7.255 scope global guestbr2
inet6 fe80::6a05:caff:fe11:82b2/64 scope link
--
28: bond2.234@bond2: <BROADCAST,MULTICAST,MASTER,UP,LOWER_UP> mtu 1500
qdisc noqueue state UP
link/ether 68:05:ca:11:82:b2 brd ff:ff:ff:ff:ff:ff
inet6 fe80::6a05:caff:fe11:82b2/64 scope link
--
29: brbond2-234: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc
noqueue state UNKNOWN
link/ether 68:05:ca:11:82:b2 brd ff:ff:ff:ff:ff:ff
inet6 fe80::6a05:caff:fe11:82b2/64 scope link
====
# ip a | grep -B1 02
30: vnet4: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc pfifo_fast
state UNKNOWN qlen 500
link/ether fe:00:56:47:00:02 brd ff:ff:ff:ff:ff:ff
====
Re: CS4.1 Lost ping between VM and virtual router - arp issue?
Posted by Milamber <mi...@apache.org>.
Hello Chiradeep,
Le 11/04/2013 18:44, Chiradeep Vittal a ecrit :
> Sounds like a misconfiguration of the bond. I would try without the bond
> first.
Yeah! it's works
Exactly, that was a misconfiguration of the bond. I change bond mode
from "balance-alb" to "balance-rr" and it's works.
Thanks.
>
> On 4/11/13 10:19 AM, "Milamber"<mi...@apache.org> wrote:
>
>> Hello,
>>
>> My setup : CS 4.1 (snapshot 2013/04/11)
>> Advanced network, 3 physical networks (public, guest, mngt/storage)
>>
>> Guest network interfaces are connected on a switch L2, with trunk
>> enabled on vlan tagged 200-300 (and 199 for untagged packets)
>>
>> Works fine:
>>
>> Node 1 : VM + Virtual Router
>> Node 2 : nothing
>>
>> I can ping VM since VR (on private network),
>> I can ping VR since VM (on private network),
>> I can ping VM on public IP.
>>
>>
>> Works with>80% ping loses
>>
>> Node 1 : Virtual Router
>> Node 2 : VM
>> I can ping VM since VR (on private network) with alot of lost packets,
>> I can ping VR since VM (on private network) with a lot of lost packets,
>> I can ping VM on public IP with a lot of lost packets.
>>
>> On VM, when I make some "arp -n" concurrent with ping VR, I found that
>> the arp address of VR changes!
>>
>> ARP address 1 matches with p5p1 / p5p2 / bond2 / guestbr2 (on Node 1)
>> (note I use IP Bonding, but currently p5p2 isn't cable connected)
>>
>> ARP address 2 matches with vnet4 (on Node 1)
>> (note this address is the mac address of VR/eth0)
>>
>>
>> Can you help me? this is a bug or misconfiguration?
>> Need add some kernel parameters on hosts related with arp?
>>
>> Thanks.
>>
>>
>> Milamber
>>
>>
>>
>> ======
>>
>> # brctl show
>> bridge name bridge id STP enabled interfaces
>> brbond2-234 8000.6805ca1182b2 no bond2.234
>> vnet4
>> cloud0 8000.fe00a9fe0333 no vnet0
>> vnet5
>> cloudbr0 8000.d4ae52e84796 no bond0
>> vnet2
>> vnet6
>> cloudbr1 8000.6805ca118352 no bond1
>> vnet1
>> vnet3
>> guestbr2 8000.6805ca1182b2 no bond2
>> virbr0 8000.5254008c282f yes virbr0-nic
>>
>>
>> =====
>>
>> # ip a | grep -B1 82
>> 8: p5p1:<BROADCAST,MULTICAST,SLAVE,UP,LOWER_UP> mtu 1500 qdisc mq
>> master bond2 state UP qlen 1000
>> link/ether 68:05:ca:11:82:b2 brd ff:ff:ff:ff:ff:ff
>> 9: p5p2:<NO-CARRIER,BROADCAST,MULTICAST,SLAVE,UP> mtu 1500 qdisc mq
>> master bond2 state DOWN qlen 1000
>> link/ether 68:05:ca:11:82:b3 brd ff:ff:ff:ff:ff:ff
>> --
>> 18: bond2:<BROADCAST,MULTICAST,MASTER,UP,LOWER_UP> mtu 1500 qdisc
>> noqueue state UP
>> link/ether 68:05:ca:11:82:b2 brd ff:ff:ff:ff:ff:ff
>> inet6 fe80::6a05:caff:fe11:82b2/64 scope link
>> --
>> 19: guestbr2:<BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc noqueue
>> state UNKNOWN
>> link/ether 68:05:ca:11:82:b2 brd ff:ff:ff:ff:ff:ff
>> inet 172.17.7.13/24 brd 172.17.7.255 scope global guestbr2
>> inet6 fe80::6a05:caff:fe11:82b2/64 scope link
>> --
>> 28: bond2.234@bond2:<BROADCAST,MULTICAST,MASTER,UP,LOWER_UP> mtu 1500
>> qdisc noqueue state UP
>> link/ether 68:05:ca:11:82:b2 brd ff:ff:ff:ff:ff:ff
>> inet6 fe80::6a05:caff:fe11:82b2/64 scope link
>> --
>> 29: brbond2-234:<BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc
>> noqueue state UNKNOWN
>> link/ether 68:05:ca:11:82:b2 brd ff:ff:ff:ff:ff:ff
>> inet6 fe80::6a05:caff:fe11:82b2/64 scope link
>>
>> ====
>> # ip a | grep -B1 02
>> 30: vnet4:<BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc pfifo_fast
>> state UNKNOWN qlen 500
>> link/ether fe:00:56:47:00:02 brd ff:ff:ff:ff:ff:ff
>>
>>
>> ====
>>
>>
>
Re: CS4.1 Lost ping between VM and virtual router - arp issue?
Posted by Chiradeep Vittal <Ch...@citrix.com>.
Sounds like a misconfiguration of the bond. I would try without the bond
first.
On 4/11/13 10:19 AM, "Milamber" <mi...@apache.org> wrote:
>Hello,
>
>My setup : CS 4.1 (snapshot 2013/04/11)
>Advanced network, 3 physical networks (public, guest, mngt/storage)
>
>Guest network interfaces are connected on a switch L2, with trunk
>enabled on vlan tagged 200-300 (and 199 for untagged packets)
>
>Works fine:
>
>Node 1 : VM + Virtual Router
>Node 2 : nothing
>
>I can ping VM since VR (on private network),
>I can ping VR since VM (on private network),
>I can ping VM on public IP.
>
>
>Works with >80% ping loses
>
>Node 1 : Virtual Router
>Node 2 : VM
>I can ping VM since VR (on private network) with alot of lost packets,
>I can ping VR since VM (on private network) with a lot of lost packets,
>I can ping VM on public IP with a lot of lost packets.
>
>On VM, when I make some "arp -n" concurrent with ping VR, I found that
>the arp address of VR changes!
>
>ARP address 1 matches with p5p1 / p5p2 / bond2 / guestbr2 (on Node 1)
>(note I use IP Bonding, but currently p5p2 isn't cable connected)
>
>ARP address 2 matches with vnet4 (on Node 1)
>(note this address is the mac address of VR/eth0)
>
>
>Can you help me? this is a bug or misconfiguration?
>Need add some kernel parameters on hosts related with arp?
>
>Thanks.
>
>
>Milamber
>
>
>
>======
>
># brctl show
>bridge name bridge id STP enabled interfaces
>brbond2-234 8000.6805ca1182b2 no bond2.234
> vnet4
>cloud0 8000.fe00a9fe0333 no vnet0
> vnet5
>cloudbr0 8000.d4ae52e84796 no bond0
> vnet2
> vnet6
>cloudbr1 8000.6805ca118352 no bond1
> vnet1
> vnet3
>guestbr2 8000.6805ca1182b2 no bond2
>virbr0 8000.5254008c282f yes virbr0-nic
>
>
>=====
>
># ip a | grep -B1 82
>8: p5p1: <BROADCAST,MULTICAST,SLAVE,UP,LOWER_UP> mtu 1500 qdisc mq
>master bond2 state UP qlen 1000
> link/ether 68:05:ca:11:82:b2 brd ff:ff:ff:ff:ff:ff
>9: p5p2: <NO-CARRIER,BROADCAST,MULTICAST,SLAVE,UP> mtu 1500 qdisc mq
>master bond2 state DOWN qlen 1000
> link/ether 68:05:ca:11:82:b3 brd ff:ff:ff:ff:ff:ff
>--
>18: bond2: <BROADCAST,MULTICAST,MASTER,UP,LOWER_UP> mtu 1500 qdisc
>noqueue state UP
> link/ether 68:05:ca:11:82:b2 brd ff:ff:ff:ff:ff:ff
> inet6 fe80::6a05:caff:fe11:82b2/64 scope link
>--
>19: guestbr2: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc noqueue
>state UNKNOWN
> link/ether 68:05:ca:11:82:b2 brd ff:ff:ff:ff:ff:ff
> inet 172.17.7.13/24 brd 172.17.7.255 scope global guestbr2
> inet6 fe80::6a05:caff:fe11:82b2/64 scope link
>--
>28: bond2.234@bond2: <BROADCAST,MULTICAST,MASTER,UP,LOWER_UP> mtu 1500
>qdisc noqueue state UP
> link/ether 68:05:ca:11:82:b2 brd ff:ff:ff:ff:ff:ff
> inet6 fe80::6a05:caff:fe11:82b2/64 scope link
>--
>29: brbond2-234: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc
>noqueue state UNKNOWN
> link/ether 68:05:ca:11:82:b2 brd ff:ff:ff:ff:ff:ff
> inet6 fe80::6a05:caff:fe11:82b2/64 scope link
>
>====
># ip a | grep -B1 02
>30: vnet4: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc pfifo_fast
>state UNKNOWN qlen 500
> link/ether fe:00:56:47:00:02 brd ff:ff:ff:ff:ff:ff
>
>
>====
>
>