[Kea-users] Running perfDHCP with interface in a bridge

Tue Jun 2 19:09:13 UTC 2015

Hi Marcin,

Hope you are doing good, and hope you enjoyed your conference!

Thanks for your reply! Very sorry for my reply.

Just wanted to let you know that perfdhcp is working perfectly fine for me :)

I am running the following command.

sudo ip netns exec test-namespace /usr/local/sbin/perfdhcp -4 -b mac=fa:16:00:00:00:00 -r 50 -R 50 -p 120 -l tapfce197b0-d5

Thanks a lot for all your help!

Regards,
Vikram Hosakote
OpenStack Software Engineer    |    vhosakot at cisco.com
Cloud and Virtualization Group  |    Cisco System

From: Marcin Siodelski <marcin at isc.org<mailto:marcin at isc.org>>
Date: Tuesday, March 24, 2015 at 5:40 PM
To: Vikram Hosakote <vhosakot at cisco.com<mailto:vhosakot at cisco.com>>, "kea-users at lists.isc.org<mailto:kea-users at lists.isc.org>" <kea-users at lists.isc.org<mailto:kea-users at lists.isc.org>>
Subject: Re: [Kea-users] Running perfDHCP with interface in a bridge

Hi Vikram,

I am currently on the conference and I will not have much time to focus
on that issue during this week.

I just had quick reading of it. From the server's standpoint the
perfdhcp acts as a relay agent and it works on the same port 67 as the
server. Binding both perfdhcp and the server to the same port on the
same interface results in a conflict and I presume this is why the
perfdhcp refuses to bind. It would be actually good if you could turn
the server off and rerun perfdhcp and see if it overcomes the "bind"
error. Obviously the perfdhcp will get no responses because the server
is down, but it would be a proof for me that this is a conflict between
the server and perfdhcp being bound to the same port.

Once you use the -L parameter and set a different local port for
perfdhcp there is no conflict anymore and it seems to work. But, in this
case the server receives the queries from perfdhcp from a different port
than standard DHCP port, in which case the server may either drop them
or may respond to port 67 which perfdhcp is not listening on.

Do you have any packet capture to see if the server is actually
responding? Do you know if dnsmasq responds to the port other than 67 or
68, if it receives the message from such port?

>From the reading of
http://www.thekelleys.org.uk/dnsmasq/docs/dnsmasq-man.html

it may be possible to change the default ports using the
"--dhcp-alternate-port" parameter.

Marcin

On 2015-03-23 20:56, Vikram Hosakote (vhosakot) wrote:
Hi Marcin,

Hope you are doing good!

I have two tap interfaces inside a namespace, and the DHCP server (dnsmasq)
30.30.30.3 is running on tap05f30e85-d2 listening on the DHCP server UDP
port 67.

Below are the IP addresses of the two tap interfaces inside the namespace
qdhcp-9a5bcbe7-7f29-4eab-9296-207ea3a6b1be.

[root at controller DHCP_CPNR_dev(keystone_admin)]# sudo ip netns exec
qdhcp-9a5bcbe7-7f29-4eab-9296-207ea3a6b1be ip a
21: tap05f30e85-d2: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc
noqueue state UNKNOWN
      link/ether fa:16:3e:25:71:de brd ff:ff:ff:ff:ff:ff
      inet 30.30.30.3/24 brd 30.30.30.255 scope global tap05f30e85-d2
      inet6 fe80::f816:3eff:fe25:71de/64 scope link
22: tap883a32b1-77: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc
noqueue state UNKNOWN
      link/ether fa:16:3e:6e:90:e7 brd ff:ff:ff:ff:ff:ff
      inet 30.30.30.101/24 brd 30.30.30.255 scope global tap883a32b1-77

[root at controller DHCP_CPNR_dev(keystone_admin)]# sudo ip netns exec
qdhcp-9a5bcbe7-7f29-4eab-9296-207ea3a6b1be ifconfig -a
tap05f30e85-d2: flags=4163<UP,BROADCAST,RUNNING,MULTICAST>  mtu 1500
          inet 30.30.30.3  netmask 255.255.255.0  broadcast 30.30.30.255
          inet6 fe80::f816:3eff:fe25:71de  prefixlen 64  scopeid 0x20<link>
          ether fa:16:3e:25:71:de  txqueuelen 0  (Ethernet)
          RX packets 105  bytes 27577 (26.9 KiB)
          RX errors 0  dropped 0  overruns 0  frame 0
          TX packets 42  bytes 6301 (6.1 KiB)
          TX errors 0  dropped 0 overruns 0  carrier 0  collisions 0
tap883a32b1-77: flags=4163<UP,BROADCAST,RUNNING,MULTICAST>  mtu 1500
          inet 30.30.30.101  netmask 255.255.255.0  broadcast 30.30.30.255
          ether fa:16:3e:6e:90:e7  txqueuelen 0  (Ethernet)
          RX packets 36  bytes 5793 (5.6 KiB)
          RX errors 0  dropped 0  overruns 0  frame 0
          TX packets 110  bytes 27995 (27.3 KiB)
          TX errors 0  dropped 0 overruns 0  carrier 0  collisions 0

Below is the DHCP server (dnsmasq) 30.30.30.3 running on tap05f30e85-d2
listening on the DHCP server UDP port 67.

[root at controller DHCP_CPNR_dev(keystone_admin)]# sudo ip netns exec
qdhcp-9a5bcbe7-7f29-4eab-9296-207ea3a6b1be netstat -pan | grep 67
udp        0      0 0.0.0.0:67              0.0.0.0:*
          18837/dnsmasq

I ran perfdhcp inside the namespace
qdhcp-9a5bcbe7-7f29-4eab-9296-207ea3a6b1be
to simulate DHCP clients in order to receive DHCP responses from the
DHCP server
(30.30.30.3 running on tap05f30e85-d2 listening on the DHCP server UDP
port 67).

But, I keep seeing the errors below. When I use the non-default local
DHCP port
(-L 10067), I don’t see any errors, but I don’t see any responses from
the DHCP server (dnsmasq).

Could you please let me know what I am missing. I am using the latest
version of
perfdhcp (0.9.1-beta). Thanks a lot!

********************************

[root at controller DHCP_CPNR_dev(keystone_admin)]# sudo ip netns exec
qdhcp-9a5bcbe7-7f29-4eab-9296-207ea3a6b1be /usr/local/sbin/perfdhcp -x a
-4 -b mac=fa:16:3e:6e:90:e7 -r 1 30.30.30.3
Running: perfdhcp -x a -4 -b mac=fa:16:3e:6e:90:e7 -r 1 30.30.30.3
IPv4
lease-type=address-only (IA_NA option added to the client's request)
rate[1/s]=1
base[0]=mac=fa:16:3e:6e:90:e7
drop-time[0]=1
drop-time[1]=1
aggressivity=1
elp-offset=-1
sid-offset=-1
rip-offset=-1
diagnostic-selectors=a
server=30.30.30.3
Set MAC to fa::16::3e::6e::90::e7
Set DUID to 000100011ca3170a000c01020304
Error running perfdhcp: Failed to bind socket 3 to 30.30.30.3/port=67

********************************

[root at controller DHCP_CPNR_dev(keystone_admin)]# sudo ip netns exec
qdhcp-9a5bcbe7-7f29-4eab-9296-207ea3a6b1be /usr/local/sbin/perfdhcp -x a
-4 -b mac=fa:16:3e:6e:90:e7 -r 1 -l tap883a32b1-77 30.30.30.3
Running: perfdhcp -x a -4 -b mac=fa:16:3e:6e:90:e7 -r 1 -l
tap883a32b1-77 30.30.30.3
IPv4
lease-type=address-only (IA_NA option added to the client's request)
rate[1/s]=1
base[0]=mac=fa:16:3e:6e:90:e7
drop-time[0]=1
drop-time[1]=1
aggressivity=1
elp-offset=-1
sid-offset=-1
rip-offset=-1
diagnostic-selectors=a
interface=tap883a32b1-77
server=30.30.30.3
Set MAC to fa::16::3e::6e::90::e7
Set DUID to 000100011ca3171a000c01020304
Error running perfdhcp: Failed to bind socket 3 to 30.30.30.101/port=67

********************************

The following works but I see no responses from the DHCP server
30.30.30.3 (dnsmasq).

[root at controller DHCP_CPNR_dev(keystone_admin)]# sudo ip netns exec
qdhcp-9a5bcbe7-7f29-4eab-9296-207ea3a6b1be /usr/local/sbin/perfdhcp -x a
-4 -b mac=fa:16:3e:6e:90:e7 -r 1 -l tap883a32b1-77 -L 10067 30.30.30.3
Running: perfdhcp -x a -4 -b mac=fa:16:3e:6e:90:e7 -r 1 -l
tap883a32b1-77 -L 10067 30.30.30.3
IPv4
lease-type=address-only (IA_NA option added to the client's request)
rate[1/s]=1
base[0]=mac=fa:16:3e:6e:90:e7
drop-time[0]=1
drop-time[1]=1
aggressivity=1
local-port=10067
elp-offset=-1
sid-offset=-1
rip-offset=-1
diagnostic-selectors=a
interface=tap883a32b1-77
server=30.30.30.3
Set MAC to fa::16::3e::6e::90::e7
Set DUID to 000100011ca31722000c01020304
^CFailed to receive DHCPv4 packet: Interrupted system call
***Rate statistics***
Rate: 0 exchanges/second, expected rate: 1 exchanges/second

***Statistics for: DISCOVER-OFFER***
sent packets: 3
received packets: 0
drops: 3