Hi, All.
There is a problem with retranslating packets in B.A.T.M.A.N. V mode. I have 3 stations, st1, st2 and st3 with one active network interface. This interface is configured so that st1 and st3 can see only st2. I expect that if I join these stations in a bat0 network, then station 2 will act as a repeater and all three stations will be visible in the bat0 network. This works fine if I selecting B.A.T.M.A.N. IV algorithm, but doesn't work in B.A.T.M.A.N.V.
-- Ermakov Alexey.
On Thursday, 21 May 2020 19:13:18 CEST Alexey Ermakov wrote: [...]
I have 3 stations, st1, st2 and st3 with one active network interface. This interface is configured so that st1 and st3 can see only st2.
[...]
This works fine if I selecting B.A.T.M.A.N. IV algorithm, but doesn't work in B.A.T.M.A.N.V.
That's not a lot to work with. I am rather sure that the B.A.T.M.A.N. V developers will not be able to figure out what is going on with this information.
I have just tried it out in my documented setup [1,2] and following test-init.sh::
#! /bin/sh
set -e export PATH="/host/batctl/:$PATH"
insmod /host/batman-adv/net/batman-adv/batman-adv.ko batctl routing_algo BATMAN_V /host/batctl/batctl if add enp0s1 ip link set up dev enp0s1 ip link set up dev bat0
MAC_PART="$(ip link show enp0s1 | awk "/ether/ {print $2}"| sed -e "s/.*://" -e "s/[\n\ ].*//"|awk "{print ("0x"$1)*1 }")" IP_PART="$(echo $MAC_PART|awk "{ print $1 }")" ip addr replace 192.168.55.${IP_PART}/24 dev bat0
The used kernel version is next-20200521 and batman-adv 2020.1.
These are the neighbor and originator tables of the devices + ping test results:
* node1
- neighbor::
[B.A.T.M.A.N. adv 2020.1, MainIF/MAC: enp0s1/02:ba:de:af:fe:01 (bat0/12:99:c8:a5:1e:27 BATMAN_V)] IF Neighbor last-seen 02:ba:de:af:fe:02 0.528s ( 1.0) [ enp0s1]
- originator::
[B.A.T.M.A.N. adv 2020.1, MainIF/MAC: enp0s1/02:ba:de:af:fe:01 (bat0/12:99:c8:a5:1e:27 BATMAN_V)] Originator last-seen ( throughput) Nexthop [outgoingIF] * 02:ba:de:af:fe:03 0.704s ( 0.8) 02:ba:de:af:fe:02 [ enp0s1] * 02:ba:de:af:fe:02 0.704s ( 1.0) 02:ba:de:af:fe:02 [ enp0s1]
- bat0 ping::
$ for i in $(seq 1 3); do ping -q -c 3 192.168.55.$i; done PING 192.168.55.1 (192.168.55.1) 56(84) bytes of data.
--- 192.168.55.1 ping statistics --- 3 packets transmitted, 3 received, 0% packet loss, time 28ms rtt min/avg/max/mdev = 0.482/0.539/0.613/0.060 ms PING 192.168.55.2 (192.168.55.2) 56(84) bytes of data.
--- 192.168.55.2 ping statistics --- 3 packets transmitted, 3 received, 0% packet loss, time 8ms rtt min/avg/max/mdev = 1.797/1.940/2.060/0.114 ms PING 192.168.55.3 (192.168.55.3) 56(84) bytes of data.
--- 192.168.55.3 ping statistics --- 3 packets transmitted, 3 received, 0% packet loss, time 10ms rtt min/avg/max/mdev = 2.105/3.021/3.962/0.760 ms
- enp1s0 ping::
$ for i in $(seq 1 3); do ping -q -c 3 192.168.251.5$i; done PING 192.168.251.51 (192.168.251.51) 56(84) bytes of data.
--- 192.168.251.51 ping statistics --- 3 packets transmitted, 3 received, 0% packet loss, time 36ms rtt min/avg/max/mdev = 0.170/0.396/0.510/0.160 ms PING 192.168.251.52 (192.168.251.52) 56(84) bytes of data.
--- 192.168.251.52 ping statistics --- 3 packets transmitted, 3 received, 0% packet loss, time 8ms rtt min/avg/max/mdev = 1.572/2.353/3.192/0.664 ms PING 192.168.251.53 (192.168.251.53) 56(84) bytes of data.
--- 192.168.251.53 ping statistics --- 3 packets transmitted, 0 received, +3 errors, 100% packet loss, time 41ms
* node2
- neighbor::
[B.A.T.M.A.N. adv 2020.1, MainIF/MAC: enp0s1/02:ba:de:af:fe:02 (bat0/32:a0:63:19:b4:44 BATMAN_V)] IF Neighbor last-seen 02:ba:de:af:fe:03 0.208s ( 1.0) [ enp0s1] 02:ba:de:af:fe:01 0.344s ( 1.0) [ enp0s1]
- originator::
[B.A.T.M.A.N. adv 2020.1, MainIF/MAC: enp0s1/02:ba:de:af:fe:02 (bat0/32:a0:63:19:b4:44 BATMAN_V)] Originator last-seen ( throughput) Nexthop [outgoingIF] * 02:ba:de:af:fe:03 0.188s ( 1.0) 02:ba:de:af:fe:03 [ enp0s1] * 02:ba:de:af:fe:01 0.016s ( 1.0) 02:ba:de:af:fe:01 [ enp0s1]
- bat0 ping::
$ for i in $(seq 1 3); do ping -q -c 3 192.168.55.$i; done PING 192.168.55.1 (192.168.55.1) 56(84) bytes of data.
--- 192.168.55.1 ping statistics --- 3 packets transmitted, 3 received, 0% packet loss, time 7ms rtt min/avg/max/mdev = 1.961/2.510/3.484/0.691 ms PING 192.168.55.2 (192.168.55.2) 56(84) bytes of data.
--- 192.168.55.2 ping statistics --- 3 packets transmitted, 3 received, 0% packet loss, time 38ms rtt min/avg/max/mdev = 0.287/0.481/0.606/0.140 ms PING 192.168.55.3 (192.168.55.3) 56(84) bytes of data.
--- 192.168.55.3 ping statistics --- 3 packets transmitted, 3 received, 0% packet loss, time 9ms rtt min/avg/max/mdev = 1.215/1.737/2.038/0.370 ms
- enp1s0 ping::
$ for i in $(seq 1 3); do ping -q -c 3 192.168.251.5$i; done PING 192.168.251.51 (192.168.251.51) 56(84) bytes of data.
--- 192.168.251.51 ping statistics --- 3 packets transmitted, 3 received, 0% packet loss, time 10ms rtt min/avg/max/mdev = 1.114/1.698/2.107/0.425 ms PING 192.168.251.52 (192.168.251.52) 56(84) bytes of data.
--- 192.168.251.52 ping statistics --- 3 packets transmitted, 3 received, 0% packet loss, time 14ms rtt min/avg/max/mdev = 0.360/0.747/1.270/0.384 ms PING 192.168.251.53 (192.168.251.53) 56(84) bytes of data.
--- 192.168.251.53 ping statistics --- 3 packets transmitted, 3 received, 0% packet loss, time 7ms rtt min/avg/max/mdev = 1.909/2.427/2.692/0.371 ms
* node3
- neighbor::
[B.A.T.M.A.N. adv 2020.1, MainIF/MAC: enp0s1/02:ba:de:af:fe:03 (bat0/7a:f4:94:ef:b6:6d BATMAN_V)] IF Neighbor last-seen 02:ba:de:af:fe:02 0.072s ( 1.0) [ enp0s1]
- originator::
[B.A.T.M.A.N. adv 2020.1, MainIF/MAC: enp0s1/02:ba:de:af:fe:03 (bat0/7a:f4:94:ef:b6:6d BATMAN_V)] Originator last-seen ( throughput) Nexthop [outgoingIF] * 02:ba:de:af:fe:01 0.764s ( 0.8) 02:ba:de:af:fe:02 [ enp0s1] * 02:ba:de:af:fe:02 0.148s ( 1.0) 02:ba:de:af:fe:02 [ enp0s1]
- bat0 ping::
$ for i in $(seq 1 3); do ping -q -c 3 192.168.55.$i; done PING 192.168.55.1 (192.168.55.1) 56(84) bytes of data.
--- 192.168.55.1 ping statistics --- 3 packets transmitted, 3 received, 0% packet loss, time 10ms rtt min/avg/max/mdev = 1.736/2.948/3.925/0.910 ms PING 192.168.55.2 (192.168.55.2) 56(84) bytes of data.
--- 192.168.55.2 ping statistics --- 3 packets transmitted, 3 received, 0% packet loss, time 11ms rtt min/avg/max/mdev = 2.081/2.175/2.361/0.136 ms PING 192.168.55.3 (192.168.55.3) 56(84) bytes of data.
--- 192.168.55.3 ping statistics --- 3 packets transmitted, 3 received, 0% packet loss, time 27ms rtt min/avg/max/mdev = 0.278/0.489/0.649/0.156 ms
- enp1s0 ping::
$ for i in $(seq 1 3); do ping -q -c 3 192.168.251.5$i; done PING 192.168.251.51 (192.168.251.51) 56(84) bytes of data.
--- 192.168.251.51 ping statistics --- 3 packets transmitted, 0 received, +3 errors, 100% packet loss, time 43ms pipe 3 PING 192.168.251.52 (192.168.251.52) 56(84) bytes of data.
--- 192.168.251.52 ping statistics --- 3 packets transmitted, 3 received, 0% packet loss, time 10ms rtt min/avg/max/mdev = 1.025/1.569/1.997/0.405 ms PING 192.168.251.53 (192.168.251.53) 56(84) bytes of data.
--- 192.168.251.53 ping statistics --- 3 packets transmitted, 3 received, 0% packet loss, time 61ms rtt min/avg/max/mdev = 0.214/0.376/0.480/0.117 ms
And it seems I can communicate with all nodes perfectly fine. And the communication between node1 and node3 seems to be blocked as expected on enp1s0 (the slave device of bat0).
Kind regards, Sven
[1] https://www.open-mesh.org/projects/open-mesh/wiki/Kernel_hacking_Debian_imag... [2] https://www.open-mesh.org/projects/open-mesh/wiki/Advanced_Bridge_virtual_ne...
Sorry, I didn't provided full information about experiment conditions. I could reproduce issue on virtual machines. I started 3 nodes with virtual interfaces eth0 which linked to host bridge interface br0. Network traffic between node 1 and node 3 drops by nft as described Sven Eckelmann above. At start moment br0 is down. After virtual machines had booted I configured them by following steps: 1) insmod /usr/batman-adv.ko 2) ifconfig eth0 up 3) batctl ra BATMAN_V 4 batctl if add eth0 5) ifconfig bat0 192.168.1.<n> up NOTE: 'n' is node numer: 1, 2 or 3 6) batctl hardif eth0 to 100 NOTE: If sklip this step, all works fine!
After this I started host bridge interface br0. NOTE: if bridge had started before batman configured all works fine
After these steps I have following results: --- NODE 1: (none)# ip -o link show dev eth0 2: eth0: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc pfifo_fast master bat0 state UP mode DEFAULT group default qlen 1000\ link/ether 00:04:23:dc:3b:01 brd ff:ff:ff:ff :ff:ff (none)# ip -o link show dev bat0 17: bat0: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc noqueue state UNKNOWN mode DEFAULT group default qlen 1000\ link/ether 62:7f:0f:87:43:78 brd ff:ff:ff:ff:ff:ff (none)# batctl o [B.A.T.M.A.N. adv 2020.1, MainIF/MAC: eth0/00:04:23:dc:3b:01 (bat0/62:7f:0f:87:43:78 BATMAN_V)] Originator last-seen ( throughput) Nexthop [outgoingIF] * 00:04:23:dc:3b:10 0.580s ( 0.1) 00:04:23:dc:3b:10 [ eth0]
--- NODE 2: (none)# ip -o link show dev eth0 2: eth0: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc pfifo_fast master bat0 state UP mode DEFAULT group default qlen 1000\ link/ether 00:04:23:dc:3b:10 brd ff:ff:ff:ff :ff:ff (none)# ip -o link show dev bat0 17: bat0: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc noqueue state UNKNOWN mode DEFAULT group default qlen 1000\ link/ether 22:86:ba:4e:9c:5c brd ff:ff:ff:ff:ff:ff (none)# batctl o [B.A.T.M.A.N. adv 2020.1, MainIF/MAC: eth0/00:04:23:dc:3b:10 (bat0/22:86:ba:4e:9c:5c BATMAN_V)] Originator last-seen ( throughput) Nexthop [outgoingIF] * 00:04:23:dc:3b:01 0.960s ( 0.1) 00:04:23:dc:3b:01 [ eth0] * 00:04:23:dc:3b:1f 0.450s ( 0.1) 00:04:23:dc:3b:1f [ eth0]
--- NODE 3: (none)# ip -o link show dev eth0 2: eth0: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc pfifo_fast master bat0 state UP mode DEFAULT group default qlen 1000\ link/ether 00:04:23:dc:3b:1f brd ff:ff:ff:ff :ff:ff (none)# ip -o link show dev bat0 17: bat0: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc noqueue state UNKNOWN mode DEFAULT group default qlen 1000\ link/ether 42:d8:87:c3:7e:6e brd ff:ff:ff:ff:ff:ff (none)# batctl o [B.A.T.M.A.N. adv 2020.1, MainIF/MAC: eth0/00:04:23:dc:3b:1f (bat0/42:d8:87:c3:7e:6e BATMAN_V)] Originator last-seen ( throughput) Nexthop [outgoingIF] * 00:04:23:dc:3b:10 0.330s ( 0.1) 00:04:23:dc:3b:10 [ eth0]
--- I didn't provided ping result because it's oblivious. In my case eth0 not actualy ethernet connection between nodes, it is connection to radio modem which configured as bridge and it can be configured for a long time. Therefore, there is no guarantee that the channel will be available at the batman setup time (but eth0 link is UP).
пт, 22 мая 2020 г. в 09:25, Sven Eckelmann sven@narfation.org:
On Thursday, 21 May 2020 19:13:18 CEST Alexey Ermakov wrote: [...]
I have 3 stations, st1, st2 and st3 with one active network interface. This interface is configured so that st1 and st3 can see only st2.
[...]
This works fine if I selecting B.A.T.M.A.N. IV algorithm, but doesn't work in B.A.T.M.A.N.V.
That's not a lot to work with. I am rather sure that the B.A.T.M.A.N. V developers will not be able to figure out what is going on with this information.
I have just tried it out in my documented setup [1,2] and following test-init.sh::
#! /bin/sh set -e export PATH="/host/batctl/:$PATH" insmod /host/batman-adv/net/batman-adv/batman-adv.ko batctl routing_algo BATMAN_V /host/batctl/batctl if add enp0s1 ip link set up dev enp0s1 ip link set up dev bat0 MAC_PART="$(ip link show enp0s1 | awk "/ether/ {print \$2}"| sed -e "s/.*://" -e "s/[\\n\\ ].*//"|awk "{print (\"0x\"\$1)*1 }")" IP_PART="$(echo $MAC_PART|awk "{ print \$1 }")" ip addr replace 192.168.55.${IP_PART}/24 dev bat0The used kernel version is next-20200521 and batman-adv 2020.1.
These are the neighbor and originator tables of the devices + ping test results:
node1
neighbor::
[B.A.T.M.A.N. adv 2020.1, MainIF/MAC: enp0s1/02:ba:de:af:fe:01 (bat0/12:99:c8:a5:1e:27 BATMAN_V)] IF Neighbor last-seen 02:ba:de:af:fe:02 0.528s ( 1.0) [ enp0s1]
originator::
[B.A.T.M.A.N. adv 2020.1, MainIF/MAC: enp0s1/02:ba:de:af:fe:01 (bat0/12:99:c8:a5:1e:27 BATMAN_V)] Originator last-seen ( throughput) Nexthop [outgoingIF]
- 02:ba:de:af:fe:03 0.704s ( 0.8) 02:ba:de:af:fe:02 [ enp0s1]
- 02:ba:de:af:fe:02 0.704s ( 1.0) 02:ba:de:af:fe:02 [ enp0s1]
bat0 ping::
$ for i in $(seq 1 3); do ping -q -c 3 192.168.55.$i; done PING 192.168.55.1 (192.168.55.1) 56(84) bytes of data.
--- 192.168.55.1 ping statistics --- 3 packets transmitted, 3 received, 0% packet loss, time 28ms rtt min/avg/max/mdev = 0.482/0.539/0.613/0.060 ms PING 192.168.55.2 (192.168.55.2) 56(84) bytes of data.
--- 192.168.55.2 ping statistics --- 3 packets transmitted, 3 received, 0% packet loss, time 8ms rtt min/avg/max/mdev = 1.797/1.940/2.060/0.114 ms PING 192.168.55.3 (192.168.55.3) 56(84) bytes of data.
--- 192.168.55.3 ping statistics --- 3 packets transmitted, 3 received, 0% packet loss, time 10ms rtt min/avg/max/mdev = 2.105/3.021/3.962/0.760 ms
enp1s0 ping::
$ for i in $(seq 1 3); do ping -q -c 3 192.168.251.5$i; done PING 192.168.251.51 (192.168.251.51) 56(84) bytes of data.
--- 192.168.251.51 ping statistics --- 3 packets transmitted, 3 received, 0% packet loss, time 36ms rtt min/avg/max/mdev = 0.170/0.396/0.510/0.160 ms PING 192.168.251.52 (192.168.251.52) 56(84) bytes of data.
--- 192.168.251.52 ping statistics --- 3 packets transmitted, 3 received, 0% packet loss, time 8ms rtt min/avg/max/mdev = 1.572/2.353/3.192/0.664 ms PING 192.168.251.53 (192.168.251.53) 56(84) bytes of data.
--- 192.168.251.53 ping statistics --- 3 packets transmitted, 0 received, +3 errors, 100% packet loss, time 41ms
node2
neighbor::
[B.A.T.M.A.N. adv 2020.1, MainIF/MAC: enp0s1/02:ba:de:af:fe:02 (bat0/32:a0:63:19:b4:44 BATMAN_V)] IF Neighbor last-seen 02:ba:de:af:fe:03 0.208s ( 1.0) [ enp0s1] 02:ba:de:af:fe:01 0.344s ( 1.0) [ enp0s1]
originator::
[B.A.T.M.A.N. adv 2020.1, MainIF/MAC: enp0s1/02:ba:de:af:fe:02 (bat0/32:a0:63:19:b4:44 BATMAN_V)] Originator last-seen ( throughput) Nexthop [outgoingIF]
- 02:ba:de:af:fe:03 0.188s ( 1.0) 02:ba:de:af:fe:03 [ enp0s1]
- 02:ba:de:af:fe:01 0.016s ( 1.0) 02:ba:de:af:fe:01 [ enp0s1]
bat0 ping::
$ for i in $(seq 1 3); do ping -q -c 3 192.168.55.$i; done PING 192.168.55.1 (192.168.55.1) 56(84) bytes of data.
--- 192.168.55.1 ping statistics --- 3 packets transmitted, 3 received, 0% packet loss, time 7ms rtt min/avg/max/mdev = 1.961/2.510/3.484/0.691 ms PING 192.168.55.2 (192.168.55.2) 56(84) bytes of data.
--- 192.168.55.2 ping statistics --- 3 packets transmitted, 3 received, 0% packet loss, time 38ms rtt min/avg/max/mdev = 0.287/0.481/0.606/0.140 ms PING 192.168.55.3 (192.168.55.3) 56(84) bytes of data.
--- 192.168.55.3 ping statistics --- 3 packets transmitted, 3 received, 0% packet loss, time 9ms rtt min/avg/max/mdev = 1.215/1.737/2.038/0.370 ms
enp1s0 ping::
$ for i in $(seq 1 3); do ping -q -c 3 192.168.251.5$i; done PING 192.168.251.51 (192.168.251.51) 56(84) bytes of data.
--- 192.168.251.51 ping statistics --- 3 packets transmitted, 3 received, 0% packet loss, time 10ms rtt min/avg/max/mdev = 1.114/1.698/2.107/0.425 ms PING 192.168.251.52 (192.168.251.52) 56(84) bytes of data.
--- 192.168.251.52 ping statistics --- 3 packets transmitted, 3 received, 0% packet loss, time 14ms rtt min/avg/max/mdev = 0.360/0.747/1.270/0.384 ms PING 192.168.251.53 (192.168.251.53) 56(84) bytes of data.
--- 192.168.251.53 ping statistics --- 3 packets transmitted, 3 received, 0% packet loss, time 7ms rtt min/avg/max/mdev = 1.909/2.427/2.692/0.371 ms
node3
neighbor::
[B.A.T.M.A.N. adv 2020.1, MainIF/MAC: enp0s1/02:ba:de:af:fe:03 (bat0/7a:f4:94:ef:b6:6d BATMAN_V)] IF Neighbor last-seen 02:ba:de:af:fe:02 0.072s ( 1.0) [ enp0s1]
originator::
[B.A.T.M.A.N. adv 2020.1, MainIF/MAC: enp0s1/02:ba:de:af:fe:03 (bat0/7a:f4:94:ef:b6:6d BATMAN_V)] Originator last-seen ( throughput) Nexthop [outgoingIF]
- 02:ba:de:af:fe:01 0.764s ( 0.8) 02:ba:de:af:fe:02 [ enp0s1]
- 02:ba:de:af:fe:02 0.148s ( 1.0) 02:ba:de:af:fe:02 [ enp0s1]
bat0 ping::
$ for i in $(seq 1 3); do ping -q -c 3 192.168.55.$i; done PING 192.168.55.1 (192.168.55.1) 56(84) bytes of data.
--- 192.168.55.1 ping statistics --- 3 packets transmitted, 3 received, 0% packet loss, time 10ms rtt min/avg/max/mdev = 1.736/2.948/3.925/0.910 ms PING 192.168.55.2 (192.168.55.2) 56(84) bytes of data.
--- 192.168.55.2 ping statistics --- 3 packets transmitted, 3 received, 0% packet loss, time 11ms rtt min/avg/max/mdev = 2.081/2.175/2.361/0.136 ms PING 192.168.55.3 (192.168.55.3) 56(84) bytes of data.
--- 192.168.55.3 ping statistics --- 3 packets transmitted, 3 received, 0% packet loss, time 27ms rtt min/avg/max/mdev = 0.278/0.489/0.649/0.156 ms
enp1s0 ping::
$ for i in $(seq 1 3); do ping -q -c 3 192.168.251.5$i; done PING 192.168.251.51 (192.168.251.51) 56(84) bytes of data.
--- 192.168.251.51 ping statistics --- 3 packets transmitted, 0 received, +3 errors, 100% packet loss, time 43ms pipe 3 PING 192.168.251.52 (192.168.251.52) 56(84) bytes of data.
--- 192.168.251.52 ping statistics --- 3 packets transmitted, 3 received, 0% packet loss, time 10ms rtt min/avg/max/mdev = 1.025/1.569/1.997/0.405 ms PING 192.168.251.53 (192.168.251.53) 56(84) bytes of data.
--- 192.168.251.53 ping statistics --- 3 packets transmitted, 3 received, 0% packet loss, time 61ms rtt min/avg/max/mdev = 0.214/0.376/0.480/0.117 ms
And it seems I can communicate with all nodes perfectly fine. And the communication between node1 and node3 seems to be blocked as expected on enp1s0 (the slave device of bat0).
Kind regards, Sven
[1] https://www.open-mesh.org/projects/open-mesh/wiki/Kernel_hacking_Debian_imag... [2] https://www.open-mesh.org/projects/open-mesh/wiki/Advanced_Bridge_virtual_ne...
On Sunday, 24 May 2020 16:43:55 CEST Alexey Ermakov wrote:
[B.A.T.M.A.N. adv 2020.1, MainIF/MAC: eth0/00:04:23:dc:3b:01 (bat0/62:7f:0f:87:43:78 BATMAN_V)] Originator last-seen ( throughput) Nexthop [outgoingIF]
- 00:04:23:dc:3b:10 0.580s ( 0.1) 00:04:23:dc:3b:10 [ eth0]
BATMAN V chooses routes based on link throughput. If the throughput drops below a certain value the route is dropped. As you can imagine, a throughput of 0.1 is pretty close to non-usable.
BATMAN V attempts to auto-detect the throughput using various methods which does not work in a virtualized environment. I suggest you go with the per- interace /sys/class/net/<iface>/batman-adv/throughput_override setting.
Kind regards, Marek Lindner
Yes, in real world I set throughput overriode for each interface, but in my example only one interface configured in batman network and if it works as you said there are several questions: - Why direct connections work properly (I mean connections Node1-Node2 and Node2-Node3 ) even with this throughput? - Why this scheme works then all nodes had seen originators before throughput overrided? - If I override throughput to 10000, for example, will it work properly?
вс, 24 мая 2020 г., 18:29 Marek Lindner mareklindner@neomailbox.ch:
On Sunday, 24 May 2020 16:43:55 CEST Alexey Ermakov wrote:
[B.A.T.M.A.N. adv 2020.1, MainIF/MAC: eth0/00:04:23:dc:3b:01 (bat0/62:7f:0f:87:43:78 BATMAN_V)] Originator last-seen ( throughput) Nexthop [outgoingIF]
- 00:04:23:dc:3b:10 0.580s ( 0.1) 00:04:23:dc:3b:10 [ eth0]
BATMAN V chooses routes based on link throughput. If the throughput drops below a certain value the route is dropped. As you can imagine, a throughput of 0.1 is pretty close to non-usable.
BATMAN V attempts to auto-detect the throughput using various methods which does not work in a virtualized environment. I suggest you go with the per- interace /sys/class/net/<iface>/batman-adv/throughput_override setting.
Kind regards, Marek Lindner
On Sunday, 24 May 2020 17:49:23 CEST Alexey Ermakov wrote:
- Why direct connections work properly (I mean connections Node1-Node2
and Node2-Node3 ) even with this throughput?
Batman originator messages are not forwarded when the throughput value is too low. Furthermore, the batman algorithm applies a hop penalty on each hop which reduces the metric with each hop. That is why one hop is still works.
- Why this scheme works then all nodes had seen originators before
throughput overrided?
I don't quite understand the question, sorry.
- If I override throughput to 10000, for example, will it work properly?
It should. Why not simply give it a try ?
Kind regards, Marek Lindner
On Sunday, 24 May 2020 16:43:55 CEST Alexey Ermakov wrote:
Sorry, I didn't provided full information about experiment conditions.
[...]
- batctl hardif eth0 to 100 NOTE: If sklip this step,
all works fine!
After this I started host bridge interface br0. NOTE: if bridge had started before batman configured all works fine
As Marek pointed out, you told batman-adv that the eth0 interfaces have for some reason a link speed of 0.1 [Mbit/s] (or 1 [100Kbit/s] - what the actual unit is for B.A.T.M.A.N. V). And unfortunately, 88% (for the hop penalty) of 1 [100 Kbit/s] is in integer arithmetic 0 [100 Kbit/s]. And 0 whatever/s is just discarded by B.A.T.M.A.N. V [1] and not considered for routing traffic - hey, this path ends up as being 0 and therefore useless.
The reason why you want to set the link to 1 [100 Kbit/s] has to be figured out by yourself. But for a (during forwarding) strictly monotonic decreasing, integer value, it is a bad starting point to set it to 1 [whatever-unit] for a multi-hop setup.
Kind regards, Sven
[1] https://git.open-mesh.org/batman-adv.git/blob/e946e0beec5764ed5182943bf764f8...
As Marek pointed out, you told batman-adv that the eth0 interfaces have for some reason a link speed of 0.1 [Mbit/s] (or 1 [100Kbit/s] - what the actual unit is for B.A.T.M.A.N. V). And unfortunately, 88% (for the hop penalty) of 1 [100 Kbit/s] is in integer arithmetic 0 [100 Kbit/s]. And 0 whatever/s is just discarded by B.A.T.M.A.N. V [1] and not considered for routing traffic - hey, this path ends up as being 0 and therefore useless. The reason why you want to set the link to 1 [100 Kbit/s] has to be figured out by yourself. But for a (during forwarding) strictly monotonic decreasing, integer value, it is a bad starting point to set it to 1 [whatever-unit] for a multi-hop setup.
I have three interfaces on each node: eth0 - radio modem, speed 145 kbit/s (see desiription below) eth1 - ethernet, speed 100mbit/s eth2 - radio modem 10kbit/s (hasn't been tested yet)
I have to say few words about radio modems. Radio modem is not builtin device with special driver, it is separate machine with linux which set up as bridge and connected to node by ethernet. This means node can send data to modem with 100mbit/s, and think that all is ok but data will dropped on modem side, it will not call netif_stop_queue, or something on node side.
All tests I have been making were with eth0 and eth1 because eth2 doesn't ready yet.
--- FIRST test conditions: - All 3 nodes see each other - Both interfaces added to batman network - no throuthput override
FIRST test steps: - Turn on all nodes and setup batman - All nodes see each other. Route goes via eth1 - Switch off eth1 (plug off ethernet cable) - Route switches to eth0, all nodes see each other - Switch eth1 on - Unstable behavior: sometimes route switch back to eth1, but sometimes - not --- SECOND test conditions: - All 3 nodes see each other - Both interfaces added to batman network - eth0 throuthput override set to 100, eth1 throuthput override set to 100000
SECOND test steps: - Turn on all nodes and setup batman - All nodes see each other. Route goes via eth1 - Switch off eth1 (plug off ethernet cable) - Route switches to eth0, all nodes see each other - Switch eth1 on - Route switch back to eth1, works as expected --- THIRD test conditions: - Node 1 And Node3 see only node 2, but not each other on eth0. All 3 nodes see each other on eth1 - Both interfaces added to batman network - no throuthput override
THIRD test steps: - Turn on all nodes and setup batman - All nodes see each other. Route goes via eth1 - Switch off eth1 (plug off ethernet cable) - Route switches to eth0, ALL NODES SEE EACH OTHER (this means speed 145 kbit/s enough to route through Node 2) - Switch eth1 on - Unstable behavior sometimes route switch back to eth1, but sometimes - not --- FOURTH test conditions: - Node 1 And Node3 see only node 2, but not each other on eth0. All 3 nodes see each other on eth1 - Both interfaces added to batman network - eth0 throuthput override set to 100, eth1 throuthput override set to 100000
FOURTH test steps: - Turn on all nodes and setup batman - All nodes see each other. Route goes via eth1 - Switch off eth1 (plug off ethernet cable) - Route switches to eth0, But Node 1 and Node 2 see only Node 3 for the reasons you described above. - Switch eth1 on - Route switch back to eth1, works as expected --- Summary: If I don't set override throughput, batman doesn't switch back to ster eth0, but if I set it, retranslation doesn't work properly. I believe, I should tweak batman and I whould be happy if you say how.
On Sunday, 24 May 2020 23:53:05 CEST Alexey Ermakov wrote:
As Marek pointed out, you told batman-adv that the eth0 interfaces have for some reason a link speed of 0.1 [Mbit/s] (or 1 [100Kbit/s] - what the actual unit is for B.A.T.M.A.N. V). And unfortunately, 88% (for the hop penalty) of 1 [100 Kbit/s] is in integer arithmetic 0 [100 Kbit/s]. And 0 whatever/s is just discarded by B.A.T.M.A.N. V [1] and not considered for routing traffic - hey, this path ends up as being 0 and therefore useless. The reason why you want to set the link to 1 [100 Kbit/s] has to be figured out by yourself. But for a (during forwarding) strictly monotonic decreasing, integer value, it is a bad starting point to set it to 1 [whatever-unit] for a multi-hop setup.
I have three interfaces on each node: eth0 - radio modem, speed 145 kbit/s (see desiription below) eth1 - ethernet, speed 100mbit/s eth2 - radio modem 10kbit/s (hasn't been tested yet)
Before we go to the next setup, can you confirm that your virtual setup works as expected once the throughput is not set to 0.1 MBit/s ?
Kind regards, Marek
On Sunday, 24 May 2020 23:53:05 CEST Alexey Ermakov wrote:
Summary: If I don't set override throughput, batman doesn't switch back to ster eth0, but if I set it, retranslation doesn't work properly. I believe, I should tweak batman and I whould be happy if you say how.
To give advice on what to change, please provide the batman-adv routing tables which allow us to understand how batman-adv sees your network. Please refer to Sven's email with various examples.
For starters, we can focus on a single setup with originator & neighbors tables. then you can explain what isn't working the way you think it should.
Kind regards, Marek Lindner
To give advice on what to change, please provide the batman-adv routing tables which allow us to understand how batman-adv sees your network. Please refer to Sven's email with various examples.
Below I provide ethernet MAC addresses, tables with throughput overrid to 100kbit/s and tables without output override. Note that in case without override batman detects throughput about 100mbit/s. This is not quiet correct, because as I say before, nodes connects to modems and throughput from node to modem is 100mbit/s, but actual speed of modem is 145 kbit/sб therefore I should set throughput override
eth0 interface MAC adresses:
Node 1:
node1# ip -o link show dev eth0 31: eth0@eth2: <BROADCAST,UP,LOWER_UP> mtu 1500 qdisc noqueue master bat0 state UP mode DEFAULT group default qlen 1000\ link/ether fe:00:00:c3:5b:13 brd ff:ff:ff:ff:ff:ff
Node 2:
node2# ip -o link show dev eth0 31: eth0@eth2: <BROADCAST,UP,LOWER_UP> mtu 1500 qdisc noqueue master bat0 state UP mode DEFAULT group default qlen 1000\ link/ether fe:00:00:3b:9b:13 brd ff:ff:ff:ff:ff:ff
---
Node 3:
node3# ip -o link show dev eth0 31: eth0@eth2: <BROADCAST,UP,LOWER_UP> mtu 1500 qdisc noqueue master bat0 state UP mode DEFAULT group default qlen 1000\ link/ether fe:00:00:31:ed:13 brd ff:ff:ff:ff:ff:ff
Tables with throughput overrid to 100kbit/s:
-- Node 1 tables --
node1# batctl o [B.A.T.M.A.N. adv 2020.1, MainIF/MAC: eth0/fe:00:00:c3:5b:13 (bat0/fe:ee:f9:fe:00:68 BATMAN_V)] Originator last-seen ( throughput) Nexthop [outgoingIF] * fe:00:00:3b:9b:13 0.750s ( 0.1) fe:00:00:3b:9b:13 [ eth0] node1# batctl n [B.A.T.M.A.N. adv 2020.1, MainIF/MAC: eth0/fe:00:00:c3:5b:13 (bat0/fe:ee:f9:fe:00:68 BATMAN_V)] IF Neighbor last-seen fe:00:00:3b:9b:13 0.410s ( 0.1) [ eth0]
-- Node 2 tables --
node2# batctl o [B.A.T.M.A.N. adv 2020.1, MainIF/MAC: eth0/fe:00:00:3b:9b:13 (bat0/6a:53:d2:c1:6f:2e BATMAN_V)] Originator last-seen ( throughput) Nexthop [outgoingIF] * fe:00:00:31:ed:13 0.190s ( 0.1) fe:00:00:31:ed:13 [ eth0] * fe:00:00:c3:5b:13 0.210s ( 0.1) fe:00:00:c3:5b:13 [ eth0] node2# batctl n [B.A.T.M.A.N. adv 2020.1, MainIF/MAC: eth0/fe:00:00:3b:9b:13 (bat0/6a:53:d2:c1:6f:2e BATMAN_V)] IF Neighbor last-seen fe:00:00:31:ed:13 0.320s ( 0.1) [ eth0] fe:00:00:c3:5b:13 0.120s ( 0.1) [ eth0]
-- Node 3 tables --
node3# batctl o [B.A.T.M.A.N. adv 2020.1, MainIF/MAC: eth0/fe:00:00:31:ed:13 (bat0/fa:be:fa:e6:5b:7a BATMAN_V)] Originator last-seen ( throughput) Nexthop [outgoingIF] * fe:00:00:3b:9b:13 0.650s ( 0.1) fe:00:00:3b:9b:13 [ eth0] node3# batctl n [B.A.T.M.A.N. adv 2020.1, MainIF/MAC: eth0/fe:00:00:31:ed:13 (bat0/fa:be:fa:e6:5b:7a BATMAN_V)] IF Neighbor last-seen fe:00:00:3b:9b:13 0.500s ( 0.1) [ eth0]
-----------------------------------------
Tables without throughput override:
-- Node 1 tables --
node1# batctl o [B.A.T.M.A.N. adv 2020.1, MainIF/MAC: eth0/fe:00:00:c3:5b:13 (bat0/26:f5:9a:68:f5:2d BATMAN_V)] Originator last-seen ( throughput) Nexthop [outgoingIF] * fe:00:00:3b:9b:13 0.090s ( 100.0) fe:00:00:3b:9b:13 [ eth0] * fe:00:00:31:ed:13 0.350s ( 88.2) fe:00:00:3b:9b:13 [ eth0] * fe:00:00:c3:5b:13 1.330s ( 88.2) fe:00:00:3b:9b:13 [ eth0] node1# batctl n [B.A.T.M.A.N. adv 2020.1, MainIF/MAC: eth0/fe:00:00:c3:5b:13 (bat0/26:f5:9a:68:f5:2d BATMAN_V)] IF Neighbor last-seen fe:00:00:3b:9b:13 0.010s ( 100.0) [ eth0]
-- Node 2 tables --
node2# batctl o [B.A.T.M.A.N. adv 2020.1, MainIF/MAC: eth0/fe:00:00:3b:9b:13 (bat0/6a:5d:4e:09:dc:f4 BATMAN_V)] Originator last-seen ( throughput) Nexthop [outgoingIF] * fe:00:00:31:ed:13 0.150s ( 100.0) fe:00:00:31:ed:13 [ eth0] * fe:00:00:c3:5b:13 0.080s ( 100.0) fe:00:00:c3:5b:13 [ eth0] node2# batctl n [B.A.T.M.A.N. adv 2020.1, MainIF/MAC: eth0/fe:00:00:3b:9b:13 (bat0/6a:5d:4e:09:dc:f4 BATMAN_V)] IF Neighbor last-seen fe:00:00:31:ed:13 0.440s ( 100.0) [ eth0] fe:00:00:c3:5b:13 0.060s ( 100.0) [ eth0]
-- Node 3 tables --
node3# batctl o [B.A.T.M.A.N. adv 2020.1, MainIF/MAC: eth0/fe:00:00:31:ed:13 (bat0/12:a7:e0:9e:f3:b6 BATMAN_V)] Originator last-seen ( throughput) Nexthop [outgoingIF] * fe:00:00:3b:9b:13 0.100s ( 100.0) fe:00:00:3b:9b:13 [ eth0] * fe:00:00:c3:5b:13 0.200s ( 88.2) fe:00:00:3b:9b:13 [ eth0] node3# batctl n [B.A.T.M.A.N. adv 2020.1, MainIF/MAC: eth0/fe:00:00:31:ed:13 (bat0/12:a7:e0:9e:f3:b6 BATMAN_V)] IF Neighbor last-seen fe:00:00:3b:9b:13 0.300s ( 100.0) [ eth0]
On Monday, 25 May 2020 16:59:23 CEST Alexey Ermakov wrote:
Below I provide ethernet MAC addresses, tables with throughput overrid to 100kbit/s and tables without output override. Note that in case without override batman detects throughput about 100mbit/s. This is not quiet correct, because as I say before, nodes connects to modems and throughput from node to modem is 100mbit/s, but actual speed of modem is 145 kbit/sб therefore I should set throughput override
I am surprised you still are configuring a throughput override of 100kbit/s. What's the idea behind this configuration ? As explained before, by configuring this low throughput value you are preventing any multi-hop route detection. Is that intended ?
Tables without throughput override:
Since the throughput is auto-detected in your setup why are you configuring a manual throughput override ?
Now that we see the tables where do we see what issue ? What is not working as per your expectation ?
Kind regards, Marek Lindner
I am surprised you still are configuring a throughput override of 100kbit/s. What's the idea behind this configuration ?
I have two channels between nodes (actually more, but this doesn't matter now) 1) Radio modem, connects as follows: NODE1(eth0)<-->modem(radio)<-->(radio)modem<-->(eth0)NODE2 eth0 interface have 100mbit/s speed and batman detects it as 100mbit/s. radio has 145kbit/s, and actual speed of this channel is 145kbit/s
2) eth1 Ethernet channel. Speed of this channel is 100mbit/s NODE1(eth1)<---->(eth1)NODE2
Now make a test: 1) Set up eth0 modem channel 2) Disconnect eth1 channel 3) Turn on nodes - batman saw eth0 channel between nodes and detected speed 100mbit/s (because speed between node and modem is 100mbit/s, as I understand) 4) plug eth1 cable in - At this moment I expect batman switch to eth1 channel because it has real 100mbit/s speed. But batman doesn't switch, because it detects eth0 as 100mbit/s too. Therefore I override throughput of eth0, actually batman detects incorrect speed on eth0.
On Monday, 25 May 2020 19:21:50 CEST Alexey Ermakov wrote:
I am surprised you still are configuring a throughput override of 100kbit/s. What's the idea behind this configuration ?
I have two channels between nodes (actually more, but this doesn't matter now) 1) Radio modem, connects as follows: NODE1(eth0)<-->modem(radio)<-->(radio)modem<-->(eth0)NODE2 eth0 interface have 100mbit/s speed and batman detects it as 100mbit/s. radio has 145kbit/s, and actual speed of this channel is 145kbit/s
- eth1 Ethernet channel. Speed of this channel is 100mbit/s NODE1(eth1)<---->(eth1)NODE2
Now make a test:
- Set up eth0 modem channel
- Disconnect eth1 channel
- Turn on nodes - batman saw eth0 channel between nodes and detected
speed 100mbit/s (because speed between node and modem is 100mbit/s, as I understand) 4) plug eth1 cable in - At this moment I expect batman switch to eth1 channel because it has real 100mbit/s speed. But batman doesn't switch, because it detects eth0 as 100mbit/s too. Therefore I override throughput of eth0, actually batman detects incorrect speed on eth0.
If you have to configure the throughput manually why not configure the link with something like 5mbit/s instead of 100kbit/s ? Batman-adv does not care that this throughput is not exact. All that matters is provide input to the algorithm to allow batman-adv to make smart choices.
Kind regards, Marek Lindner
If you have to configure the throughput manually why not configure the link with something like 5mbit/s instead of 100kbit/s ? Batman-adv does not care that this throughput is not exact. All that matters is provide input to the algorithm to allow batman-adv to make smart choices.
Yes, today I have changed throughput to 1mbit/s and it works as expected. I just wanted to make sure it was the right way. Thnks to all
Hi Alexey,
There is a problem with retranslating packets in B.A.T.M.A.N. V mode. I have 3 stations, st1, st2 and st3 with one active network interface. This interface is configured so that st1 and st3 can see only st2. I expect that if I join these stations in a bat0 network, then station 2 will act as a repeater and all three stations will be visible in the bat0 network. This works fine if I selecting B.A.T.M.A.N. IV algorithm, but doesn't work in B.A.T.M.A.N.V.
can you please provide more information as to what does not work exactly ? The neighbors don't find each other ? Payload traffic does not flow ? ...
Kind regards, Marek Lindner
b.a.t.m.a.n@lists.open-mesh.org
-
Alexey Ermakov -
Marek Lindner -
Sven Eckelmann