Fix macaddr bug #102

Sabaniki · 2022-12-05T10:27:08Z

このパッチは nodes > interface > addr フィールドを利用してインターフェイスの MAC アドレスを指定する機能のバグの修正です．

このバグは，yaml ファイルで指定された MAC アドレスを割り振る処理を N2nLink 関数の中で行っていたために発生していました．

nodes:
- name: Router
  image: sabaniki/frr:latest
  interfaces:
  - { name: vm1, type: direct, args: VM1#rt, addr: 10:00:00:00:00:10 }
  - { name: vm2, type: direct, args: VM2#rt, addr: 20:00:00:00:00:20 }
  - { name: vm3, type: direct, args: VM3#rt, addr: 30:00:00:00:00:30 }

- name: VM1
  image: sabaniki/frr:latest
  interfaces:
  - { name: rt, type: direct, args: Router#vm1, addr: 10:10:10:10:10:10 }

- name: VM2
  image: sabaniki/frr:latest
  interfaces:
  - { name: rt, type: direct, args: Router#vm2, addr: 20:20:20:20:20:20 }

- name: VM3
  image: sabaniki/frr:latest
  interfaces:
  - { name: rt, type: direct, args: Router#vm3, addr: 30:30:30:30:30:30 }

上記の spec.yaml に対して tinet up を実行すると，出力は以下の通りとなり，VM 側のインターフェイスの MAC アドレスが指定されていません．

docker run -td --net none --name Router --rm --privileged --hostname Router -v /tmp/tinet:/tinet sabaniki/frr:latest > /dev/null
mkdir -p /var/run/netns > /dev/null
PID=`docker inspect Router --format '{{.State.Pid}}'` > /dev/null
ln -s /proc/$PID/ns/net /var/run/netns/Router > /dev/null
docker run -td --net none --name VM1 --rm --privileged --hostname VM1 -v /tmp/tinet:/tinet sabaniki/frr:latest > /dev/null
mkdir -p /var/run/netns > /dev/null
PID=`docker inspect VM1 --format '{{.State.Pid}}'` > /dev/null
ln -s /proc/$PID/ns/net /var/run/netns/VM1 > /dev/null
docker run -td --net none --name VM2 --rm --privileged --hostname VM2 -v /tmp/tinet:/tinet sabaniki/frr:latest > /dev/null
mkdir -p /var/run/netns > /dev/null
PID=`docker inspect VM2 --format '{{.State.Pid}}'` > /dev/null
ln -s /proc/$PID/ns/net /var/run/netns/VM2 > /dev/null
docker run -td --net none --name VM3 --rm --privileged --hostname VM3 -v /tmp/tinet:/tinet sabaniki/frr:latest > /dev/null
mkdir -p /var/run/netns > /dev/null
PID=`docker inspect VM3 --format '{{.State.Pid}}'` > /dev/null
ln -s /proc/$PID/ns/net /var/run/netns/VM3 > /dev/null
ip link add vm1 netns Router type veth peer name rt netns VM1 > /dev/null
ip netns exec Router ip link set vm1 up > /dev/null
ip netns exec VM1 ip link set rt up > /dev/null
ip netns exec Router ip link set vm1 address 10:00:00:00:00:10 > /dev/null
ip link add vm2 netns Router type veth peer name rt netns VM2 > /dev/null
ip netns exec Router ip link set vm2 up > /dev/null
ip netns exec VM2 ip link set rt up > /dev/null
ip netns exec Router ip link set vm2 address 20:00:00:00:00:20 > /dev/null
ip link add vm3 netns Router type veth peer name rt netns VM3 > /dev/null
ip netns exec Router ip link set vm3 up > /dev/null
ip netns exec VM3 ip link set rt up > /dev/null
ip netns exec Router ip link set vm3 address 30:00:00:00:00:30 > /dev/null
ip netns del Router > /dev/null
ip netns del VM1 > /dev/null
ip netns del VM2 > /dev/null
ip netns del VM3 > /dev/null

N2n関数は CmdUp 関数の中でノードを接続するために実行されます．Router#vm1 と VM1#rt をつなげる際，Router#vm1 に対してのみ実行され，VM1#rt に対してはスキップされます．そのため，VM1#rt に対して MAC アドレスを指定する処理が行われません．

バグ修正後の実行結果は以下のとおりです．

docker run -td --net none --name Router --rm --privileged --hostname Router -v /tmp/tinet:/tinet sabaniki/frr:latest > /dev/null
mkdir -p /var/run/netns > /dev/null
PID=`docker inspect Router --format '{{.State.Pid}}'` > /dev/null
ln -s /proc/$PID/ns/net /var/run/netns/Router > /dev/null
docker run -td --net none --name VM1 --rm --privileged --hostname VM1 -v /tmp/tinet:/tinet sabaniki/frr:latest > /dev/null
mkdir -p /var/run/netns > /dev/null
PID=`docker inspect VM1 --format '{{.State.Pid}}'` > /dev/null
ln -s /proc/$PID/ns/net /var/run/netns/VM1 > /dev/null
docker run -td --net none --name VM2 --rm --privileged --hostname VM2 -v /tmp/tinet:/tinet sabaniki/frr:latest > /dev/null
mkdir -p /var/run/netns > /dev/null
PID=`docker inspect VM2 --format '{{.State.Pid}}'` > /dev/null
ln -s /proc/$PID/ns/net /var/run/netns/VM2 > /dev/null
docker run -td --net none --name VM3 --rm --privileged --hostname VM3 -v /tmp/tinet:/tinet sabaniki/frr:latest > /dev/null
mkdir -p /var/run/netns > /dev/null
PID=`docker inspect VM3 --format '{{.State.Pid}}'` > /dev/null
ln -s /proc/$PID/ns/net /var/run/netns/VM3 > /dev/null
ip link add vm1 netns Router type veth peer name rt netns VM1 > /dev/null
ip netns exec Router ip link set vm1 up > /dev/null
ip netns exec VM1 ip link set rt up > /dev/null
ip netns exec Router ip link set vm1 address 10:00:00:00:00:10 > /dev/null
ip link add vm2 netns Router type veth peer name rt netns VM2 > /dev/null
ip netns exec Router ip link set vm2 up > /dev/null
ip netns exec VM2 ip link set rt up > /dev/null
ip netns exec Router ip link set vm2 address 20:00:00:00:00:20 > /dev/null
ip link add vm3 netns Router type veth peer name rt netns VM3 > /dev/null
ip netns exec Router ip link set vm3 up > /dev/null
ip netns exec VM3 ip link set rt up > /dev/null
ip netns exec Router ip link set vm3 address 30:00:00:00:00:30 > /dev/null
+ ip netns exec VM1 ip link set rt address 10:10:10:10:10:10 > /dev/null
+ ip netns exec VM2 ip link set rt address 20:20:20:20:20:20 > /dev/null
+ ip netns exec VM3 ip link set rt address 30:30:30:30:30:30 > /dev/null
ip netns del Router > /dev/null
ip netns del VM1 > /dev/null
ip netns del VM2 > /dev/null
ip netns del VM3 > /dev/null

また，プログラム上は MAC アドレスの指定はaddr: XX:XX:XX:XX:XX:XXで行われることを想定していますが，ドキュメント上はmac: XX:XX:XX:XX:XX:XXとなっていたため，これについても修正しました．

slankdev · 2022-12-05T11:57:40Z

ありがとうございます!
方針として完全に理解していて, 良いと思うんですが, このshell scriptって成功しますかね.
というのも, ip link set NAME address ADDR って, link downしていないと確か成功しなかった記憶があります.

それ以外はLGTMです

Sabaniki · 2022-12-05T12:16:56Z

ありがとうございます! 方針として完全に理解していて, 良いと思うんですが, このshell scriptって成功しますかね. というのも, ip link set NAME address ADDR って, link downしていないと確か成功しなかった記憶があります.

それ以外はLGTMです

以下が実際の実行結果です．

+ docker run -td --net none --name Router --rm --privileged --hostname Router -v /tmp/tinet:/tinet sabaniki/frr:latest
+ mkdir -p /var/run/netns
+ docker inspect Router --format {{.State.Pid}}
+ PID=605795
+ ln -s /proc/605795/ns/net /var/run/netns/Router
+ docker run -td --net none --name VM1 --rm --privileged --hostname VM1 -v /tmp/tinet:/tinet sabaniki/frr:latest
+ mkdir -p /var/run/netns
+ docker inspect VM1 --format {{.State.Pid}}
+ PID=605997
+ ln -s /proc/605997/ns/net /var/run/netns/VM1
+ docker run -td --net none --name VM2 --rm --privileged --hostname VM2 -v /tmp/tinet:/tinet sabaniki/frr:latest
+ mkdir -p /var/run/netns
+ docker inspect VM2 --format {{.State.Pid}}
+ PID=606239
+ ln -s /proc/606239/ns/net /var/run/netns/VM2
+ docker run -td --net none --name VM3 --rm --privileged --hostname VM3 -v /tmp/tinet:/tinet sabaniki/frr:latest
+ mkdir -p /var/run/netns
+ docker inspect VM3 --format {{.State.Pid}}
+ PID=606475
+ ln -s /proc/606475/ns/net /var/run/netns/VM3
+ ip link add vm1 netns Router type veth peer name rt netns VM1
+ ip netns exec Router ip link set vm1 up
+ ip netns exec VM1 ip link set rt up
+ ip netns exec Router ip link set vm1 address 10:00:00:00:00:10
+ ip link add vm2 netns Router type veth peer name rt netns VM2
+ ip netns exec Router ip link set vm2 up
+ ip netns exec VM2 ip link set rt up
+ ip netns exec Router ip link set vm2 address 20:00:00:00:00:20
+ ip link add vm3 netns Router type veth peer name rt netns VM3
+ ip netns exec Router ip link set vm3 up
+ ip netns exec VM3 ip link set rt up
+ ip netns exec Router ip link set vm3 address 30:00:00:00:00:30
+ ip netns exec VM1 ip link set rt address 10:10:10:10:10:10
+ ip netns exec VM2 ip link set rt address 20:20:20:20:20:20
+ ip netns exec VM3 ip link set rt address 30:30:30:30:30:30
+ ip netns del Router
+ ip netns del VM1
+ ip netns del VM2
+ ip netns del VM3

  ~/test ································································································ sabaniki@dev-local
❯ docker exec -it Router ip -br l
lo               UNKNOWN        00:00:00:00:00:00 <LOOPBACK,UP,LOWER_UP>
pimreg@NONE      UNKNOWN        <NOARP,UP,LOWER_UP>
vm1@if3          UP             10:00:00:00:00:10 <BROADCAST,MULTICAST,UP,LOWER_UP>
vm2@if3          UP             20:00:00:00:00:20 <BROADCAST,MULTICAST,UP,LOWER_UP>
vm3@if2          UP             30:00:00:00:00:30 <BROADCAST,MULTICAST,UP,LOWER_UP>

  ~/test ································································································ sabaniki@dev-local
❯ docker exec -it VM1 ip -br l
lo               UNKNOWN        00:00:00:00:00:00 <LOOPBACK,UP,LOWER_UP>
pimreg@NONE      UNKNOWN        <NOARP,UP,LOWER_UP>
rt@if3           UP             10:10:10:10:10:10 <BROADCAST,MULTICAST,UP,LOWER_UP>

  ~/test ································································································ sabaniki@dev-local
❯ docker exec -it VM2 ip -br l
lo               UNKNOWN        00:00:00:00:00:00 <LOOPBACK,UP,LOWER_UP>
pimreg@NONE      UNKNOWN        <NOARP,UP,LOWER_UP>
rt@if4           UP             20:20:20:20:20:20 <BROADCAST,MULTICAST,UP,LOWER_UP>

  ~/test ································································································ sabaniki@dev-local
❯ docker exec -it VM3 ip -br l
lo               UNKNOWN        00:00:00:00:00:00 <LOOPBACK,UP,LOWER_UP>
rt@if5           UP             30:30:30:30:30:30 <BROADCAST,MULTICAST,UP,LOWER_UP>
pimreg@NONE      UNKNOWN        <NOARP,UP,LOWER_UP>

正常に MAC アドレスが変更されていることがわかります．
また，以下のコマンドの実行結果から，link を down せずとも MAC アドレスを変更できることを確認しました

❯ docker exec -it VM1 ip -br l
lo               UNKNOWN        00:00:00:00:00:00 <LOOPBACK,UP,LOWER_UP>
pimreg@NONE      UNKNOWN        <NOARP,UP,LOWER_UP>
rt@if3           UP             10:10:10:10:10:10 <BROADCAST,MULTICAST,UP,LOWER_UP>

  ~/test ································································································ sabaniki@dev-local
❯ docker exec -it VM1 ip link set addr 10:10:10:10:10:21 dev rt


  ~/test ································································································ sabaniki@dev-local
❯ docker exec -it VM1 ip -br l
lo               UNKNOWN        00:00:00:00:00:00 <LOOPBACK,UP,LOWER_UP>
pimreg@NONE      UNKNOWN        <NOARP,UP,LOWER_UP>
rt@if3           UP             10:10:10:10:10:21 <BROADCAST,MULTICAST,UP,LOWER_UP>

また，以下のように Network Namespace 内で IP コマンドを実行しても同様に変更されることを確認しました．

  ~/test ································································································ sabaniki@dev-local
❯ sudo ip netns add test1

  ~/test ··························································································· sabaniki@dev-local
❯ sudo ip netns add test2

  ~/test ·························································································· ✘ 255 sabaniki@dev-local
❯ sudo ip link add eth0-t1 type veth peer name eth0-t2

  ~/test ································································································ sabaniki@dev-local
❯ sudo ip link set eth0-t1 netns test1

  ~/test ································································································ sabaniki@dev-local
❯ sudo ip link set eth0-t2 netns test2

  ~/test ································································································ sabaniki@dev-local
❯ sudo ip netns exec test1 ip link set up dev eth0-t1

  ~/test ································································································ sabaniki@dev-local
❯ sudo ip netns exec test2 ip link set up dev eth0-t2

  ~/test ································································································ sabaniki@dev-local
❯ sudo ip netns exec test1 ip -br l
lo               DOWN           00:00:00:00:00:00 <LOOPBACK>
eth0-t1@if33     UP             72:9f:e4:ea:02:ed <BROADCAST,MULTICAST,UP,LOWER_UP>

  ~/test ································································································ sabaniki@dev-local
❯ sudo ip netns exec test2 ip -br l
lo               DOWN           00:00:00:00:00:00 <LOOPBACK>
eth0-t2@if34     UP             1a:47:d2:d2:c5:1e <BROADCAST,MULTICAST,UP,LOWER_UP>

  ~/test ································································································ sabaniki@dev-local
❯ sudo ip netns exec test1 ip link set addr 10:00:00:00:00:10 dev eth0-t1

  ~/test ································································································ sabaniki@dev-local
❯ sudo ip netns exec test1 ip -br l
lo               DOWN           00:00:00:00:00:00 <LOOPBACK>
eth0-t1@if33     UP             10:00:00:00:00:10 <BROADCAST,MULTICAST,UP,LOWER_UP>

これはあくまでも私の予想ですが，veth に関しては MAC アドレスを変更する際，リンクダウンさせることは必須ではないのではないか，と考えています．

追加の調査などが必要であれば行います．

slankdev

LGTM

tinynetwork#102

Sabaniki added 2 commits December 5, 2022 09:34

fix bug

aaa333a

fix docs

1f71972

Sabaniki requested review from ak1ra24 and slankdev as code owners December 5, 2022 10:27

slankdev approved these changes Dec 7, 2022

View reviewed changes

slankdev merged commit 6d9b552 into tinynetwork:master Dec 7, 2022

kimitoboku added a commit to kimitoboku/tinet-go that referenced this pull request Sep 12, 2023

Fix test fail N2nLink is not handle MAC addr now

0c64089

tinynetwork#102

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Fix macaddr bug #102

Fix macaddr bug #102

Sabaniki commented Dec 5, 2022 •

edited

Loading

slankdev commented Dec 5, 2022

Sabaniki commented Dec 5, 2022 •

edited

Loading

slankdev left a comment

Fix macaddr bug #102

Fix macaddr bug #102

Conversation

Sabaniki commented Dec 5, 2022 • edited Loading

slankdev commented Dec 5, 2022

Sabaniki commented Dec 5, 2022 • edited Loading

slankdev left a comment

Choose a reason for hiding this comment

Sabaniki commented Dec 5, 2022 •

edited

Loading

Sabaniki commented Dec 5, 2022 •

edited

Loading