【Troubleshooting and Questions】Kubernetes Pod-to-Pod Communication Cross Nodes

Question

Troubleshooting Pod-to-Pod Communication Cross Nodes

Prerequisite:

I use VirtualBox to setup 2 VMs. VirtualBox version is 6.1.16 r140961 (Qt5.6.2). The OS is CentOS-7(CentOS-7-x86_64-Minimal-2003.iso). One VM as kubernetes master node, named k8s-master1. Another VM as kubernetes worker node, named k8s-node1.
Attach 2 network interfaces, NAT + Host-Only. NAT is used for internet facing. Host-Only is used for connections among VMs.

The network information on k8s-master1.

[root@k8s-master1 ~]# ip addr
1: lo: <LOOPBACK,UP,LOWER_UP> mtu 65536 qdisc noqueue state UNKNOWN group default qlen 1000
    link/loopback 00:00:00:00:00:00 brd 00:00:00:00:00:00
    inet 127.0.0.1/8 scope host lo
       valid_lft forever preferred_lft forever
    inet6 ::1/128 scope host 
       valid_lft forever preferred_lft forever
2: enp0s3: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc pfifo_fast state UP group default qlen 1000
    link/ether 08:00:27:d1:41:4b brd ff:ff:ff:ff:ff:ff
    inet 192.168.56.101/24 brd 192.168.56.255 scope global noprefixroute enp0s3
       valid_lft forever preferred_lft forever
    inet6 fe80::62ae:c676:da76:cbff/64 scope link noprefixroute 
       valid_lft forever preferred_lft forever
3: enp0s8: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc pfifo_fast state UP group default qlen 1000
    link/ether 08:00:27:8a:2d:63 brd ff:ff:ff:ff:ff:ff
    inet 10.0.3.15/24 brd 10.0.3.255 scope global noprefixroute dynamic enp0s8
       valid_lft 80380sec preferred_lft 80380sec
    inet6 fe80::952e:9af8:a1cb:8a07/64 scope link noprefixroute 
       valid_lft forever preferred_lft forever
4: docker0: <NO-CARRIER,BROADCAST,MULTICAST,UP> mtu 1500 qdisc noqueue state DOWN group default 
    link/ether 02:42:04:6e:c7:dc brd ff:ff:ff:ff:ff:ff
    inet 172.17.0.1/16 brd 172.17.255.255 scope global docker0
       valid_lft forever preferred_lft forever
5: flannel.1: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1450 qdisc noqueue state UNKNOWN group default 
    link/ether 06:cd:b1:24:62:fc brd ff:ff:ff:ff:ff:ff
    inet 10.244.0.0/32 scope global flannel.1
       valid_lft forever preferred_lft forever
    inet6 fe80::4cd:b1ff:fe24:62fc/64 scope link 
       valid_lft forever preferred_lft forever
6: cni0: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1450 qdisc noqueue state UP group default qlen 1000
    link/ether 06:75:06:d0:17:ee brd ff:ff:ff:ff:ff:ff
    inet 10.244.0.1/24 brd 10.244.0.255 scope global cni0
       valid_lft forever preferred_lft forever
    inet6 fe80::475:6ff:fed0:17ee/64 scope link 
       valid_lft forever preferred_lft forever
8: veth221fb276@if3: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1450 qdisc noqueue master cni0 state UP group default 
    link/ether 62:b9:61:95:0f:73 brd ff:ff:ff:ff:ff:ff link-netnsid 1
    inet6 fe80::60b9:61ff:fe95:f73/64 scope link 
       valid_lft forever preferred_lft forever

The network information on k8s-node1.

[root@k8s-node1 ~]# ip addr
1: lo: <LOOPBACK,UP,LOWER_UP> mtu 65536 qdisc noqueue state UNKNOWN group default qlen 1000
    link/loopback 00:00:00:00:00:00 brd 00:00:00:00:00:00
    inet 127.0.0.1/8 scope host lo
       valid_lft forever preferred_lft forever
    inet6 ::1/128 scope host 
       valid_lft forever preferred_lft forever
2: enp0s3: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc pfifo_fast state UP group default qlen 1000
    link/ether 08:00:27:ba:34:4f brd ff:ff:ff:ff:ff:ff
    inet 192.168.56.102/24 brd 192.168.56.255 scope global noprefixroute enp0s3
       valid_lft forever preferred_lft forever
    inet6 fe80::660d:ec2:cb1c:49de/64 scope link noprefixroute 
       valid_lft forever preferred_lft forever
3: enp0s8: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc pfifo_fast state UP group default qlen 1000
    link/ether 08:00:27:48:6d:8b brd ff:ff:ff:ff:ff:ff
    inet 10.0.3.15/24 brd 10.0.3.255 scope global noprefixroute dynamic enp0s8
       valid_lft 80039sec preferred_lft 80039sec
    inet6 fe80::62af:7770:3b6d:6576/64 scope link noprefixroute 
       valid_lft forever preferred_lft forever
4: docker0: <NO-CARRIER,BROADCAST,MULTICAST,UP> mtu 1500 qdisc noqueue state DOWN group default 
    link/ether 02:42:e4:65:f4:28 brd ff:ff:ff:ff:ff:ff
    inet 172.17.0.1/16 brd 172.17.255.255 scope global docker0
       valid_lft forever preferred_lft forever
5: flannel.1: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1450 qdisc noqueue state UNKNOWN group default 
    link/ether 8e:d3:79:63:40:22 brd ff:ff:ff:ff:ff:ff
    inet 10.244.1.0/32 scope global flannel.1
       valid_lft forever preferred_lft forever
    inet6 fe80::8cd3:79ff:fe63:4022/64 scope link 
       valid_lft forever preferred_lft forever
6: cni0: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1450 qdisc noqueue state UP group default qlen 1000
    link/ether 8e:22:8d:e5:29:24 brd ff:ff:ff:ff:ff:ff
    inet 10.244.1.1/24 brd 10.244.1.255 scope global cni0
       valid_lft forever preferred_lft forever
    inet6 fe80::8c22:8dff:fee5:2924/64 scope link 
       valid_lft forever preferred_lft forever
8: veth5314dbaf@if3: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1450 qdisc noqueue master cni0 state UP group default 
    link/ether 82:58:1c:3a:a1:a3 brd ff:ff:ff:ff:ff:ff link-netnsid 1
    inet6 fe80::8058:1cff:fe3a:a1a3/64 scope link 
       valid_lft forever preferred_lft forever
10: calic849a8cefe4@if3: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1450 qdisc noqueue state UP group default 
    link/ether ee:ee:ee:ee:ee:ee brd ff:ff:ff:ff:ff:ff link-netnsid 2
    inet6 fe80::ecee:eeff:feee:eeee/64 scope link 
       valid_lft forever preferred_lft forever
11: cali82f9786c604@if3: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1450 qdisc noqueue state UP group default 
    link/ether ee:ee:ee:ee:ee:ee brd ff:ff:ff:ff:ff:ff link-netnsid 3
    inet6 fe80::ecee:eeff:feee:eeee/64 scope link 
       valid_lft forever preferred_lft forever
12: caliaa01fae8214@if3: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1450 qdisc noqueue state UP group default 
    link/ether ee:ee:ee:ee:ee:ee brd ff:ff:ff:ff:ff:ff link-netnsid 4
    inet6 fe80::ecee:eeff:feee:eeee/64 scope link 
       valid_lft forever preferred_lft forever
13: cali88097e17cd0@if3: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1450 qdisc noqueue state UP group default 
    link/ether ee:ee:ee:ee:ee:ee brd ff:ff:ff:ff:ff:ff link-netnsid 5
    inet6 fe80::ecee:eeff:feee:eeee/64 scope link 
       valid_lft forever preferred_lft forever
15: cali788346fba46@if3: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1450 qdisc noqueue state UP group default 
    link/ether ee:ee:ee:ee:ee:ee brd ff:ff:ff:ff:ff:ff link-netnsid 0
    inet6 fe80::ecee:eeff:feee:eeee/64 scope link 
       valid_lft forever preferred_lft forever
16: caliafdfba3871a@if3: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1450 qdisc noqueue state UP group default 
    link/ether ee:ee:ee:ee:ee:ee brd ff:ff:ff:ff:ff:ff link-netnsid 6
    inet6 fe80::ecee:eeff:feee:eeee/64 scope link 
       valid_lft forever preferred_lft forever
18: cali329803e4ee5@if3: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1450 qdisc noqueue state UP group default 
    link/ether ee:ee:ee:ee:ee:ee brd ff:ff:ff:ff:ff:ff link-netnsid 7
    inet6 fe80::ecee:eeff:feee:eeee/64 scope link 
       valid_lft forever preferred_lft forever

I make a table to simplify the network information.

| k8s-master1                               | k8s-node1                                  |
| ----------------------------------------- | ------------------------------------------ |
| NAT               10.0.3.15/24    enp0s8  | NAT                10.0.3.15/24    enp0s8  |
| Host-Only     192.168.56.101/24    enp0s3 | Host-Only      192.168.56.102/24    enp0s3 |
| docker0        172.17.0.1/16              | docker0         172.17.0.1/16              |
| flannel.1       10.244.1.0/32             | flannel.1        10.244.1.0/32             |
| cni0               10.244.1.1/24          | cni0                10.244.1.1/24          |

Problem Description

Test case1:

Expose Kubernetes Service nginx-svc as NodePort, and Pod nginx is running on k8s-node1。

View the NodePort of nginx-svc

[root@k8s-master1 ~]# kubectl get svc 
NAME                    TYPE        CLUSTER-IP       EXTERNAL-IP   PORT(S)        AGE
nginx-svc               NodePort    10.103.236.60    <none>        80:30309/TCP   9h

Run command curl http://localhost:30309 on k8s-master1. Finally, it returns Connection timed out.

Run command curl http://localhost:30309 on k8s-node1. It returns response.

Test case 2:

I created another Pod nginx on k8s-master1. There is already a Pod nginx on k8s-node1. Make a call from the Pod nginx on k8s-master1 to the Pod nginx on k8s-node1. It also Connection timed out. The problem is very likely to could not communicate cross node's pod in kubernetes cluster using flannel

Test case 3:

If 2 nginx pod on the same node, whatever k8s-master1 or k8s-node1. The connection is ok. The pod can get response from another pod.

Test case 4:

I did several ping requests from k8s-master1 to k8s-node1 using different IP address.

Ping k8s-node1's Host-Only IP address: Connection Success

ping 192.168.56.102

Ping k8s-node1's flannel.1 IP address: Connection Failed

ping 10.244.1.0

Ping k8s-node1's cni0 IP address: Connection Failed

ping 10.244.1.1

Test case 5:

I setup a new k8s cluster using bridge network on the same laptop. Test case 1,2,3,4 cannot be reproduced. No problem of 'Pod-to-Pod Communication Cross Nodes'.

Guess

Firewall. I disabled the firewall on all VMs. Also the firewall on laptop(Win 10), including any antivirus software. The connection problem is still there.
Compare the iptables between k8s cluster attached NAT & Host-Only with k8s cluster attached bridge network. Not found any big difference.
```
yum install -y net-tools
iptables -L
```

Solution

# install route
[root@k8s-master1 tzhong]# yum install -y net-tools

# View routing table after k8s cluster installed thru. kubeadm.
[root@k8s-master1 tzhong]# route
Kernel IP routing table
Destination     Gateway         Genmask         Flags Metric Ref    Use Iface
default         gateway         0.0.0.0         UG    100    0        0 enp0s3
10.0.2.0        0.0.0.0         255.255.255.0   U     100    0        0 enp0s3
10.244.0.0      0.0.0.0         255.255.255.0   U     0      0        0 cni0
10.244.1.0      10.244.1.0      255.255.255.0   UG    0      0        0 flannel.1
172.17.0.0      0.0.0.0         255.255.0.0     U     0      0        0 docker0
192.168.56.0    0.0.0.0         255.255.255.0   U     101    0        0 enp0s8

Focus on the below routing table rule. This rule means that all the requests from k8s-master1, and the destination is 10.244.1.0/24 will be forwarded by flannel.1.

But seems that the flannel.1 has some problems. So I just have a try. Add a new routing table rule that all the requests from k8s-master1, and the destination is 10.244.1.0/24 will be forwarded by enp0s8.

# Run below commands on k8s-master1 to add routing table rules. 10.244.1.0 is the IP address of flannel.1 on k8s-node1.
route add -net 10.244.1.0 netmask 255.255.255.0 enp0s8
route add -net 10.244.1.0 netmask 255.255.255.0 gw 10.244.1.0

[root@k8s-master1 ~]# route
Kernel IP routing table
Destination     Gateway         Genmask         Flags Metric Ref    Use Iface
default         gateway         0.0.0.0         UG    100    0        0 enp0s3
10.0.2.0        0.0.0.0         255.255.255.0   U     100    0        0 enp0s3
10.244.0.0      0.0.0.0         255.255.255.0   U     0      0        0 cni0
10.244.1.0      10.244.1.0      255.255.255.0   UG    0      0        0 enp0s8
10.244.1.0      0.0.0.0         255.255.255.0   U     0      0        0 enp0s8
10.244.1.0      10.244.1.0      255.255.255.0   UG    0      0        0 flannel.1
172.17.0.0      0.0.0.0         255.255.0.0     U     0      0        0 docker0
192.168.56.0    0.0.0.0         255.255.255.0   U     101    0        0 enp0s8

Two additional routing table rules has been added.

10.244.1.0      10.244.1.0      255.255.255.0   UG    0      0        0 enp0s8
10.244.1.0      0.0.0.0         255.255.255.0   U     0      0        0 enp0s8

Do the same way on k8s-node1.

# Run below commands on k8s-node1 to add routing table rules. 10.244.0.0 is the IP address of flannel.1 on k8s-master1.
route add -net 10.244.0.0 netmask 255.255.255.0 enp0s8
route add -net 10.244.0.0 netmask 255.255.255.0 gw 10.244.0.0

Do test case 1,2,3,4,5 again. All passed. I am confused, why? I have a couple of questions.

Questions:

Why the connection of flannel.1 is blocked? Any way/methods to investigation?
Is that a limitation of Host-Only network interface with flannel? If I change to other CNI plugins, the connection will be ok?
After add routing tables using enp0s8, why the connection becomes ok?

Looks like CNI-specific problem. Did you try different CNI e.g. Calico ? — mario, Nov 25 '20 at 21:20
Hi, any progress with solving this issue ? Was the try with Calico successfull ? — mario, Dec 17 '20 at 22:28
Sorry. Busy with other things. I will give feedback when I do a try, — Cain, Dec 18 '20 at 08:04
Finally, I found the root cause. I use VirtualBox to setup the VM, there is 2 network interface. One is NAT for the internet-facing, and another is Host-Only for k8s cluster communication. Flannel always uses the NAT. This is not correct. We need to config the correct network interface vaule in kube-flannel.yaml — Cain, Jan 23 '21 at 07:36

score 1 · Answer 1 · answered Jan 23 '21 at 07:39

Finally, I found the root cause. I use VirtualBox to setup the VM, there is 2 network interface. One is NAT for the internet-facing, and another is Host-Only for k8s cluster communication. Flannel always uses the NAT. This is not correct. We need to config the correct network interface vaule in kube-flannel.yaml

apiVersion: apps/v1
kind: DaemonSet
metadata:
  name: kube-flannel-ds
  namespace: kube-system
  ...
spec:
  ...
  template:
    ...
    spec:
      ...
      containers:
      - name: kube-flannel
        image: quay.io/coreos/flannel:v0.13.1-rc1
        command:
        - /opt/bin/flanneld
        args:
        - --ip-masq
        - --kube-subnet-mgr
        # THIS LINE
        - --iface=enp0s8