Any suggestion for combining enb, mme, sgw into a container of a pod?

[JC1O24]

Hi, I try to combine enb, mme, sgw into a container of a pod. And see you example in: https://github.com/electrocucaracha/gw-tester/tree/master/k8s/nsm/charts

The tunnels can be created, but seems can not send u plane to via gtp-sgw-s5. I think it may miss some network configuration.
Any ideas?

ip a

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
2: tunl0@NONE: <NOARP> mtu 1480 qdisc noop state DOWN group default qlen 1000
    link/ipip 0.0.0.0 brd 0.0.0.0
6: net1@if366: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc noqueue state UP group default
    link/ether ba:95:a1:45:8e:f8 brd ff:ff:ff:ff:ff:ff link-netnsid 0
    inet 172.22.0.13/24 brd 172.22.0.255 scope global net1
       valid_lft forever preferred_lft forever
    inet 172.22.0.12/24 scope global secondary net1
       valid_lft forever preferred_lft forever
8: net2@if367: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc noqueue state UP group default
    link/ether f6:9e:c8:bc:fa:2b brd ff:ff:ff:ff:ff:ff link-netnsid 0
    inet 172.21.0.13/24 brd 172.21.0.255 scope global net2
       valid_lft forever preferred_lft forever
    inet 172.21.0.16/24 scope global secondary net2
       valid_lft forever preferred_lft forever
10: net3@if368: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc noqueue state UP group default
    link/ether 2a:ae:54:de:54:9e brd ff:ff:ff:ff:ff:ff link-netnsid 0
    inet 172.25.0.13/24 brd 172.25.0.255 scope global net3
       valid_lft forever preferred_lft forever
12: net4@if369: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc noqueue state UP group default
    link/ether a2:2d:16:85:5d:cd brd ff:ff:ff:ff:ff:ff link-netnsid 0
    inet 172.25.1.13/24 brd 172.25.1.255 scope global net4
       valid_lft forever preferred_lft forever
14: net5@if370: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc noqueue state UP group default
    link/ether 52:05:45:da:8d:e6 brd ff:ff:ff:ff:ff:ff link-netnsid 0
    inet 172.21.1.12/24 brd 172.21.1.255 scope global net5
       valid_lft forever preferred_lft forever
    inet 172.21.1.16/24 scope global secondary net5
       valid_lft forever preferred_lft forever
16: net6@if371: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc noqueue state UP group default
    link/ether 1a:4f:b7:7c:f3:64 brd ff:ff:ff:ff:ff:ff link-netnsid 0
    inet 10.0.0.254/24 brd 10.0.0.255 scope global net6
       valid_lft forever preferred_lft forever
    inet 10.0.0.202/24 brd 10.0.0.255 scope global secondary net6
       valid_lft forever preferred_lft forever
    inet 10.0.0.201/24 scope global secondary net6
       valid_lft forever preferred_lft forever
17: gtp-sgw-s1: <POINTOPOINT,MULTICAST,NOARP,UP,LOWER_UP> mtu 1500 qdisc noqueue state UNKNOWN group default
    link/none
18: gtp-sgw-s5: <POINTOPOINT,MULTICAST,NOARP,UP,LOWER_UP> mtu 1500 qdisc noqueue state UNKNOWN group default
    link/none
24: gtp-enb: <POINTOPOINT,MULTICAST,NOARP,UP,LOWER_UP> mtu 1500 qdisc noqueue state UNKNOWN group default
    link/none

ip r

10.0.0.0/24 dev net6 proto kernel scope link src 10.0.0.254
172.21.0.0/24 dev net2 proto kernel scope link src 172.21.0.13
172.21.1.0/24 dev net5 proto kernel scope link src 172.21.1.12
172.22.0.0/24 dev net1 proto kernel scope link src 172.22.0.13
172.25.0.0/24 dev net3 proto kernel scope link src 172.25.0.13
172.25.1.0/24 dev net4 proto kernel scope link src 172.25.1.13

More details: cncf/cnf-testbed#370 (comment)

[electrocucaracha]

Hi @JC1O24,

First of all, I'm not sure why you're combining those services in a single pod but I'm sure that you have your own reasons. These are few things that came to my mind,

• eNodeB pod must start after the rest of the components are up and running. I assume that you have consider this before given that your logs are showing a successful creation of sessions.
• External clients and HTTP servers must have additional routes to address properly the cURL calls.
• eNodeB pod has to include the IP address of the external client in the subscribers list.

Maybe you can initially test your setup using docker-compose files like this one.

Last but not least, I have included some TravisCI jobs to validate different scenarios. They provide valuable logs information which may deserved to take a look.