Bug 2060932

Description of problem:-

Failing to build images due to mtu issue on cni-podman bridge in latest version of buildah(buildah-1.24.1-1.el9.x86_64), Higher mtu set on cni-podman0 than the interface on instance by default.

* dnf install during image build is failing with timeout - Unable to curl the repo metadata:-
~~~  
masked-repo                         2.7 kB/s | 365 kB     02:16    
Errors during downloading metadata for repository 'masked-repo':
  - Curl error (28): Timeout was reached for <url masked for sanity> [Operation too slow. Less than 1000 bytes/sec transferred the last 30 seconds]
~~~

During our checks, we found in commit[A] - buildah have moved from using Host Namespace to using private Namespace. Now buildah reads config from /usr/share/containers/containers.conf which defaults to private .

[A] https://github.com/containers/buildah/commit/05d5d22dc6b76efa4fd1b1c1eccec2fbbb2f404a

buildah inspect output (when using older buildah-1.24.0-0.15.el9.x86_64)
~~~
         {
             "Name": "network",
             "Host": true,   >>>>>>>>>>>
             "Path": ""
        }
~~~   

buildah inspect output (when using newer buildah-1.24.1-1.el9.x86_64)

~~~
        {
             "Name": "network",
             "Host": false,    >>>>>>>>
             "Path": ""
        }
~~~      

Actual results:

* Our instance where we run build images for Openstack is hosted in an internal Openstack cloud which uses private network for connectivity (where mtu is 1450).
* With new buildah, When we try to build images as root user, a cni-podman0 bridge is created which have mtu of 1500. In older version cni-podman0 was not used to create and host networking was used instead.

~~~
# ip link | grep mtu
1: lo: <LOOPBACK,UP,LOWER_UP> mtu 65536 qdisc noqueue state UNKNOWN mode DEFAULT group default qlen 1000
2: eth0: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1450 qdisc fq_codel state UP mode DEFAULT group default qlen 1000 >>>> 1450 on actual interface
3: ovs-system: <BROADCAST,MULTICAST> mtu 1500 qdisc noop state DOWN mode DEFAULT group default qlen 1000
4: br-ex: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1350 qdisc noqueue state UNKNOWN mode DEFAULT group default qlen 1000
48: cni-podman0: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc noqueue state UP mode DEFAULT group default qlen 1000 >>>> 1500 mtu on cni-podman0
54: veth9646aed9@if2: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc noqueue master cni-podman0 state UP mode DEFAULT group default 
~~~
* We could not curl the repo metadata url because of higher mtu set on cni-podman bridge / ping to outside network with mtu > 1422 was also failing.



Expected results:-

* Buildah should set correct mtu on cni-podman based on interface on the machine


Version-Release number of selected component (if applicable):

RHEL-9 Beta 

Last good known version:-
buildah-1.24.0-0.15.el9.x86_64

Affected version:-
buildah-1.24.1-1.el9.x86_64

How reproducible:

Everytime

Steps to Reproduce:
1) Run an instance in Openstack environment where mtu on interface is set lower than 1500. (It was 1450 in our case)
2) Install buildah-1.24.0-0.15.el9.x86_64
3) Build container using buildah and try Pull content from outside / Ping outside network with mtu 1500



Additional info:

Workaround:-

1) When we set lower mtu(1400/1450) on cni-podman0 - we were able to successfully curl/ pull metadata for repos and image build was successful.

buidah provides way to configuring mtu on bridge via setting mtu in  /etc/cni/net.d/87-podman-bridge.conflist

2) Downgrade to older version buildah-1.24.0-0.15.el9.x86_64
3) Use --network=host when running buildah bud