Update parsing of ip addr show (SC-723)

[TheRealFalcon]

Proposed Commit Message

Update parsing of ip addr show

"ip addr show" may include a metric on the inet line.
Parse this while also keeping backwards compatibility
for lines without.

Additional Context

Example ip addr show on Jammy:

root@jammy:~# ip addr show
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
118: eth0@if119: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc noqueue state UP group default qlen 1000
    link/ether 00:16:3e:80:4f:ea brd ff:ff:ff:ff:ff:ff link-netnsid 0
    inet 10.85.130.57/24 metric 100 brd 10.85.130.255 scope global dynamic eth0
       valid_lft 2912sec preferred_lft 2912sec
    inet6 fd42:baa2:3dd:17a:216:3eff:fe80:4fea/64 scope global dynamic mngtmpaddr noprefixroute 
       valid_lft 3545sec preferred_lft 3545sec
    inet6 fe80::216:3eff:fe80:4fea/64 scope link 
       valid_lft forever preferred_lft forever

Notice the metric 100 on the inet line.

Test Steps

Can be seen in jammy integration tests on LXD as a new unexpected warning.

Checklist:

• My code follows the process laid out in the documentation
• I have updated or added any unit tests accordingly
• I have updated or added any documentation accordingly

[TheRealFalcon]

pytest parametrization doesn't work with the base class, so I removed the base class from the tests.

The only functional test change is in test_netdev_iproute_pformat

[holmanb]

Are the backslashes on lines 3 and 4 a typo?

[TheRealFalcon]

I just copied the sample-ipaddrshow-output file and added the metric part

[holmanb]

I see, nevermind. It doesn't matter what it says after scope since the regex ends with .*.

[holmanb]

I did a little bit of manual testing with this regex and didn't spot any issues.

[blackboxsw]

The bigger this regex gets, the more it makes me question whether we should really be extending a fragile regex parsing of human-readable output of ip. While I originally weighed in to the contrary because cloud-init is supported even more broadly than just Ubuntu, I think it would probably be preferable given that we attempt to try ip -json format first, and fallback to the legacy parsing if -json param is not present. This way most of the latest *nix distros with recent ip tooling can benefit from simple parsing of machine-readable output and any older distributions will continue to parse correctly with the older ip tooling and output formats.

[holmanb]

lgtm

[mwhudson]

Is there a reason you are not parsing the output of 'ip -json addr show' instead?

[TheRealFalcon]

@mwhudson It was proposed, but we decided against it due to lack of availability everywhere.

[mwhudson]

@mwhudson It was proposed, but we decided against it due to lack of availability everywhere.

Ah yes I see it's not supported in xenial.

[TheRealFalcon]

Closing in favor of #1210