Implement YAML state tracking and use it in the DBus API and netplan-try (LP: #1943120) (FR-1745)

[slyon]

Description

Allow to pass an optional --state parameter to netplan try/apply describing a directory that contains a netplan configuration tree/state (i.e. /{etc,run,lib}/netplan/*.yaml).
Netplan will make use of this "old state" to calculate the delta of dropped interface definitions, like bridges/bonds/vlans/tunnels that have been configured before but are not part of the current YAML configuration anymore. It will then try to delete those virtual interfaces (via ip link del dev IFACE) if they still exist.

The same functionality is used to roll back a netplan try command that failed or was rejected.

Generally, the state needs to be provided manually. The DBus API (using io.netplan.Netplan.Config.Try/Apply) is an exception, as the previous state can be backed up automatically in this case.

Checklist

• Runs make check successfully.
• Retains 100% code coverage (make check-coverage).
• New/changed keys in YAML format are documented.
• (Optional) Adds example YAML for new feature.
• (Optional) Closes an open bug in Launchpad: LP#1943120

[codecov-commenter]

Codecov Report

Merging #231 (3f859f0) into main (17d3848) will increase coverage by 0.00%.
The diff coverage is 100.00%.

@@           Coverage Diff           @@
##             main     #231   +/-   ##
=======================================
  Coverage   99.04%   99.05%           
=======================================
  Files          57       57           
  Lines        9524     9592   +68     
=======================================
+ Hits         9433     9501   +68     
  Misses         91       91           

Impacted Files	Coverage Δ
netplan/cli/commands/apply.py	100.00% <100.00%> (ø)
netplan/cli/commands/try_command.py	100.00% <100.00%> (ø)
netplan/configmanager.py	100.00% <100.00%> (ø)
src/dbus.c	100.00% <100.00%> (ø)
tests/dbus/test_dbus.py	100.00% <100.00%> (ø)
tests/test_cli_units.py	100.00% <100.00%> (ø)
tests/test_configmanager.py	100.00% <100.00%> (ø)

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update 17d3848...3f859f0. Read the comment docs.

[schopin-pro]

I have a few non-blocking nitpicks but all in all LGTM.

Have you run the integration tests?

[slyon]

Thanks for your feedback. I've made the devices parameter a named argument and added the extra code path and relevant testing around it.

As this have only been nitpicks and the integration tests are passing (incl. the new one that explicitly checks for this case), too, let's get this merged.

$ time autopkgtest ../netplan.io_0.103-0ubuntu6_amd64.changes -U -- qemu /data/autopkgtest-impish-amd64.img
autopkgtest [15:39:14]: starting date: 2021-09-27
autopkgtest [15:39:14]: version 5.16ubuntu1
autopkgtest [15:39:14]: host abaconcy; command line: /usr/bin/autopkgtest ../netplan.io_0.103-0ubuntu6_amd64.changes -U -- qemu /data/autopkgtest-impish-amd64.img
[...]
autopkgtest [16:12:15]: @@@@@@@@@@@@@@@@@@@@ summary
ovs                  PASS
ethernets            PASS
bridges              PASS
bonds                PASS
routing              PASS
vlans                PASS
wifi                 FLAKY non-zero exit status 1
tunnels              PASS
scenarios            PASS
regressions          PASS
autostart            PASS
cloud-init           PASS
qemu-system-x86_64: terminating on signal 15 from pid 375488 (/usr/bin/python3)

real	33m1,326s
user	9m28,119s
sys	1m2,647s