[hostcfgd] [202012] Enhance hostcfgd to check feature state and run less system calls

[shlomibitton]

Signed-off-by: Shlomi Bitton shlomibi@nvidia.com

Why I did it

Currently hostcfgd is implemented in a way each feature which is enabled/disabled triggering execution of systemctl enable/unmask commands which eventually trigger 'systemctl daemon-reload' command.
Each call like this cost 0.6s and overall add a overhead of ~12 seconds of CPU time.
This change will verify the desired state of a feature and the current state of this feature on systemd and trigger a system call only when must.

How I did it

Check each feature status on systemd before executing a system call to enable and reload the systemctl daemon.

How to verify it

Build an image with this change and observe less system calls are executed.

Which release branch to backport (provide reason below if selected)

• 201811
• 201911
• 202006
• 202012
• 202106

Description for the changelog

A picture of a cute animal (not mandatory but encouraged)

[shlomibitton]

@yxieca can you please check the warm_boot test failure? It does not seems to be related to this change.
I see in the test failure: "WARNING: No route found for IPv6 destination"

[vaibhavhd]

@yxieca can you please check the warm_boot test failure? It does not seems to be related to this change.
I see in the test failure: "WARNING: No route found for IPv6 destination"

The real failure here is the exceeded downtime:
downtime"FAILED:dut:Longest downtime period must be less then 200 seconds. It was 0:03:24.487656"

The "WARNING: No route found for IPv6 destination" is benign and appears on every run of this test.

I have triggered a re-run and this should likely pass.