Fixes for mac2iface tests with esoteric ifaces

[ogayot]

Hello,

I know I said I wouldn't be doing more tests on nvme-stas in the near future, but the CI failed in Ubuntu on runners that have gre network interfaces.. so sending you my findings :)

There are two fixes to the test-suite in the PR. Well, one that I would call a proper fix and one which is more of a workaround..

Patch 1

mac2iface takes a MAC address as argument and returns the corresponding interface (if any). The mac2iface tests will however invoke mac2iface with invalid MAC addresses when esoteric network interfaces are present on the system. As an example,armhf autopkgtest runners in Ubuntu have gre interfaces configured so the test-suite fails.

ip link add type gre
ip -j addr show dev gre0 | jq '.[]["address"]'
> "0.0.0.0"

We now ensure that the test-suite calls mac2iface with only valid MAC addresses. We also skip interfaces that have 00:00:00:00:00:00 since it is theoretically invalid and can be assigned to multiple type of interfaces.

Patch 2 (more of a workaround)

Sometimes the same MAC address is assigned to more than one interface on the system (this is true for VLAN interfaces for instance). This confuses mac2iface, which returns only the first match.

This scenario is possibly more of a nvme-stas bug than a test-suite bug. I am not sure how to fix it properly (should a mac2ifaceS function be provided?). So for now I chose to skip the test for interfaces that share the same MAC address.

ip link add link eth0 name eth0.222 type vlan id 222
ip -j addr show dev eth0 | jq '.[]["address"]'
> "00:16:3e:36:b5:0e"
ip -j addr show dev eth0.222 | jq '.[]["address"]'
> "00:16:3e:36:b5:0e"

Thank you,
Olivier

[codecov-commenter]

Codecov Report

All modified and coverable lines are covered by tests ✅

Comparison is base (db9d90b) 58.08% compared to head (95346c1) 58.08%.
Report is 2 commits behind head on main.

❗ Your organization needs to install the Codecov GitHub app to enable full functionality.

Additional details and impacted files

@@           Coverage Diff           @@
##             main     #411   +/-   ##
=======================================
  Coverage   58.08%   58.08%           
=======================================
  Files          16       16           
  Lines        2627     2627           
=======================================
  Hits         1526     1526           
  Misses       1101     1101           

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

[martin-belanger]

Thanks for the fix.

I need to revisit this at some point. I could have used different Python libraries (psutil, netifaces, or get-nic) to retrieve the interface information, but wanted to limit the number of dependencies (i.e. I was encouraged by other distros to limit the number of dependencies). That's how I ended up writing my own library (iputil.py). But now, with the issues you're finding, I'm wondering if that was the right choice. Maybe I should look into a standard Python library even if that adds more dependencies.