Add OUI model

[stveit]

Closes #2957

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 56.36%. Comparing base (0a90566) to head (0029ccb).
Report is 598 commits behind head on master.

Additional details and impacted files

@@            Coverage Diff             @@
##           master    #2896      +/-   ##
==========================================
+ Coverage   56.35%   56.36%   +0.01%     
==========================================
  Files         603      604       +1     
  Lines       43862    43858       -4     
  Branches       48       48              
==========================================
+ Hits        24719    24722       +3     
+ Misses      19131    19124       -7     
  Partials       12       12              

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

[github-actions]

Test results

     12 files       12 suites   11m 41s ⏱️
3 324 tests 3 324 ✔️ 0 💤 0 ❌
9 447 runs  9 447 ✔️ 0 💤 0 ❌

Results for commit 1b333b0.

♻️ This comment has been updated with latest results.

[lunkwill42]

I think you should consider the option to store the OUI prefix as a regular PostgreSQL macaddr data type, which has several benefits when it comes to indexing, sorting, joining in pure SQL etc.

Just ensure that the last three bytes are always zero (this could also be accomplished by a constraint on the table. The built-in trunc function can also truncate mac address values).

See https://www.postgresql.org/docs/13/datatype-net-types.html and https://www.postgresql.org/docs/13/functions-net.html

[stveit]

Made some changes according to your suggestions

[lunkwill42]

Looks ok to me. We would usually put the model in nav.models.manage, since the database object is placed in the manage namespace - but we could really stand to split up this giant module anyway, so let's roll with it.

My only nitpick is the naming of the SQL change file itself. It should follow the established pattern of corresponding with the major.minor of the NAV release we expect it to land in, so it should either be prefixed sc.05.10. or sc.05.11....

[CLAassistant]

All committers have signed the CLA.

[stveit]

Why do you do the validation of the last three bytes being zero on the SQL level and not earlier how you validated the length before?

We got it super free this way with absolute minimal code and maximum speed. Better to do it with postgres code thats made to do this stuff super fast than manually with slow python code.

The max length might not be necessary tbh since postgres MACADDR type should only allow a valid mac address anyways

[johannaengland]

Why do you do the validation of the last three bytes being zero on the SQL level and not earlier how you validated the length before?

We got it super free this way with absolute minimal code and maximum speed. Better to do it with postgres code thats made to do this stuff super fast than manually with slow python code.

The max length might not be necessary tbh since postgres MACADDR type should only allow a valid mac address anyways

I noticed the comment by @lunkwill42 too late, which is why I hid this comment 😅

[lunkwill42]

Short and sweet!