ospf6d: set adv_router for self-originated external routes

[jaredmauch]

Ensure that adv_router is set to the current router-id for self-originated external routes. This is needed for proper self-originated route filtering when sending routes to zebra, especially if the router-id has changed.

The change sets both adv_router and type fields in the route origin structure for both existing and newly created external routes to maintain consistency and proper route identification.

(discovered during testing and errors from cicd pipeline)

[ton31337]

Can this be tested?

[jaredmauch]

@ton31337 this was exposed in other testing as cicd failures, not sure how long it had been hiding out. Most of these minor things came as part of me touching other things and they became the new side-quest. We could instead try to instrument those tests further to understand the failure cases better

[ton31337]

this was exposed in other testing as cicd failures

We need more context on how this change fixes "other testing".

[donaldsharp]

to cut to the chase I think Donatas is asking for a topotest to show the problem is fixed. We have absolutely no visibility into whatever CI/CD system you may be using. Part of the requirement for more topotests is that it is helping us to ensure that code works as expected as well as people's uses cases continue to work.

[jaredmauch]

to cut to the chase I think Donatas is asking for a topotest to show the problem is fixed. We have absolutely no visibility into whatever CI/CD system you may be using. Part of the requirement for more topotests is that it is helping us to ensure that code works as expected as well as people's uses cases continue to work.

This is in your existing CICD pipeline. I have discovered numerous bugs while trying to make other changes and in some cases come back and linked to the example failed tests. I understand wanting to know that these are all not going to introduce other breakages, but a few examples are here:

#20081

which is exhibited in the test for
#20030
https://github.com/FRRouting/frr/pull/20030/checks?check_run_id=55778828467

The same is true for this PR:
#20026

which shows up in other test failures from other branches, so instead of giving you a mega-PR with here's everything latent in the codebase that I'm hitting in other efforts to contribute to FRR, you have a lot of small PRs fixing the CICD failures that are unrelated to code being changed. The LDP IPv4 failure is a race condition failure that would only be exposed in a subset of the full suite of tests.

I'm honestly getting a bit frustrated in trying to contribute to the project because I can't get a clean test result of my changes as narrow as they may be due to these, so I don't know if I'll hit this ospf6, ldp6, ospf6 bug in PR#20081 and if you look at 20081 that also has CICD failures that are unrelated to what I've been working on as well.

These are all failures exposed in the FRRouting CICD running your pipelines on your servers, not mine. I hope this provides the clarity you are seeking. Happy to have a live discussion to answer questions as well.