combine proxymode = shared and inpod_mode = true

[bleggett]

We currently have ProxyMode::Dedicated | Shared and InpodEnabled: true|false and I can't think of any scenarios where those are not basically indicating the same thing.

We also sort of don't use ProxyMode at all, and I think we should - the tenancy model is more important to express than how we do the tenancy model (currently we have only one multitenant model anyway, so why make it optional?).

[istio-testing]

Skipping CI for Draft Pull Request.
If you want CI signal for your change, please convert it to an actual PR.
You can still manually trigger a test run with /test all

[bleggett]

slapping a hold on until @keithmattix @jaellio @justinpettit have a chance to weigh in.

[howardjohn]

Hmm, how does direct.rs work still?

This seems to break local development flows which I didn't realize would happen when initially discussing

[bleggett]

Hmm, how does direct.rs work still?

By setting proxy_mode: config::ProxyMode::Dedicated in the test config it uses.

This seems to break local development flows which I didn't realize would happen when initially discussing

Local development flows as in proxy_mode: config::ProxyMode::Dedicated flows?

Or local development flows as in proxy_mode: config::ProxyMode::Shared flows but without any inpod machinery?

[howardjohn]

Doesn't dedicated not allow multiple clients with unique source though? do we just not test that

[bleggett]

Doesn't dedicated not allow multiple clients with unique source though? do we just not test that

I don't believe we test that. Probably because we have too many flags that influence how tenancy works.

That's kind of what I'm trying to say with this PR, we really only have two tenancy modes as far as I can see - multi tenant and single tenant.

Multitenant is only supported, tested, exercised and fully functional with inpod.

And if we don't test it and we don't use it, I don't know why we carry it/am waiting for an argument as to why we should carry (and thus test) it, or not.

[howardjohn]

And if we don't test it and we don't use it, I don't know why we carry it/am waiting for an argument as to why we should carry (and thus test) it, or not.

Well we do test it, this PR removed the tests 🙂. This currently breaks how develop and ~exclusively run ztunnel. Mind if I take a bit to poke around and see if there are viable alternatives with this change to develop locally?

[bleggett]

Well we do test it, this PR removed the tests 🙂. This currently breaks how develop and ~exclusively run ztunnel. Mind if I take a bit to poke around and see if there are viable alternatives with this change to develop locally?

Fair, we lightly test it, sorta, in one spot with 3 tests, and don't support the mode. Yep, please - I don't develop with standalone local ztunnel so I'm not the best testbed for that.

(Am also OK with just not doing this if it's a pain - just feels like we should drop or consolidate these flags)

[istio-testing]

PR needs rebase.

Instructions for interacting with me using PR comments are available here. If you have questions or suggestions related to my behavior, please file an issue against the kubernetes-sigs/prow repository.