Effi Ofer
Effi Ofer
Re-iterating Ezra's comments - this is a good change, but lets keep the ability to run each independently. If you prefer, you can change the default for ginkgo's -skip-setup to...
The e2e tests used to support building OTP separately and making the path available via an env variable. But rather then reinvent that mechanism or something like it, it would...
The bottom line is that we need to wait both for the SYNCED and the READY to be set to true?
> @effi-ofer After looking at the code, I'm not sure ginkgo is a good choice for the performance test. I'm quite sure that for such very simple base line tests,...
> > The purpose is to identify early major regressions resulting in 10x elapsed time. > > That may be an eventual goal, or maybe an expectation for some future...
> (A) The explanation in the PR and in the README does not make the exact sequencing clear. What is not clearly said is that first the BindingPolicy is created,...
@pdettori can I take this over?
I just tried to repro this on my system. I didn't had the metrics server setup so I couldn't track the memory in sufficient graduality. But I did notice that...
Sorry I wasn't clear. The manifestwork object is deleted when empty. The binding control object remains. I see the binding object is deleted when the bindingpolicy is deleted. I expect...
Running the `ginkgo -v -focus "sets singleton status when a singleton bindingpolicy/deployment is created" -- -skip-setup` test in a loop I can see the memory used by wec1 (aka cluster1...