refactor(pytest): make test more robust and easier to debug

[tiagolobocastro]

refactor(nix): make shell nix depend on other shells

Signed-off-by: Tiago Castro <[email protected]>

---

ci(bdd): use larger runner for testing

Signed-off-by: Tiago Castro <[email protected]>

---

test(bdd): make test more robust

Ensure install is properly completed
Ensure upgrade is properly completed
Compare against v-next and not v-next since there's no current way
of knowing which version is the current helm chart.
This may get addressed if we start building the images and the
chart as v-next from the outset.

Signed-off-by: Tiago Castro <[email protected]>

---

test: ensure hugepages allocated

When deploying a kind cluster, ensure we allocate the requested hugepages.

Signed-off-by: Tiago Castro <[email protected]>

---

refactor(pytest): make debugging easier

Change a lot of return Nones for exceptions which aids debugging by
failing at the point of failure
Reduce verbose logging when install/upgrades takes some time due to
cpu or network constraints
Remove extra logging
Allow using kubectl plugin from path

I think the tests are currently broken for release branches as its
asserting for develop.
Also we're modifying the chart itself, which is not ideal.
todo: improve this

Signed-off-by: Tiago Castro <[email protected]>

[tiagolobocastro]

bors try

[bors-openebs-mayastor]

try

Build failed:

• bors-ci