RFE - Provide option to test against a real target array

[martin-gpy]

Instead of the current soft target, it would help if there was an option available in blktests to test against a NVMe-oF block device provisioned from a real target array (like say an ONTAP array).

[kawasaki]

@martin-gpy I can understand such usage of blktests. Both soft test targets and real targets have benefits. Soft targets allow to run tests without real hardware, so more people can run tests easily. Real hardware allow to catch more realistic and important failures, so the test runs will be more benefitial.

I guess pull request #86 has same motivation for rdma hardware. Unfortunately, its activity stalls. From the comment in the pull request, I guess more script work will be required. I wish someone's contribution to it.

[igaw]

FWIW, I'll try to look into this.

[igaw]

FWIW, the refactoring of the nvme tests so that we have helper functions for setting up and tearing down the target was the first step to get this working.

[hreinecke]

Next step would be to abstract away printing the subsystem NQN in the *.out files; that will break when moving to a pre-defined subsystem setup.

[hreinecke]

I've now created #127 which should address the above issue.