Disable SPDK discovery controller on GW initialization

[leonidc]

when Gateway is created it by default creates discovery subsystem
server.py when spdk is started, calls rpc nvmf_delete_subsystem.

Tests:

The rpc was issued before applying the new GW configuration :
[root@rhel8-leo-dev ceph-nvmeof]# nvmeof-cli get_subsystems
INFO:main:Get subsystems:
[] <= the discovery subsystem does not appear anymore

]# nvme discover -t tcp -a 192.168.13.3 -s 4420
Failed to write to /dev/nvme-fabrics: Input/output error

The PR has no impact on the current Gateway behavior. Checked also IOs from the nvme-tcp initiator.

[gbregman]

LGTM

[baum]

@leonidc 🖖 awesome PR! Please sign off the commit, see here, to color DCO check green 🟩.

Also multi_gateway test fails consistently with this change, any idea why?

[gbregman]

@leonidc the multi gateway test runs "get_subsystems" and checks that it got 2 subsystem. As there is now only one subsystem you need to adjust the test.

[leonidc]

if multi gateway test checks that 2 subsystems are configured and we want to add configuration parameter in config file that in some cases server.py would not delete discovery subsystems - request of Aviv , it looks so many dependencies for such a small change.

[caroav]

@leonidc you can check the config flag in the test, and fix accordingly.

[leonidc]

@leonidc you can check the config flag in the test, and fix accordingly.

where can I find this flag ? it does not appear in conf file

[baum]

@leonidc 🖖 awesome 👍

could you please sign your existing commit:

git commit --amend --signoff --no-edit

And re-push this change, to make DCO check happy, see here

Generally, to sign off new commit, use --signoff