What were you trying to do that didn't work?

Reboot after configuring an nvme subsystem shared via tcp

What is the impact of this issue to you?

No real impact, a trivial fix by modifying the nvmet.service unit is possible

Please provide the package NVR for which the bug is seen:

nvmetcli-0.7-3.el9.noarch

How reproducible is this bug?:

Always

Steps to reproduce

1. Setup an nvme target subsystem with tcp ports. An example layout below: 

   # nvmetcli ls
   o- / ......................................................................................................................... [...]
     o- hosts ................................................................................................................... [...]
     | o- nqn.2014-08.org.nvmexpress:uuid:d92766d9-487e-4bfe-8fdf-d47134a8050d .................................................. [...]
     o- ports ................................................................................................................... [...]
     | o- 1 ................................................ [trtype=tcp, traddr=192.168.141.197, trsvcid=4420, inline_data_size=16384]
     | | o- ana_groups .......................................................................................................... [...]
     | | | o- 1 ..................................................................................................... [state=optimized]
     | | o- referrals ........................................................................................................... [...]
     | | o- subsystems .......................................................................................................... [...]
     | |   o- nqn.2024-12.local.lab:sn.123456789 ................................................................................ [...]
     | o- 2 ................................................ [trtype=tcp, traddr=192.168.142.197, trsvcid=4420, inline_data_size=16384]
     |   o- ana_groups .......................................................................................................... [...]
     |   | o- 1 ..................................................................................................... [state=optimized]
     |   o- referrals ........................................................................................................... [...]
     |   o- subsystems .......................................................................................................... [...]
     |     o- nqn.2024-12.local.lab:sn.123456789 ................................................................................ [...]
     o- subsystems .............................................................................................................. [...]
       o- nqn.2024-12.local.lab:sn.123456789 .................................. [version=1.3, allow_any=0, serial=1a5b978928f51adbc438]
         o- allowed_hosts ....................................................................................................... [...]
         | o- nqn.2014-08.org.nvmexpress:uuid:d92766d9-487e-4bfe-8fdf-d47134a8050d .............................................. [...]
         o- namespaces .......................................................................................................... [...]
           o- 1 ................................ [path=/dev/sharedvg/LUN1, uuid=45ef94ec-d280-4ea0-9ba1-5140688ee111, grpid=1, enabled]
           o- 2 ................................ [path=/dev/sharedvg/LUN2, uuid=bcad08b6-ba5b-4a0f-b120-a3ff15beb3fe, grpid=1, enabled]
           o- 3 ................................ [path=/dev/sharedvg/LUN3, uuid=7033abf6-f894-42b8-a40c-fc6f0f7172cc, grpid=1, enabled]

1. Save the configuration in /etc/nvmet/config.json 

   saveconfig savefile=/etc/nvmet/config.json

1. Enable nvmet.service 

   systemctl enable nvmet.service

1. Reboot

Expected results

nvmet restores the configuration and starts normally

Actual results

nvmet races with network setup and fails because NICs don't have their IP addresses assigned yet