Unable to remove / disassociate a disk from initiator

[wwdillingham]

I have an initiator who can be seen in gwcli as such:

 o- iqn.1991-05.com.microsoft:host.example.com ........................................... [Auth: None, Disks: 3(6.0T)]
  o- lun 68 ............................................... [rbd/disk0_2021-12-15(2.0T), Owner: p3plcephigw002.example.com]
  o- lun 69 ............................................... [rbd/disk1_2021_-12-15(2.0T), Owner: p3plcephigw003.example.com]
  o- lun 70 ............................................... [rbd/disk2_2021-12-15(2.0T), Owner: p3plcephigw004.example.com]

From here I attempt to remove the disk using disk remove rbd/disk2_2021-12-15

If I monitor the changes in configuration object rbd/gateway.conf I see that that "epoch" line and the "updated" timestamps at the bottom of the json gets updated/increments when the command is issued but the disk is not removed from the json file and my gwcli command hangs forever.

Further If i look in the rbd-target-api log I see for the disk that fails to delete:
2022-01-05 10:58:45,372 INFO [_internal.py:87:_log()] - IP_REDACTED - - [05/Jan/2022 10:58:45] "DELETE /api/disk/rbd/disk2_2021-12-15 HTTP/1.1" 400 -

If I compare this to a cluster where disk removals are working I see:

2022-01-05 11:38:46,846 INFO [_internal.py:87:_log()] - IP_REDACTED - - [05/Jan/2022 11:38:46] "DELETE /api/_disk/rbd/disk9_2021-11-20 HTTP/1.1" 200 -

noting _disk vs disk in the path between working/non-working example.

I am concerned to restart rbd-target-api to see if it fixes issue as I dont want to exacerbate and Im not sure how to reproduce issue.

running ceph-iscsi-3.4-1.el7.noarch

[wwdillingham]

Other, non-lun-disassociate changes are functional on this cluster, for example additional luns can be created and associated with initiators.