rook-ceph-osd-prepare device filter

[tpaoletto]

Is this a bug report or feature request?

• Bug Report

Deviation from expected behavior:
OSD preparation failed on udev triggered device.

Expected behavior:
OSD preparation success.

How to reproduce it (minimal and precise):

If the "deviceFilter" is specified in the cluster.yaml file and it is pointing to udev-triggered devices
the rook-ceph-osd-prepare will fail.

storage: # cluster level storage configuration and selection

useAllNodes: true

useAllDevices: false

deviceFilter: "^rook."

location:

In the rook-ceph-osd-prepare job the list of "available" devices is built by issuing:
lsblk --all --noheadings --list --output KNAME

In this list all udev-triggered devices are not present.

If you want to have a persistent-binded name on your OSDs names across cluster nodes
(eg. /dev/rook1 that points to /dev/sdh on nodeA, /dev/sdy on nodeB and /dev/sdg on nodeC)
the preparation task will not work

Environment:

• OS (e.g. from /etc/os-release): CentOS Linux release 7.5.1804
• Kernel (e.g. uname -a): 3.10.0-862.14.4.el7.x86_64
• Cloud provider or hardware configuration: on-prem nutanix setup
• Rook version (use rook version inside of a Rook Pod): 0.8
• Kubernetes version (use kubectl version):1.8.13
• Kubernetes cluster type (e.g. Tectonic, GKE, OpenShift): K8S
• Storage backend status (e.g. for Ceph use ceph health in the [Rook Ceph toolbox] (https://rook.io/docs/Rook/master/toolbox.html)): VM virtual disk on Nutanix AHV

[tpaoletto]

You can find a similar behavior on issue #1228

[galexrt]

@travisn Added to 1.0 milestone, to ensure this has either been fixed or needs to be fixed.

[travisn]

The device filter was actually just fixed with #2703, which will be included in 1.0.