rook: devicePathFilter doesn't work with /dev/disk/by-partlabel

• Bug Report

Deviation from expected behavior: Devices are ignored. Expected behavior: Devices are used in cluster. How to reproduce it (minimal and precise): Create a simple partition on a disk. Use gdisk to give the partition a label. Use: “devicePathFilter: /dev/disk/by-partlabel/<the-label>”

File(s) to submit:

Fresh 1.2.7 deployment.

Log from the discovery (cut useless info):

2020-05-18 13:19:40.022717 I | exec: Running command: udevadm info --query=property /dev/sdb
2020-05-18 13:19:40.027761 I | exec: Running command: lsblk --noheadings --list -o NAME /dev/sdb
2020-05-18 13:19:40.033102 I | exec: Running command: udevadm info --query=property /dev/sdb1
2020-05-18 13:19:40.037794 I | inventory: skipping device "sdb" because it has child, considering the child instead.
2020-05-18 13:19:40.037829 I | exec: Running command: lsblk /dev/sdb1 --bytes --nodeps --pairs --output SIZE,ROTA,RO,TYPE,PKNAME
2020-05-18 13:19:40.041742 I | exec: Running command: udevadm info --query=property /dev/sdb1

udevadm info --query=property /dev/sdb1:

DEVPATH=/devices/pci0000:00/0000:00:1d.0/usb2/2-1/2-1.1/2-1.1:1.0/host6/target6:0:0/6:0:0:0/block/sdb/sdb1
DEVNAME=/dev/sdb1
DEVTYPE=partition
PARTN=1
PARTNAME=CEPH_DATA
MAJOR=8
MINOR=17
SUBSYSTEM=block
USEC_INITIALIZED=83937007236
ID_VENDOR=Crucial_
ID_VENDOR_ENC=Crucial_
ID_VENDOR_ID=152d
ID_MODEL=CT256MX100SSD1
ID_MODEL_ENC=CT256MX100SSD1\x20\x20
ID_MODEL_ID=1576
ID_REVISION=0204
ID_SERIAL=Crucial__CT256MX100SSD1_987654321029-0:0
ID_SERIAL_SHORT=987654321029
ID_TYPE=disk
ID_INSTANCE=0:0
ID_BUS=usb
ID_USB_INTERFACES=:080650:080662:
ID_USB_INTERFACE_NUM=00
ID_USB_DRIVER=uas
ID_PATH=pci-0000:00:1d.0-usb-0:1.1:1.0-scsi-0:0:0:0
ID_PATH_TAG=pci-0000_00_1d_0-usb-0_1_1_1_0-scsi-0_0_0_0
ID_PART_TABLE_UUID=0e5b541c-593f-364e-aad0-9e14a35330d2
ID_PART_TABLE_TYPE=gpt
ID_FS_UUID=909d7a94-518f-437c-96ad-0081cd2c0b92
ID_FS_UUID_ENC=909d7a94-518f-437c-96ad-0081cd2c0b92
ID_FS_VERSION=1.0
ID_FS_TYPE=ext4
ID_FS_USAGE=filesystem
ID_PART_ENTRY_SCHEME=gpt
ID_PART_ENTRY_NAME=CEPH_DATA
ID_PART_ENTRY_UUID=69b2a0e3-78c8-fc4d-968e-27cf536b05b1
ID_PART_ENTRY_TYPE=0fc63daf-8483-4772-8e79-3d69d8477de4
ID_PART_ENTRY_NUMBER=1
ID_PART_ENTRY_OFFSET=2048
ID_PART_ENTRY_SIZE=500116111
ID_PART_ENTRY_DISK=8:16
DEVLINKS=/dev/disk/by-uuid/909d7a94-518f-437c-96ad-0081cd2c0b92 /dev/disk/by-id/usb-Crucial__CT256MX100SSD1_987654321029-0:0-part1 /dev/disk/by-partlabel/CEPH_DATA /dev/disk/by-partuuid/69b2a0e3-78c8-fc4d-968e-27cf536b05b1 /dev/disk/by-path/pci-0000:00:1d.0-usb-0:1.1:1.0-scsi-0:0:0:0-part1
TAGS=:systemd:

So both PARTNAME and DEVLINKS show the correct label (CEPH_DATA), but i guess because the ‘raw’ devpath is still sdb, it’s ignored.

This causes issues, because I want to generalize the config for all machines and use autodiscover. I just want to be able to use different devices in different machines with different paths with a generic configuration. All my machines have different disks attached, but all the Ceph partitions are labeled properly. In the future, I’m probably going to use CEPH_DATA[0-90-9] as a filter, so I can easily and dynamically add OSD’s.

Environment:

• OS (e.g. from /etc/os-release):
• Kernel (e.g. uname -a): Debian 10, fully updated
• Cloud provider or hardware configuration: Intel NUC’s
• Rook version (use rook version inside of a Rook Pod): 1.2.7
• Storage backend version (e.g. for ceph do ceph -v): latest for 1.2.7
• Kubernetes version (use kubectl version):1.17.3
• Kubernetes cluster type (e.g. Tectonic, GKE, OpenShift): kubeadm
• Storage backend status (e.g. for Ceph use ceph health in the Rook Ceph toolbox): healthy