Resource provider not identifying NICs on s390 LPAR

[Jeff Lane ]

While running 20.04 cert on an s390x (z13) LPAR I noticed that none of the network tests are run. Looking at the tests executed this seems to be because NETWORK devices are not discovered by the resource jobs.

The only network job that ran was this:
ethernet/info_automated_server test skip job cannot be started: resource expression "device.category == 'NETWORK'" evaluates to false

Running udev_resource on the LPAR resulted in:
ubuntu@s1lp10-jefflane:~/.cache/plainbox/sessions/certify-20.04-2020-04-27T12.38.42.session/io-logs$ sudo /usr/lib/plainbox-provider-resource-generic/bin/udev_resource
path: /devices/css0/0.0.0266/0.0.1606
name: dasdb
bus: ccw
category: DISK
driver: dasd-eckd
product: IBM s390 Virtual Disk
product_slug: IBM_s390_Virtual_Disk

path: /devices/css0/0.0.0266/0.0.1606/block/dasdb/dasdb1
name: dasdb1
bus: block
category: DISK
product: IBM s390 Virtual Disk
product_slug: IBM_s390_Virtual_Disk

path: /devices/css0/0.0.0266/0.0.1606/block/dasdb/dasdb2
name: dasdb2
bus: block
category: DISK
product: IBM s390 Virtual Disk
product_slug: IBM_s390_Virtual_Disk

path: /devices/css0/0.0.027a/0.0.161a
name: dasda
bus: ccw
category: DISK
driver: dasd-eckd
product: IBM s390 Virtual Disk
product_slug: IBM_s390_Virtual_Disk

path: /devices/css0/0.0.027a/0.0.161a/block/dasda/dasda1
name: dasda1
bus: block
category: DISK
product: IBM s390 Virtual Disk
product_slug: IBM_s390_Virtual_Disk

It properly saw the disks, but did not identify the network device:
ubuntu@s1lp10-jefflane:~/.cache/plainbox/sessions/certify-20.04-2020-04-27T12.38.42.session/io-logs$ ip link show
1: lo: <LOOPBACK,UP,LOWER_UP> mtu 65536 qdisc noqueue state UNKNOWN mode DEFAULT group default qlen 1000
    link/loopback 00:00:00:00:00:00 brd 00:00:00:00:00:00
2: encc000: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc mq state UP mode DEFAULT group default qlen 1000
    link/ether 2e:38:26:ef:4c:42 brd ff:ff:ff:ff:ff:ff
3: encc000.2653@encc000: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc noqueue state UP mode DEFAULT group default qlen 1000
    link/ether 2e:38:26:ef:4c:42 brd ff:ff:ff:ff:ff:ff
4: lxdbr0: <NO-CARRIER,BROADCAST,MULTICAST,UP> mtu 1500 qdisc noqueue state DOWN mode DEFAULT group default qlen 1000
    link/ether 00:00:00:00:00:00 brd ff:ff:ff:ff:ff:ff

Doing a manual dump of the udevdb (sudo udevadm info --export-db) shows that encc000 is listed:
P: /devices/qeth/0.0.c000/net/encc000
L: 0
E: DEVPATH=/devices/qeth/0.0.c000/net/encc000
E: SUBSYSTEM=net
E: INTERFACE=encc000
E: IFINDEX=2
E: USEC_INITIALIZED=9632783
E: ID_MM_CANDIDATE=1
E: ID_NET_NAMING_SCHEME=v245
E: ID_NET_NAME_PATH=encc000
E: ID_PATH=ccwgroup-0.0.c000
E: ID_PATH_TAG=ccwgroup-0_0_c000
E: SYSTEMD_ALIAS=/sys/subsystem/net/devices/encc000
E: TAGS=:systemd:

as is encc000.2653:
P: /devices/virtual/net/encc000.2653
L: 0
E: DEVPATH=/devices/virtual/net/encc000.2653
E: SUBSYSTEM=net
E: DEVTYPE=vlan
E: INTERFACE=encc000.2653
E: IFINDEX=3
E: USEC_INITIALIZED=21303091
E: ID_MM_CANDIDATE=1
E: SYSTEMD_ALIAS=/sys/subsystem/net/devices/encc000.2653
E: TAGS=:systemd:

[Jeff Lane  (bladernr)]

Full udev dump from the LPAR

[Jeff Lane  (bladernr)]

Need to modify the udev parser so it properly identifies the devices... likely by keying on this:

SUBSYSTEM=net

for s390 systems.

[Maksim Beliaev (beliaev-maksim)]

Bug was migrated to GitHub: https://github.com/canonical/checkbox/issues/166.
Bug is no more monitored here.