Open danielwieberneit opened 5 years ago
bgilbert
commented
5 years ago I'm not sure whether Container Linux would successfully boot, even if the install completed normally. Could you try modifying the check to accept mpath and see whether you get a bootable system?
danielwieberneit
commented
5 years ago we copied coreos-install to a tmp Folder, added mpath. System booted after Execution.
lucab
commented
5 years ago @danielwieberneit can you please also double-check that it boots fine on both first-boot and on subsequent reboots?
Additionally, is the /etc/modules-load.d/multipath.conf above configured for the installer or for the final OS?
danielwieberneit
commented
5 years ago I tried to reproduce. I run in BLKRRPART: Device or resource busy. The partitions are already in use by the kernel and I get problems restarting the changed coreos-install Skript
danielwieberneit
commented
5 years ago There seam to be a lot of problems to make CoreOS boot from a multi-path device. I found 2 further problems in coreos-install Skript. It is probably not all.
line 508 blockdev --rereadpt "${DEVICE}" && unset try && break ||
blockdev --rereadpt "/dev/dm-0" blockdev: ioctl error on BLKRRPART: Invalid argument -> dm devices seam to be incompatible to ioctl
Line 636:
local OEM_DEV=$(blkid -t "LABEL=OEM" -o device "${DEVICE}"*)
blkid returns not /dev/dm-0, but /dev/mapper/3600507680c80840e1800000000000032
After working around I can boot. But our configuration from ignition.json still will not work effective. In consequence I just boot from /dev/sda without multipathing. Thy is why multiple restarts seemed to work. I could not login...
Now I see blkid is used in line 611 and 614 too...
danielwieberneit
commented
5 years ago After some workarounds I was able to install to /dev/dm0. I have the following messages in journal when booting the installed system: kernel: device-mapper: multipath service-time: version 0.3.0 loaded kernel: device-mapper: table: 254:1: multipath: error getting device kernel: device-mapper: ioctl: error adding target to table multipathd[4207]: 3600507680c80840e1800000000000032: ignoring map
HBA and disks are discovered 10s before.
Looks like it makes no difference installing to /dev/dm-0
I had the same results when installing to /dev/sda
Disks are already in use. Even though Multipathing is loaded the partitions are spread over the path:
NAME MAJ:MIN RM SIZE RO TYPE MOUNTPOINT
sda 8:0 0 20G 0 disk
|-sda1 8:1 0 128M 0 part /boot
|-sda2 8:2 0 2M 0 part
|-sda3 8:3 0 1G 0 part
|-sda4 8:4 0 1G 0 part
|-sda6 8:6 0 128M 0 part /usr/share/oem
|-sda7 8:7 0 64M 0 part
-sda9 8:9 0 17.7G 0 part / sdb 8:16 0 20G 0 disk |-sdb1 8:17 0 128M 0 part |-sdb2 8:18 0 2M 0 part |-sdb3 8:19 0 1G 0 part |-usr 254:0 0 1016M 1 crypt /usr
|-sdb4 8:20 0 1G 0 part
|-sdb6 8:22 0 128M 0 part
|-sdb7 8:23 0 64M 0 part
-sdb9 8:25 0 17.7G 0 part sdc 8:32 0 20G 0 disk |-sdc1 8:33 0 128M 0 part |-sdc2 8:34 0 2M 0 part |-sdc3 8:35 0 1G 0 part |-sdc4 8:36 0 1G 0 part |-sdc6 8:38 0 128M 0 part |-sdc7 8:39 0 64M 0 part -sdc9 8:41 0 17.7G 0 part
sdd 8:48 0 20G 0 disk
|-sdd1 8:49 0 128M 0 part
|-sdd2 8:50 0 2M 0 part
|-sdd3 8:51 0 1G 0 part
|-sdd4 8:52 0 1G 0 part
|-sdd6 8:54 0 128M 0 part
|-sdd7 8:55 0 64M 0 part
`-sdd9 8:57 0 17.7G 0 part
Is there something like a kernel boot option for multipathing?
Issue Report
MPATH LUN ist not recognized as full Disk by coreos-install Skript. Error: Target block device /dev/dm-0 is not a full disk."
Environment
HP ProLiant DL 380 Gen9 2 x HP Store Fabric 8 Gb Single Port PCI-e FC HBA IBM SAN Volume Controller (SVC)
We load dm-multipath Modul:
Desired Feature
A FC LUN / Multipath Device (/dev/dm-0) should be choosable as install-disk mpath type should be recognized as full disk.
Other Information
Line 446 in coreos-install: if ! [[ $(lsblk -n -d -o TYPE "${DEVICE}") =~ ^(disk|loop|lvm)$ ]]; then echo "$0: Target block device (${DEVICE}) is not a full disk." >&2 exit 1 fi lsblk -n -d -o TYPE "${DEVICE}" returns mpath