ppc64el node maas install failing talking to mpath0

Bug #1579586 reported by Brad Marshall on 2016-05-09
12
This bug affects 1 person
Affects Status Importance Assigned to Milestone
multipath-tools (Ubuntu)
High
Mathieu Trudel-Lapierre

Bug Description

I'm trying to install a PPC64EL node with Xenial via MAAS 1.9.2 using Juju 1.25.5 and it seems to be failing due to a possible multipath issue. The node is setup in maas with a default single partition. The boot logs show:

  Begin: Running /scripts/init-premount ... done.
  Begin: Mounting root file system ... Begin: Running /scripts/local-top ... lvmetad is not active yet, using direct activation during sysinit
    Volume group "mpath0" not found
    Cannot process volume group mpath0
  done.

and then multiple repeats of:

  Begin: Waiting for root file system ... Begin: Running /scripts/local-block ... lvmetad is not active yet, using direct activation during sysinit
    Volume group "mpath0" not found
    Cannot process volume group mpath0
  done.

To finally fail with:

  Gave up waiting for root device. Common problems:
   - Boot args (cat /proc/cmdline)
     - Check rootdelay= (did the system wait long enough?)
     - Check root= (did the system wait for the right device?)
   - Missing modules (cat /proc/modules; ls /dev)
  ALERT! /dev/mapper/mpath0-part2 does not exist. Dropping to a shell!
  Rebooting automatically due to panic= boot argument

$ dpkg-query -W maas
maas 1.9.2+bzr4568-0ubuntu1~trusty1
$ dpkg-query -W juju-core
juju-core 1.25.5-0ubuntu1~14.04.2~juju1

Kernel version booting on the PPC64EL node is Linux version 4.4.0-21-generic.

Please let me know what further information you require to debug this.

Thanks,
Brad

Steve Langasek (vorlon) on 2016-05-17
Changed in multipath-tools (Ubuntu):
assignee: nobody → Mathieu Trudel-Lapierre (cyphermox)

Please attach the full logs for boot (you may need to boot with "verbose" tacked on to the kernel command-line); including full logs for the deployment via MAAS might help too.

multipath setup should be happening before LVM; and requires the multipath-boot package to be installed. It's not unlikely for LVM to fail to find "mpath0" in the case where multipath did not bring up the devices since all these names will come straight from device-mapper.

Furthermore, this looks slightly off. On Xenial, the first multipath device node should be (by default, unless this was changed by configuration) named "mpatha", not "mpath0".

Changed in multipath-tools (Ubuntu):
status: New → Incomplete
importance: Undecided → High

Any news on this? Could you include all the logs for the boot attempt?

Changed in multipath-tools (Ubuntu):
status: Incomplete → New
status: New → Incomplete
milestone: none → ubuntu-17.01
Changed in multipath-tools (Ubuntu):
milestone: ubuntu-17.01 → ubuntu-17.02

Still no update on this bug; I'm removing the milestone for now.

Changed in multipath-tools (Ubuntu):
milestone: ubuntu-17.02 → none
To post a comment you must log in.
This report contains Public information  Edit
Everyone can see this information.

Other bug subscribers