deployment HANGS on removing pre-existing devices

Bug #1882964 reported by David Andruczyk
6
This bug affects 1 person
Affects Status Importance Assigned to Milestone
MAAS
Fix Released
Undecided
Unassigned
2.6
Triaged
Medium
Unassigned

Bug Description

Maas 2.6.2-7841-ga10625be3-0ubuntu1~18.04.1

When deploying a machine, maas is blocking on cleaning up old stuff off of drives (why?).
i.e. I redeploy a machine that has a large RAID10 of SSD's that is partitioned into two two PV's for two different volume groups, on re-deploy, maas starts up mdadm, continues the previous resync (if it hadn't completed) and then proceeds to block completely on lvremoving the old volumes.

Logging into the instance into the deploying ephemeral shows the RAID10 is rebuilding at 4KB/sec

# uname -a
Linux use1-oscomp-104 5.4.0-37-generic #41-Ubuntu SMP Wed Jun 3 18:57:02 UTC 2020 x86_64 x86_64 x86_64 GNU/Linux

# cat /proc/mdstat
Personalities : [linear] [multipath] [raid0] [raid1] [raid6] [raid5] [raid4] [raid10]
md125 : active raid10 sdh2[7] sdf2[5] sdi2[8] sdg2[6] sdj2[9] sdd2[3] sdl2[11] sda2[0] sdb2[1] sdk2[10] sde2[4] sdc2[2]
      22503598080 blocks super 1.2 512K chunks 2 near-copies [12/12] [UUUUUUUUUUUU]
      [>....................] resync = 0.4% (90218368/22503598080) finish=79106046.0min speed=4K/sec
      bitmap: 168/168 pages [672KB], 65536KB chunk

md126 : active (auto-read-only) raid1 sdn2[1] sdm1[0]
      228436992 blocks super 1.2 [2/2] [UU]
      bitmap: 0/2 pages [0KB], 65536KB chunk

md127 : active (auto-read-only) raid1 sdn3[1] sdm2[0]
      5848064 blocks super 1.2 [2/2] [UU]

unused devices: <none>

lvremove is hung
root 6720 4937 0 14:33 ? 00:00:04 lvremove --force --force <vg_name>/lv0

This makes maas completely unusable for re-imaging a machine with a pre-existing disk configuration, I know the word "should" is bad in a bug report, but the ephemeral for imaging shouldn't start mdadm, and should just wipe out the drives it's configured to touch, and build the machine, not spin up unnecessary stuff and block indefinitely.

The only way I can fix this, is to boot it into rescue, manually wipe the drives so that MDADM won't start on booting the installer ephemeral and then re-deploy it.

Revision history for this message
Alberto Donato (ack) wrote :

Does this still happen with more recent versions of MAAS?

Changed in maas:
status: New → Incomplete
Revision history for this message
David Andruczyk (dandruczyk) wrote :

This does not appear to happen on Maas 2.9.x

Changed in maas:
status: Incomplete → Fix Released
To post a comment you must log in.
This report contains Public information  
Everyone can see this information.

Other bug subscribers

Remote bug watches

Bug watches keep track of this bug in other bug trackers.