Ask Your Question
0

Issue with LVM

asked 2017-02-16 23:09:48 -0600

CDOT gravatar image

I have installed Compute and Block Storage over Ubuntu Server 16.04.1 on same nodes. A few days ago there was a power failure in our lab, on account of which the entire OpenStack Cloud had an abrupt shutdown. When the power supply regained, some of the Nova-cum-Cinder nodes came up all right with instances and volumes. However, some nodes dropped to BusyBox shell (initramfs) during boot. The boot process could not find the /dev/mapper/nodename--vg-root partition.

I tried booting up the system with older kernel (4.4.0-31) from the Advanced option of Ubuntu Grub and it starts up fine.

Please suggest a way so that the nodes become functional with higher kernel versions as well.

edit retag flag offensive close merge delete

1 answer

Sort by ยป oldest newest most voted
1

answered 2017-02-17 15:24:19 -0600

With the small amount of data provided, it might be hard to track down. The initrd is created on the package install and can pull things from your environment such as lvm configs. If the older kernel works, chances are that something is up with the initrd image.

What I have seen happen in the past is if an os update happens and something happened to the kernel update it would have caused what you are seeing. A common thing with ubuntu is that the /boot directory fills up due to the way ubuntu handles upgrades and the initrd doesn't get created correctly.

A reboot should happen to test this out after an update, but this doesn't always happen. It sits like that for a while till an event happens(say a power outage) then doesn't come back up on the newer kernel with the broken initrd.

You might want to make sure your /boot space isn't full and run some cleanup. This may involve removing some older kernel images and re-installing the new one on those broken nodes. If that still doesn't work, you may have changed something in your configs that is causing bad configurations to be set up in you initrd on install of the kernel. This may require updating the initrd and testing which may be more involved. You can find tons of information via google to recreate the initrd. As long as you don't touch the older known working kernel and its initrd, you can always fall back to it after testing.

edit flag offensive delete link more

Your Answer

Please start posting anonymously - your entry will be published after you log in or create a new account.

Add Answer

Get to know Ask OpenStack

Resources for moderators

Question Tools

1 follower

Stats

Asked: 2017-02-16 23:09:48 -0600

Seen: 89 times

Last updated: Feb 17 '17