After a week of downtime from a thin pool that had 100% full metadata I decided 
to rebuild my single node.  My post was formerly titled: Ovirt 4.2.7 won't 
start and drops to emergency console.  I have backups of nearly all my VMs and 
I am working on importing them.  I went from version to 4.3.6.  I am 
comfortable with LVM but I am new to administering thin pool LVM.  I am asking 
myself what can I do to avoid this headache in the future?  What practical 
steps should I take?
-Regular backups
-Proper configuration

It's the last one that is the most difficult for me.  You don't know what you 
don't know.... Can anyone share tips and advice?

I learned that /etc/LVM has backups.  It should be part of the backup strategy. 
 I now know that thin pools have a special metadata volume that should be 
monitored for disk space.

At the time I built the system I used the gedploy gluster config file.  I had 
assumed the values in there were "safe".  As I found out the hard way, reality 
proved otherwise.  I was reading elsewhere about allowing the thin volume to 
grow 20% as needed but no real mention of how to do that or a good example.  As 
I understand it, the config to make that happen exists in lvm.conf.  Is that 

Have there been any changes in Ansible GlusterFS setup  since 4.2.7 to choose 
different "safe" values with newer versions?  

I hope my comments won't be taken as critical by any Dev Team members regarding 
Ansible.  My sole intent for this post is so I learn what to do better/right 
next time.  I do have most of my VMs and the few I lost are not critical.  It 
was lucky that I learned a hard lesson the easy way.

Thanks for your input!
Users mailing list --
To unsubscribe send an email to
Privacy Statement:
oVirt Code of Conduct:
List Archives:

Reply via email to