Ask Your Question
0

How to add space for glance service

asked 2015-09-29 15:11:12 -0500

sacha-m gravatar image

I have 100GB for Glance service. I need to reinstall everything, so I want to create images of my existing instances and Volumes. For volumes I use Cinder. From Cinder I can upload the volumes as images to Glance. The problem every volume is 1TB at least. So, what I wanted to do is to "give" more space to Cinder. How?

  • Move every image file from /var/lib/glance/images/ to /tmp
  • Install/format/mount a new 4TB HDD on /var/lib/glance/images/ (mount /dev/sdc1 /var/lib/glance/images/)
  • Change owner of /var/lib/glance/images/ to glance.glance
  • Move back all image files from /tmp

After that, every single creation, deletion or anything related with images failed. If I delete an image, this is what I get from command line:

# glance image-delete a8818cdd-9737-4294-9e48-06f995c5b873
HTTPInternalServerError (HTTP 500): Unable to delete image a8818cdd-9737-4294-9e48-06f995c5b873

And in /var/log/glance/api.log this is what I see:

 File "/usr/lib/python2.7/site-packages/glance_store/_drivers/filesystem.py", line 501, in delete
    raise exceptions.Forbidden(_("You cannot delete file %s") % fn)

I don't know where can be the problem. Any clue?

edit retag flag offensive close merge delete

1 answer

Sort by ยป oldest newest most voted
0

answered 2015-09-29 21:22:12 -0500

Moe gravatar image

Hi,

I suggest to change the glance mechanism from "Saving to files" to "Clustered RBD Mode" in this way you will need to install CEPH and add some HDDs all around your Servers and cluster every spare disks together this need refactoring the architecture but brings lots of flexibility with itself for example in your case instead of moving data from a disk to another or to mount a bigger hard disk you will install ceph client and install a separated storage servers in your server farm which are responsible for delivering the storage service.

you can have a look at the link below: ceph & glance configuration

Have in mind you should have active + clean state in ceph cluster.

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: 2015-09-29 15:11:12 -0500

Seen: 543 times

Last updated: Sep 29 '15