yahoo-eng-team team mailing list archive

[Acoss69 <1885252@xxxxxxxxxxxxxxxxxx>]

Public bug reported:

OpenStack version : Stein
Deployemnt tool : kolla-ansible
Glance version : 2.16.0

We have implemented an external CEPH backend and we try to use it as a glance store.
The setup is ok and we could upload small images to this target.

But we encountered a problem with bigger images. Every time the process is kill bye the "oom killer".
After analasys, the image is uploaded to the RBD store by using only memory as cache.

Maybe we have messed up our configuration but it's seems completly off to only use memory.
Could it be related to the fact that v2 api dosen't used the total size of the image and upload until it reached the end and this message "since image size is zero we will be doing resize-before-write for each chunk which will be considerably slower than normal" ?

This behaviour has been encountered with the openstack cli and the
glance cli.

Thanks for your help and advices.

P.S. : I've attached an anonimized version of our glance-api.conf

** Affects: glance
     Importance: Undecided
         Status: New

** Attachment added: "glance-api.conf"
   https://bugs.launchpad.net/bugs/1885252/+attachment/5387337/+files/glance-api.conf

-- 
You received this bug notification because you are a member of Yahoo!
Engineering Team, which is subscribed to Glance.
https://bugs.launchpad.net/bugs/1885252

Title:
  "In memory" upload with RBD store

Status in Glance:
  New

Bug description:
  OpenStack version : Stein
  Deployemnt tool : kolla-ansible
  Glance version : 2.16.0

  We have implemented an external CEPH backend and we try to use it as a glance store.
  The setup is ok and we could upload small images to this target.

  But we encountered a problem with bigger images. Every time the process is kill bye the "oom killer".
  After analasys, the image is uploaded to the RBD store by using only memory as cache.

  Maybe we have messed up our configuration but it's seems completly off to only use memory.
  Could it be related to the fact that v2 api dosen't used the total size of the image and upload until it reached the end and this message "since image size is zero we will be doing resize-before-write for each chunk which will be considerably slower than normal" ?

  This behaviour has been encountered with the openstack cli and the
  glance cli.

  Thanks for your help and advices.

  P.S. : I've attached an anonimized version of our glance-api.conf

To manage notifications about this bug go to:
https://bugs.launchpad.net/glance/+bug/1885252/+subscriptions