Open Stack

On Thu, 2021-03-11 at 18:54 +0530, Midhunlal Nb wrote:
> Hi all,
> I successfully installed the openstack rocky version and everything is
> working properly.
> 
> ---->Uploaded all images to openstack and I can see all images listed
> through openstack CLI.
> +--------------------------------------+---------------------+--------+
> > ID                                                                    |
> Name                          | Status |
> +--------------------------------------+---------------------+--------+
> > 1912a80a-1139-4e00-b0d9-e98dc801e54b    | CentOS-7-x86_64         active |
> > 595b928f-8a32-4107-b6e8-fea294d3f7f1       | CentOS-8-x86_64
> active|
> > f531869a-e0ac-4690-8f79-4ff64d284576       | Ubuntu-18.04-x86_64
>  active|
> > 7860193b-81bd-4a22-846c-ae3a89117f9d     | Ubuntu-20.04-x86_64     active
> > 1c3a9878-6db3-441e-9a85-73d3bc4a87a7    | cirros
>      active
> > d321f3a3-08a3-47bf-8f53-6aa46a6a4b52       | windows10
>   active
> 
> --->same images I can see through horizon dashboard,but i am not able to
> launch windows VM from horizon dashboard .Balance all images launching from
> dashboard.
> 
> Getting below error;
> Error: Failed to perform requested operation on instance "windows", the
> instance has an error status: Please try again later [Error: Build of
> instance 15a3047d-dfca-4615-9d34-c46764b703ff aborted: Volume
> 5ca83635-a2fe-458c-9053-cb7a9804156b did not finish being created even
> after we waited 190 seconds or 61 attempts. And its status is downloading.].
> 
> --->I am able to launch windows VM from openstack CLI
> 
> What is the issue ?why I am not able to launch from dashboard?please help
your cinder sotrage is likel not that fast and the windows image is much larger then the other images
so its takeing longer then our default timeout to create the volume.
im assuming your using somthing like the cinder LVM backend.

try adding 

[DEFAULT]
block_device_allocate_retries_interval=10

to your nova.conf or nova-cpu.conf if this is a devstack install
if you look in the cinder driver log you will likely see its still creating the volume form the image and if
you use htop/ps on the host with the cinder volume driver you will see the qemu-img command running.

other ways to work around this is to confgiure glance to use cider for image storage that will allow the cinder backend
to create a volumn snapshot for the new volume instaed of making a full copy. over all that is much more effeicnt and works
similar to when you use ceph for both glance and cinder.

> me.
> 
> Thanks & Regards
> Midhunlal N B
> +918921245637