[Starlingx-discuss] Questions related to Startlingx & Openstack

bnovickovs at weecodelab.com bnovickovs at weecodelab.com
Wed Jun 17 17:14:12 UTC 2020


Hello,

First of all, I would like to say thank you very much for your effort in 
this amazing project, and congratulations for being a top-level project 
at Openstack Foundation.

I have been testing StarlingX for 5 days now and wanted to ask / specify 
some questions of mine related to Starlingx in general and Openstack 
deployment on it in particular.

StarlingX:
1. I know that the 4th release is upcoming in July. Also, from what I've 
heard in StarlingX irc channel, you can't upgrade from version 3 to 
version 4. Thus, I am wondering if I deploy the 4th version, would I be 
able to upgrade it in the future with up-to-date releases?
2. I really enjoy StarlingX management UI. However, I can see that it's 
not fully-functional yet. I am wondering, if Host Inventory will be 
upgraded so the host can be added via UI? Is Software Management can be 
used for patching anything within infrastructure and StarlingX related?
3. Since Kubernetes is using docker, and docker is highly dependent on 
Linux kernel; and as far as I know, it is the best practice to use newer 
versions of kernel with docker, do you have any plans of upgrading 
kernel version til 5.x version?

Openstack:
1. I've been working a lot with Mirantis MOSK (Openstack on Kubernetes) 
https://docs.mirantis.com/mosk/beta/index.html & 
https://github.com/Mirantis/release-openstack-k8s ; which uses 
operator's mechanism for both openstack and ceph. Their package is very 
nice, still in beta though, but it is consuming a lot more resources 
compared to Starlingx package. However, mosk provides 
designate/octavia/cells2 services by default. Thus, I am wondering if 
the StarlingX Openstack package can offer additional services deployment 
if required, like octavia or magnum?
2. In documentation it is specified that having an additional disk for 
Openstack nova-local ephemeral images is a good practice. However, I 
could not figure out how to use a separate disk for that since it's not 
specified anywhere in the docs. Default procedure is to use a root disk 
(screenshot: https://prnt.sc/t1fqwr). Therefore, how can I use a 
separate disk for nova-local images and avoid having those errors in 
openstack when I am creating an instance without volume (screenshot: 
https://prnt.sc/t1fs41).
3. I had a similar issue at the start 
https://bugs.launchpad.net/starlingx/+bug/1867731 ; However, I am a bit 
confused why you would use services externally, using external FQDN, 
while you use horizon. When setup external dns for openstack services 
and open horizon UI, it works very slowly. If you create volume in 
horizon and create volume using internal CLI, its totally different 
speeds but it should not be like that. I am wondering, if Horizon can be 
used while all the services will be connected and used internally? That 
is what I can see using MOSK, and Horizon UI of theirs is working very 
fast.
4. I think you are aware of it, but you can't upload images (glance 
sends python errors) using horizon ui therefore, only internal CLI can 
be used for that.
5. I would really appreciate if you can help me with troubleshooting 
this: https://bugs.launchpad.net/starlingx/+bug/1883911

Thank you very much and regards



More information about the Starlingx-discuss mailing list