instance panel takes much more time when there are too many volumes

[mashuting]

Project -> instance table takes lots of time when there are too many volume resources, I found that horizon have a new change in stein version: https://review.opendev.org/#/c/648325/

Now we need one more api request for getting cinder list, this is the root cause.

Suggestion: We can request cinder api when we need this data, otherwise we won’t request it.

[Akihiro Motoki (amotoki)]

As you may know, the response of volume_list() is used to show image name for boot-from-volume instances (i.e. servers).

Your suggestion works only when all instances are not booted from volumes. If one instance is booted from a volume, volume_list() will be called after server_list() and it means the instance panel takes more time than now.

The possible solutions I see are:

(1) to keep the current implementation as-is
(2) to add an option not to show image names for boot-from-volume instances (this is pre-stein behavior)
(3) to follow mashuting's suggestion (it only works when *all* instances do not use boot-from-volume)

Thought?

[Launchpad Janitor (janitor)]

[Expired for OpenStack Dashboard (Horizon) because there has been no activity for 60 days.]