Unshelve with az update the requestSpec.availability_zone this can be inconsistent if schedule fails
Affects | Status | Importance | Assigned to | Milestone | |
---|---|---|---|---|---|
OpenStack Compute (nova) |
In Progress
|
Medium
|
Rajesh Tailor |
Bug Description
Description
===========
Unshelving a shelved offloaded server will go through the
scheduler to pick a new host, so we update the
RequestSpec.
fails the RequestSpec will remain updated, which is not great.
https:/
Steps to reproduce
==================
NA
Expected result
===============
If schedule fails, restore the original az.
Actual result
=============
RequestSpec inconsistent if schedule fails.
Environment
===========
This was introduced in Train (Aug 28, 2019) with the following note:
If we want to change that we need to defer updating the
RequestSpec until conductor which probably means RPC changes to
pass the new_az variable to conductor. This is likely low
priority since the RequestSpec.
offloaded server does not mean much anyway and clearly the user
is trying to put the server in the target AZ.
Logs & Configs
==============
NA
Changed in nova: | |
status: | New → Triaged |
importance: | Undecided → Medium |
tags: | added: scheduler shelve |
Changed in nova: | |
assignee: | nobody → Rajesh Tailor (ratailor) |
Fix proposed to branch: master /review. opendev. org/c/openstack /nova/+ /911108
Review: https:/