VMAX should not tie volume extra specs exclusively to driver-wide config
Affects | Status | Importance | Assigned to | Milestone | |
---|---|---|---|---|---|
Cinder |
Fix Released
|
Critical
|
Xing Yang |
Bug Description
The change recently merged here, https:/
This *breaks* PowerVC concurrency support because of the PowerVC requirement to support volume-based extra_specs from "volume types". For example, storagetype:
But with https:/
Ideally, the VMAX driver would add a facility like other drivers do, to obtain a set of extra_specs from the volume type, accessed via volume[
UPDATE: I looked at the EMC code a little closer and it appears to be broken in general, not just because we override _initial_setup(). The code does get the volume type extra specs:
extraSpecs = self.utils.
The issue is that these extraSpecs are overwritten in-memory by what is contained in the config xml file. With FAST policy for example, if it is not in the config file, the in-memory extra spec still gets overwritten with None. So the volume type extra spec does not get honored. And in the case of storagetype:
description: | updated |
tags: | added: drivers emc vmax |
Changed in cinder: | |
assignee: | nobody → Xing Yang (xing-yang) |
Changed in cinder: | |
milestone: | none → kilo-3 |
status: | New → In Progress |
importance: | Undecided → Critical |
Changed in cinder: | |
status: | In Progress → Fix Committed |
Changed in cinder: | |
status: | Fix Committed → Fix Released |
Changed in cinder: | |
milestone: | kilo-3 → 2015.1.0 |
Fix proposed to branch: master /review. openstack. org/162658
Review: https:/