This is mostly a brain-dump spec on the topic of compute hosts and
how fragile they are in terms of hostname handling. There has long
been a requirement that computes can NEVER change hostnames, and we
have few tools to even detect the problem before we corrupt the
database if it happens.
Here I have documented some of the things we could do to make that
more robust, should we choose to do so. This is based on a recent
near-catastrophe and thus reflects things that would have avoided
pain in a real scenario.
Per discussion at the PTG, I am adding this as a backlog spec, to be
an overarching guide for multiple smaller specs to provide more
detailed progress towards the goals described here.
Change-Id: I72fa3f605cfcf7c3dd0ff4c791be7df8f19f058b
Scrubbing the backlog directory.
This spec has been here since 2015.
It is probably related to [1], but I haven't dug into that yet. The
authors of either spec should feel free to modify this commit to
clarify.
[1] https://review.openstack.org/#/c/635408/
Change-Id: I1e5267a828133e9a6579a6790f2996c64a88fc8d
Scrubbing the backlog directory.
This spec has been here since 2015.
The original authors should feel free to modify or downvote this commit
if the spec should continue to be considered for implementation now or
in the future.
Change-Id: I6e243b34dbe82aab2f91887a8a8f46e44d10afbb
Scrubbing the backlog directory.
This spec has been here since 2015.
The original authors should feel free to modify or downvote this commit
if the spec should continue to be considered for implementation now or
in the future.
Change-Id: Idec4df4b3fe689d73da4786993aaf8aa13c2dbf0
Scrubbing the backlog directory.
This spec has been here since 2015.
The original authors should feel free to modify or downvote this commit
if the spec should continue to be considered for implementation now or
in the future.
Change-Id: If15ca7f542774aefcf6b173196f72012fc4c062e
Adds tooling and enhances README documentation around the backlog specs
process.
- To move a spec from the backlog to the current release, we can now use
the ``move-spec`` tox target, e.g.
tox -e move-spec -- [-n] [-v] specs/backlog/approved/great-idea.rst specs/train/approved
- To abandon a backlog spec - i.e. move it from specs/backlog/approved
to the (new) specs/abandoned directory, we can now use the
``abandon-spec`` tox target, e.g.
tox -e abandon-spec -- [-n] [-v] specs/backlog/it-was-a-great-idea.rst
These utilities will move the specified spec into the target directory
and create an appropriate redirect for it.
To make it so, this commit factors out a helper method that a) moves a
spec from one subdirectory to another and b) adds a redirect for it.
This is used by the existing ``move-implemented-specs`` utility as well
as the new ``move-spec`` and ``abandon-spec``.
While I was in here, I spruced up the verbose output (including for
move-implemented-specs) to be a bit more readable.
Change-Id: I322eecbacd5dc52accf6ac69c9fe1116be8c216f
TrivialFix:
"automaticly" should be "automatically"
"instnaces" should be "instances"
"verison" should be "version"
Change-Id: If470b65277718a4c6d1c56b4d215b565babe67d0
Removing Project Priority from the backlog template, so tests pass for
new backlogs that don't include that heading. This has been done by
moving to use the Mitaka template for the backlog template.
This change also removes the heading from all the existing backlog
specs for consistency.
Change-Id: Iaa3d44d568e6a236c894af2c1afbba99bfc8aac3
A short history of what lead to the idea of instance tasks, and some of
the thinking that has taken place around it.
Change-Id: I4136d411ae078e3cfbd0a8494cf86b41dcc9a8c6
This describes a concept to group flavors together in order use those
groupings to scope policy and quotas.
Co-Authored-By: Vilobh Meshram <vilobhmm@yahoo-inc.com>
Change-Id: Ia0b2ac4b42f02c821ae5b36f008d66d5ef963a34
To help share ideas, adding a backlog spec to document the problems
with the current filter scheduler.
It also documents a number of possible solutions that have been
discussed in recent years, that fit together to fix the issues for
public clouds that currently use cells v1 and are looking at moving to
cells v2 in the M release.
Please note: there is currently no blueprint associated with this
specification, as its still in the problem definition and idea gathering
phase.
Change-Id: Ie52b862ace7801c17e3844d5c9ed8cb25a0592f1
Simple use case: Reduce the licence cost
Windows licences are billed by host or by socket. The goal is to stack all
Windows instances on the same hosts as much as possible without creating a
new host aggregate.
As much as possible means this is a best effort feature, if nova can place
this new Windows instance on a compute where other Windows are running, do
it. If not, spawn it where you want or on a empty compute to identify it
as « preferred widows compute » for future windows spawn.
This use case could be extended to many other situations, every time you
have a preferred situation with no hard restrictions.
Co-Authored-By: Michael Still <mikal@stillhq.com>
Change-Id: I261bcb802573406f0aca5a87606e4dcca877301b
* Link to Liberty template
* Add empty backlog/approved directory
* Update backlog index
* Since we cannot have an empty approved directory, link to
backlog-template for now.
Change-Id: I5f91e28f8d5d55397ba7d9f0c9881cce477a4e6b