Add prelude to Rocky releasenotes

Also removes the bullet point prelude that was in b3.

Change-Id: I3a5c71a93ab404f4a3bb236270706e97e0b12b0d
This commit is contained in:
Erno Kuvaja 2018-08-09 20:10:15 +01:00
parent 3e187ac364
commit 6884266251
3 changed files with 34 additions and 7 deletions

View File

@ -1,7 +1,4 @@
---
prelude: >
- The ``glance-manage`` utility has been updated to address OpenStack
Security Note OSSN-0075.
security:
- |
The ``glance-manage`` tool has been updated to address `OSSN-0075`_.

View File

@ -1,8 +1,4 @@
---
prelude: >
Automatic image conversion plugin for Interoperable Image Import. This
release introduces a new plugin that can be used to convert images to
specific format automatically upon image import.
features:
- |
Automatic image conversion plugin for Interoperable Image Import. With

View File

@ -0,0 +1,34 @@
---
prelude: |
This release of OpenStack Glance introduces 2 new API versions. Images
API v2.7 adds support and modifications for the Hidden Images and
Multihash features introduced during Rocky cycle. Version 2.8 is
included as an optional EXPERIMENTAL API for testing and preparing for
multiple back-end support.
Rocky development cycle marks long waited milestone on Glance work. The
Images API v1 which has been deprecated for years is finally removed and
not available at all in Glance version 17.0.0 forward.
Some security aspects were tackled for this release. Multihash, providing
secure hashing for image data with future proof options marks the end of
relying upon MD5 checksums when verifying image payloads. OSSN-0075
migitation lessens the risk of ID reusability on those very rare cases
when a database purge is necessary.
When delayed delete is enabled operators are able to recover image records
if the scrubber has been stopped before the data removal interval. While
the image metadata is still not preserved in these cases, this provides a
way to save the image data on accidental deletes.
Due to increasing core counts on modern servers Glance services started
consuming huge amounts of resources as the default was to spin up equal
amount of workers to logical CPUs seen on the host. This automatic scaling
was capped to 8 workers limiting the resources consumed. While each worker
can handle a pool of connections this limit should be sufficient for most
clouds. Large deployments should monitor their performance after upgrade.
When using Interoperable Image Import workflow, the cloud operators can
now enable automatic image conversion to desired format. When the plugin
is enabled end-users do not have any input to its operation but their
local checksum might not match with checksums recorded in Glance.