- When upgrading HA-clustered deployments with pause/resume,
hacluster units need to be paused first so VIP are migrated and
the unit being upgraded does not receive API requests.
- Currently nova-related charms do not support RPC version
pinning or auto-negotiation, so it is not recommended that
nova-compute is upgraded separately than nova-cloud-controller
(should be upgraded immediately after), as RPC version mismatch
will cause internal errors and API unavailability.
Change-Id: Ib63999848bae754e8fe61cd5f55345a000bc847c
Enabling RBD mirroring makes use of the journal feature of RBD's.
This has some significant overheads for write performance so
make sure this is apparent in the documentation for this feature.
Change-Id: I0e10d641606e25f6711d07eed75ba23214ec9bd7
Update metadata.yaml snippet for neutron-gateway.
Add 'NICs' column to a table on the install-maas page.
Clarify what 'data-port' is and how to configure all MAAS nodes so that
an unused network interface is left in the 'Unconfigured' state.
It is odd to have the user do this but the alternative was to be invasive in
terms of changes, which would result in a high risk of breakage later on. This
entire guide will be revisited and reviewed at a later date.
Closes-Bug: #1833625
Change-Id: I373eb238f10ab31ee43fdea251747393bc95b44b
There's no .md file in here, so change links. Looking at the used links,
these are all external links and thus sphinx cannot test whether they
work. Use internal links so that all links work - and work even after
renames.
This shows up as e.g.:
WARNING: unknown document: install-maas.html
Update tox.ini so that these are hard errors.
Remove dead requirement oslosphinx, it's not needed anymore here since
openstackdocstheme is used.
Change-Id: Ieae398aecb48c2e26de5c50261e700f8790e64ce
Now etcd and vault have two relations. Also, it's nice to have explicit
relations for others for the future.
Change-Id: Ic2652ba7d317e5adb93a5cf93a813d37f29efd87
Close-Bug: #1835165
Add new appendix detailing how to configure the Ceph RADOS
Gateway to make use of Erasure Coding and automatic storage
device classing.
Change-Id: I85ff5deae37e2471434a499fd4d97d9b577386ce
Ensure its clear that the authorize-charm action should be run
on the lead vault unit in a HA deployment by using the /leader
syntax to auto-identify the lead unit.
Add '--wait' parameter to example to ensure users see any action
failures.
Change-Id: I7dff6c3188d1877529e879adea4af9ffb8dee7c8
Closes-Bug: 1818660
Add a new appendix to detail how to use the multisite replication
features of the ceph-radosgw charm.
Change-Id: I3ac1a5ecab1449b700b2b41948ffcb5deb9d8fc1
Add instructions for migrating to using the cinder-ceph charm.
Depends-On: I93eb801ca4a166f862d5d86711d9476c61851344
Change-Id: Ibf98e98f938c186b8b7701fc763e7d1acb0521fb
Add pre-upgrade step to discover and purge and old compute service
entries; in later openstack releases, the DB model will throw
exceptions for service entries with versions less than the minimum
requirement, causing API requests to error post upgrade.
Change-Id: Id739f0f23aaadf08dff216bb9f17552deee1b0a4
Closes-Bug: 1809260
This change adds a new section to the upgrade appendix to list known
issues at the time that might not be resolvable via charm changes; i.e.
that need operator manual intervention.
Change-Id: I0250ac7dbb289e7e63703818a475f1854bf0f8a4
openstack-dev was decomissioned this night in https://review.openstack.org/621258
Update openstack-dev to openstack-discuss
Change-Id: I233cc3eb926c5bd0c0f6787e65c2bb5b74dc80f4
Closes-Bug: #1808310
Add new appendix for deployment, configuration and usage of
Octavia as part of an OpenStack Rocky deployment.
Change-Id: Id488b250055c1ec4e07821dda9d652e72e68e9ca
Ryan Beisner