| Commit message (Collapse) | Author | Age | Files | Lines |
|\
| |
| | |
WIP: update excluders to latest by default, in non-upgrade scenarios do not update
|
| |
| |
| |
| |
| |
| | |
- check both available excluder versions are at most of upgrade target version
- get excluder status through status command
- make excluders enablement configurable
|
|/ |
|
|
|
|
| |
upgrade README file
|
|
|
|
| |
Fixes Bug 1423425
|
|\
| |
| | |
Fixed issue where upgrade fails when using daemon sets (e.g. aggregated logging)
|
| | |
|
| | |
|
|\ \
| | |
| | | |
Add upgrade job step after the entire upgrade performs
|
| |/ |
|
|/ |
|
| |
|
|
|
|
|
| |
So that excluder is disabled and reset within the scope of each of those
in addition to the overall playbook
|
| |
|
| |
|
|\
| |
| | |
Correct usage of draining nodes
|
| | |
|
|/
|
|
|
|
|
| |
The add_host: task does not change any data on the host and as practice
has been configured to changed_when: False. This commit standardizes
that usage in the byo and common playbooks. Additionally, taks names
are added to each task to improve troubleshooting.
|
| |
|
|\
| |
| | |
Correct consistency between upgrade playbooks
|
| | |
|
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
This was done far into the process potentially leaving the user in a
difficult situation if they had now considered they were running the
upgrade playbook on a host that would be restarted. Instead check
configuration and what host we're running on in pre-upgrade and allow
the user to abort before making any substantial changes.
This is a step towards merging master upgrade into one serial process.
|
|\ \
| |/
|/| |
Begin requiring Docker 1.12.
|
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
Building off the work done for Docker 1.10, we now require Docker 1.12
by default.
The upgrade process was already set to ensure you are running the latest
docker during upgrade, and the standalone docker upgrade playbook can
also be used if desired.
As before, you can override this Docker 1.12 requirement by setting a
docker_version=1.10.3 (or similar), and you can skip the default to
upgrade docker by setting docker_upgrade=False.
|
| |
| |
| |
| | |
Closes #3070
|
|\ \
| | |
| | | |
Deprecate node 'evacuation' with 'drain'
|
| | |
| | |
| | |
| | | |
* https://trello.com/c/TeaEB9fX/307-3-deprecate-node-evacuation
|
|/ /
| |
| |
| | |
hook run.
|
| |
| |
| |
| |
| | |
* Added checks to make ci for yaml linting
* Modified y(a)ml files to pass lint checks
|
| | |
|
| | |
|
|\ \
| | |
| | | |
Remove duplicate when key
|
| | | |
|
|/ /
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
Router/registry update and re-deploy was recently reordered to
immediately follow control plane upgrade, right before we proceed to
node upgrade.
In some situations (small or single host clusters) it appears possible
that the deployer pods are running when the node in question is
evacuated for upgrade. When the deployer pod dies the deployment is
failed and the router/registry continue running the old version, despite
the deployment config being updated correctly.
This change re-orderes the router/registry upgrade to follow node
upgrade. However for separate control plane upgrade, the router/registry
still occurs at the end. This is because router/registry seems like they
should logically be included in a control plane upgrade, and presumably
the user will not manually launch node upgrade so quickly as to trigger
an evac on the node in question.
Workaround for this problem when it does occur is simply to:
oc deploy docker-registry --latest
|
|/
|
|
|
| |
On Fedora we just blindly upgrade to the latest.
On RHEL we do stepwise upgrades 2.0,2.1,2.2,2.3,3.0
|
| |
|
|
|
|
| |
This reverts commit 1f2276fff1e41c1d9440ee8b589042ee249b95d7.
|
|\
| |
| | |
Update link to latest versions upgrade README
|
| | |
|
|\ \
| | |
| | | |
Add support for 3.4 upgrade.
|
| |/
| |
| |
| |
| | |
This is a direct copy of 3.3 upgrade playbooks, with 3.3 specific hooks
removed and version numbers adjusted appropriately.
|
|\ \
| |/
|/| |
Fix and reorder control plane service restart.
|
| |
| |
| |
| |
| |
| |
| |
| |
| | |
This was missed in the standalone upgrade control plane playbook.
However it also looks to be out of order, we should restart before
reconciling and upgrading nodes. As such moved the restart directly into
the control plane upgrade common code, and placed it before
reconciliation.
|
|\ \
| | |
| | | |
Switch from "oadm" to "oc adm" and fix bug in binary sync.
|
| |/
| |
| |
| |
| |
| |
| |
| | |
Found bug syncing binaries to containerized hosts where if a symlink was
pre-existing, but pointing to the wrong destination, it would not be
corrected.
Switched to using oc adm instead of oadm.
|
|/ |
|
|
|
|
|
| |
Transition to being able to specify nodes to upgrade caused standalone
nodes to get skipped in this playbook.
|
| |
|
| |
|
|
|
|
|
|
|
|
| |
This improves the situation further and prevents configuration changes
from accidentally triggering docker restarts, before we've evacuated
nodes. Now in two places, we skip the role entirely, instead of previous
implementation which only skipped upgrading the installed version.
(which did not catch config issues)
|