| Age | Commit message (Collapse) | Author |
|---|
|
This commit continues moving plugins into lib_utils.
This commit does not move any plugins for add-on roles
such as logging and metrics.
|
|
Variables that are specifically booleans should be
cast to bool. This is because users may sometimes
pass them as string values. This is particularly
prevalent when using ini-style inventories.
Affected-by: https://github.com/ansible/ansible/issues/34591
Fixes: https://bugzilla.redhat.com/show_bug.cgi?id=1531592
|
|
We set these variables using facts in init, no need
to duplicate the logic all around the codebase.
|
|
This commit relocates filter_plugings to lib_utils,
changes the namespacing to prevent unintended use of
older versions that may be present in filter_plugins/
directory on existing installs.
Add lib_utils to meta depends for roles
Also consolidate some plugins into lib_utils from
various other areas.
Update rpm spec, obsolete plugin rpms.
|
|
|
|
|
|
This commit removes openshift.common.service_type
in favor of openshift_service_type.
This commit also removes r_openshift_excluder_service_type
from plays in favor of using the role's defaults.
|
|
Replace with `oc adm`
|
|
|
|
When a package install/update fails due to network blips or other spotty
availability, retry it. If the failure is a real failure (e.g. package
is really not there) it still fails after 3 tries (Ansible default).
|
|
This commit removes openshift.common.cli_image in
favor of openshift_cli_image.
|
|
Simplify the logic for this variable.
|
|
Signed-off-by: Steve Milner <smilner@redhat.com>
|
|
Signed-off-by: Giuseppe Scrivano <gscrivan@redhat.com>
|
|
When CRI-O is used and the CLI image is already pulled into Docker
then use it also for copying the CLI files to the host instead of
pulling it once again in the ostree storage.
Signed-off-by: Giuseppe Scrivano <gscrivan@redhat.com>
|
|
Most of this role's purpose was to set facts. The vast majority
of these facts were simply redefining user-supplied variables.
This commit also removes various artifacts leftover from
previous versions, as well as variables that seem to be
entirely unused.
|
|
crio: rename openshift_docker_use_crio to openshift_use_crio
|
|
It is confusing to have _docker_ in the name, since they are two
different backends.
Signed-off-by: Giuseppe Scrivano <gscrivan@redhat.com>
|
|
openshift_cli_facts only runs a single task, and otherwise
the only existence is to serve as a link in the dependency
chain towards openshift_facts.
This commit flattens openshift_cli into one role.
|
|
With origin 1.5, /usr/local/bin/oc was a symlink to
/usr/local/bin/openshift. During the container_binary_sync updated
versions of both binaries are copied to the host. First openshift is
copied to /usr/local/bin/openshift followed by copying oc to
/usr/local/bin/oc. Since oc is a symlink back to
/usr/local/bin/openshift the end result was everything linked to oc.
This change adds a check before copying a binary. If the destination is
a symlink then said symlink is removed before copying the new binary
over.
Fixed #4965
Reference: https://github.com/openshift/openshift-ansible/issues/4965
|
|
Use atomic to copy the CLI binaries to the host.
Signed-off-by: Giuseppe Scrivano <gscrivan@redhat.com>
|
|
In a project where contributors are free to use whatever editor they
want and we have linting tools that verify the proper formatting of
Python files, it should not be required to have a vim-specific line in
Python files.
|
|
|
|
The Ansible package module will call the correct package manager for the
underlying OS.
|
|
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.
|
|
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)
|
|
For containerized masters, we previously create a wrapper script to run
the docker CLI image and clean up afterward, but this approach is much
slower than having the binary on the native system.
Instead we now use an ansible module to handle the logic of syncing the
various OpenShift binaries and symlinks for client tools out onto the
host. The module will correctly report changed if modifications were
needed.
Substantial speed improvement for containerized installs which requires
many openshift/oc commands.
|
|
|
|
|
|
|
|
|
|
|
|
Allows the use of arbitrary tags, precise control over containers and
rpms, and likely mixed environments.
|
|
|
|
Docker 1.10 is no longer tolerant of commands like "docker pull
myimage:" when we do not have an image tag in play.
Adjust all occurrences with one that only includes the : if a version is
defined.
Adjust the containerized CLI wrappers for a similar problem.
|
|
Very early in playbooks we must init the openshift_version for each
host. First we determine it for the master, logic now is pushed into the
openshift_docker role which we run only on first master via
openshift_cli. Facts are reloaded leaving us with a first master with
openshift.common.version fact we can then re-use on all other hosts. The
correct version of docker should be installed as well.
We then set openshift_version for all other hosts by re-using the master
fact.
|
|
openshift_docker role was largely useless now, but also almost did what
we needed. (deps ordering still needs to be changed)
Remove defaulting of openshift_version.
|
|
Starting to remove openshift.docker.openshift_version fact usage.
openshift_version should no longer contain a leading 'v' for
containerized installs, just a version number.
|
|
|
|
|
|
|
|
QE found that for fresh installs we were basing the docker version facts of the
images that could be pulled prior to configuring /etc/sysconfig/docker. This
is an edge case but something we need to fix.
|
|
Fixing bugs 1322788 and 1323123
|
|
|
|
and atomic-openshift-master-controllers
|
|
- gather facts requiring docker only if docker is present and running
- Update reference to etcd role in playbooks/common/openshift-etcd/config.yml
to use openshift_etcd
|
|
- Prevents roles that need common facts from needing to require
openshift_common, which pulls in the openshift binary.
- Add dependency on openshift_facts to os_firewall, since it uses
openshift.common facts
|
|
- refactors the docker role to push generic config into docker role and wrap
openshift specific variables into an openshift_docker role and it's
dependent openshift_docker_facts role
- adds support for setting --confirm-def-push flag (Resolves
https://github.com/openshift/openshift-ansible/issues/1014)
- moves docker related facts from common/node roles to a new docker role
- renames cli_docker_* role varialbes to openshift_docker-* (maintaining
backward compatibility)
- update role dependencies to pull in openshift_docker conditionally based on
is_containerized
- remove playbooks/common/openshift-docker since the docker role is now
conditionally included
|
|
|
|
This is the containerized openshift_pkg_version equivalent. Originally I was
hoping to reuse openshift_pkg_version for containerized installs but the fact
that it's very coupled to yum made that pretty ugly.
However, I did opt to rely on the previously existing 'openshift_version'
variable. Containerized and RPM installs can both use that variable and it
will be set appropriately if either openshift_pkg_version or
openshift_image_tag are set. I suspect someday containerized installs will be
the only option and I didn't can to have thinkgs like openshift_pkg_version and
openshift_image_tag in the playbooks anymore the necessary.
|
