Since the default of `osd_objectstore` has changed as of 3.2, some
deployments might have a mix of filestore and bluestore OSDs on a same
node. In some specific cases, there's a possibility that a filestore OSD
shares a journal/db device with a bluestore OSD. We shouldn't try to
redeploy in this context because ceph-volume will complain. (either
because in lvm batch you can't pass partition or about gpt header).
The safest option is to skip the migration on the node when such a mix
is detected or force all osds including those already using bluestore
(option `force_filestore_to_bluestore=True` has to be passed as an extra var).
If all OSDs are using filestore, then they will be migrated to
bluestore.
Closes: https://bugzilla.redhat.com/show_bug.cgi?id=1875777
Signed-off-by: Guillaume Abrioux <gabrioux@redhat.com>
The current check makes no sense because it checks any of other monitor
than the one being played (either a previous one already converted or a
next that isn't yet converted) is present on the quorum.
Closes: https://bugzilla.redhat.com/show_bug.cgi?id=1909011
Signed-off-by: Guillaume Abrioux <gabrioux@redhat.com>
Instead of iterate over the host list for adding the node/label to the
host orchestrator configuration then we can do it parallelly.
Signed-off-by: Dimitri Savineau <dsavinea@redhat.com>
This adds cephadm_adopt ansible module for replacing the command module
usage with the cephadm adopt command.
Signed-off-by: Dimitri Savineau <dsavinea@redhat.com>
Let's discard the ansible lint error 306 and add a "# noqa 306" on tasks
where we don't need `set -o pipefail`
Fixes: #6090
Signed-off-by: Guillaume Abrioux <gabrioux@redhat.com>
We should always use the ceph_volume ansible module when possible.
This patch replace the ceph-volume inventory and lvm {list,zap} commands
called via the command/shell modules by the corresponding call with the
ceph_volume module.
Signed-off-by: Dimitri Savineau <dsavinea@redhat.com>
This adds cephadm_bootstrap ansible module for replacing the command module
usage with the cephadm bootstrap command.
Signed-off-by: Dimitri Savineau <dsavinea@redhat.com>
This adds ceph_osd_flag ansible module for replacing the command module
usage with the ceph osd set/unset commands.
Signed-off-by: Dimitri Savineau <dsavinea@redhat.com>
This adds ceph_osd ansible module for replacing the command module
usage with the ceph osd destroy/down/in/out/purge/rm commands.
Signed-off-by: Dimitri Savineau <dsavinea@redhat.com>
This adds ceph_mgr_module ansible module for replacing the command module
usage with the ceph mgr module enable/disable commands.
Signed-off-by: Dimitri Savineau <dsavinea@redhat.com>
`ceph.target` should be disabled only. Otherwise, in collocation
scenario you stop other collocated services in the OSD play which isn't
what we want to do. Each daemon has its corresponding play for managing
the transition to container.
Closes: https://bugzilla.redhat.com/show_bug.cgi?id=1901865
Signed-off-by: Guillaume Abrioux <gabrioux@redhat.com>
This adds ceph_volume_simple_{activate,scan} ansible modules for replacing
the command module usage with the ceph-volume simple activate/scan commands.
Signed-off-by: Dimitri Savineau <dsavinea@redhat.com>
adding monitor is no longer possible because we generate a new mon
keyring each time the playbook is run.
Fixes: #5864
Signed-off-by: Guillaume Abrioux <gabrioux@redhat.com>
ignore 302,303 and 505 errors
[302] Using command rather than an argument to e.g. file
[303] Using command rather than module
[505] referenced files must exist
they aren't relevant on these tasks.
Signed-off-by: Guillaume Abrioux <gabrioux@redhat.com>
fa2bb3a only fix the symlink owner/group issue in the OSD play. If the
OSDs are collocated with other services like MONs and MGRs then the
chown command will fail.
$ find /var/lib/ceph/osd/ceph-0 -not -user 167 -execdir chown 167:167 {} +
chown: cannot dereference './block': Permission denied
Closes: https://bugzilla.redhat.com/show_bug.cgi?id=1896448
Signed-off-by: Dimitri Savineau <dsavinea@redhat.com>
When deploying the ceph OSD via the packages then the ceph-osd@.service
unit is configured as enabled-runtime.
This means that each ceph-osd service will inherit from that state.
The enabled-runtime systemd state doesn't survive after a reboot.
For non containerized deployment the OSD are still starting after a
reboot because there's the ceph-volume@.service and/or ceph-osd.target
units that are doing the job.
$ systemctl list-unit-files|egrep '^ceph-(volume|osd)'|column -t
ceph-osd@.service enabled-runtime
ceph-volume@.service enabled
ceph-osd.target enabled
When switching to containerized deployment we are stopping/disabling
ceph-osd@XX.servive, ceph-volume and ceph.target and then removing the
systemd unit files.
But the new systemd units for containerized ceph-osd service will still
inherit from ceph-osd@.service unit file.
As a consequence, if an OSD host is rebooting after the playbook execution
then the ceph-osd service won't come back because they aren't enabled at
boot.
This patch also adds a reboot and testinfra run after running the switch
to container playbook.
Closes: https://bugzilla.redhat.com/show_bug.cgi?id=1881288
Signed-off-by: Dimitri Savineau <dsavinea@redhat.com>
cec994b introduced a regression when a mgr is collocated with a mon.
During the mon upgrade, the mgr service is masked to avoid to be
restarted on packages update.
Then the start mgr task is failing because the service is still masked.
Instead we should unmask it.
Fixes: #5983
Signed-off-by: Dimitri Savineau <dsavinea@redhat.com>
bd611a7 introduced the new ceph_fs module but missed some tasks in
rolling_update and shrink-mds playbooks.
Signed-off-by: Dimitri Savineau <dsavinea@redhat.com>
The ceph status command returns a lot of information stored in variables
and/or facts which could consume resources for nothing.
When checking the cluster health, we're using the health structure in the
ceph status output.
To optimize this, we could use the ceph health command which contains
the same needed information.
$ ceph status -f json | wc -c
2001
$ ceph health -f json | wc -c
46
Signed-off-by: Dimitri Savineau <dsavinea@redhat.com>
The ceph status command returns a lot of information stored in variables
and/or facts which could consume resources for nothing.
When checking the rgw/rbdmirror services status, we're only using the
servicmap structure in the ceph status output.
To optimize this, we could use the ceph service dump command which contains
the same needed information.
This command returns less information and is slightly faster than the ceph
status command.
$ ceph status -f json | wc -c
2001
$ ceph service dump -f json | wc -c
1105
$ time ceph status -f json > /dev/null
real 0m0.557s
user 0m0.516s
sys 0m0.040s
$ time ceph service dump -f json > /dev/null
real 0m0.454s
user 0m0.434s
sys 0m0.020s
Signed-off-by: Dimitri Savineau <dsavinea@redhat.com>
The ceph status command returns a lot of information stored in variables
and/or facts which could consume resources for nothing.
When checking the quorum status, we're only using the quorum_names
structure in the ceph status output.
To optimize this, we could use the ceph quorum_status command which contains
the same needed information.
This command returns less information.
$ ceph status -f json | wc -c
2001
$ ceph quorum_status -f json | wc -c
957
$ time ceph status -f json > /dev/null
real 0m0.577s
user 0m0.538s
sys 0m0.029s
$ time ceph quorum_status -f json > /dev/null
real 0m0.544s
user 0m0.527s
sys 0m0.016s
Signed-off-by: Dimitri Savineau <dsavinea@redhat.com>
The ceph status command returns a lot of information stored in variables
and/or facts which could consume resources for nothing.
When checking the pgs state, we're using the pgmap structure in the ceph
status output.
To optimize this, we could use the ceph pg stat command which contains
the same needed information.
This command returns less information (only about pgs) and is slightly
faster than the ceph status command.
$ ceph status -f json | wc -c
2000
$ ceph pg stat -f json | wc -c
240
$ time ceph status -f json > /dev/null
real 0m0.529s
user 0m0.503s
sys 0m0.024s
$ time ceph pg stat -f json > /dev/null
real 0m0.426s
user 0m0.409s
sys 0m0.016s
The data returned by the ceph status is even bigger when using the
nautilus release.
$ ceph status -f json | wc -c
35005
$ ceph pg stat -f json | wc -c
240
Signed-off-by: Dimitri Savineau <dsavinea@redhat.com>
Instead of using ceph auth get command via the ansible command module
then we can use the ceph_key module and the info state.
Signed-off-by: Dimitri Savineau <dsavinea@redhat.com>
This playbook isn't needed anymore, we can achieve this operation by
running main playbook with `--limit` option.
Signed-off-by: Guillaume Abrioux <gabrioux@redhat.com>
This adds the ceph_fs ansible module for replacing the command module
usage with the ceph fs command.
Signed-off-by: Dimitri Savineau <dsavinea@redhat.com>
This commit adds the `osd_auto_discovery` scenario support in the
filestore-to-bluestore playbook.
Closes: https://bugzilla.redhat.com/show_bug.cgi?id=1881523
Signed-off-by: Guillaume Abrioux <gabrioux@redhat.com>
Co-authored-by: Dimitri Savineau <dsavinea@redhat.com>
This change default value of grafana-server group name.
Adding some tasks in ceph-defaults in order to keep backward
compatibility.
Signed-off-by: Guillaume Abrioux <gabrioux@redhat.com>
Otherwise this will generate an ansible warning about the missing
filter.
[DEPRECATION WARNING]: evaluating xxx as a bare variable, this behaviour
will go away and you might need to add |bool to the expression in the
future.
Also see CONDITIONAL_BARE_VARS configuration toggle.. This feature will
be removed in version 2.12.
Signed-off-by: Dimitri Savineau <dsavinea@redhat.com>
In Pacific we're are sure that users already achieved the msgr2 because
that was introduced in Nautilus.
Signed-off-by: Dimitri Savineau <dsavinea@redhat.com>
There's no need ot have a copy of this file in infrastructure-playbooks
directory.
playbooks in that directory can be run from the root dir of
ceph-ansible.
Signed-off-by: Guillaume Abrioux <gabrioux@redhat.com>
As of 2.10, group names containing a dash are invalid.
However, setting this option makes it still possible to use a dash in
group names and prevent this warning to show up.
It might need to be definitely addressed in a future ansible release.
Closes: https://bugzilla.redhat.com/show_bug.cgi?id=1880476
Signed-off-by: Guillaume Abrioux <gabrioux@redhat.com>
When running the switch2container playbook on a Debian based system
then the systemd unit path isn't the same than Red Hat based system.
Because the systemd unit files aren't removed then the new container
systemd unit isn't take in count.
Signed-off-by: Dimitri Savineau <dsavinea@redhat.com>
Most ansible module using a state parameter default to the present
value (when available) instead of using it as a mandatory option.
Signed-off-by: Dimitri Savineau <dsavinea@redhat.com>
We already do this in the site-container.yml playbook because we don't
need docker/podman installed on all client nodes and having the
container image only on the first client node.
Signed-off-by: Dimitri Savineau <dsavinea@redhat.com>
When running the rolling_update playbook with an inventory without
monitor nodes defined (like external scenario) then we can't retrieve
the cluster fsid from the running monitor.
In this scenario we have to pass this information manually (group_vars
or host_vars).
Closes: https://bugzilla.redhat.com/show_bug.cgi?id=1877426
Signed-off-by: Dimitri Savineau <dsavinea@redhat.com>
In the OSP context, during the rolling update the playbook fails
with the following error:
'''
ERROR! The field 'hosts' has an invalid value, which includes an
undefined variable. The error was: list object has no element 0
'''
This PR just change the hosts field providing a valid mons group
value.
Closes: https://bugzilla.redhat.com/1876803
Signed-off-by: Francesco Pantano <fpantano@redhat.com>
On DCN environments, or when multiple ceph cluster are configured,
we need to specify the cluster name before running the command or
the rolling_update playbook will fail during minor updates.
Closes: https://bugzilla.redhat.com/1876447
Signed-off-by: Francesco Pantano <fpantano@redhat.com>
There's no need to use `ignore_errors: true` on these tasks.
Using a loop on the task stopping mon daemons allows us to avoid
duplicating this task, the `ignore_errors` isn't needed here because it
won't fail the playbook if one of the ID doesn't exist (shortname vs. fqdn)
Using the right condition on the task starting the mgr daemon allows us
to avoid using an `ignore_errors: true` as well.
Signed-off-by: Guillaume Abrioux <gabrioux@redhat.com>
When using fqdn in inventory host file, this task will fail because the
mds is registered with its shortname.
It means we must use `mds_to_kill_hostname` in this task.
Closes: https://bugzilla.redhat.com/show_bug.cgi?id=1869837
Signed-off-by: Guillaume Abrioux <gabrioux@redhat.com>
ceph-volume can generate large logs at some point.
debug logs by definition should be enabled only when debugging.
Let's make it customizable with a variable which is set to `False` by
default.
Signed-off-by: Guillaume Abrioux <gabrioux@redhat.com>
When running `infrastructure-playbooks/purge-cluster.yml` twice, it fails the
second time on the `ensure rbd devices are unmapped` task, because `rbdmap`
isn't installed anymore at that point.
This commit adds a check that ensures `rbdmap` is available, and skips the
`ensure rbd devices are unmapped` task if it isn't.
Signed-off-by: Benoît Knecht <bknecht@protonmail.ch>
This handles missing /etc/ceph/osd, by ensuring we actually found files in
`/etc/ceph/osd` before trying to slurp their content.
This also add a missing `| default(False)` to avoid fowlloing error:
```
fatal: [ceph01]: FAILED! =>
msg: |-
The conditional check 'ceph_osd_data_json[item.2]['encrypted'] | bool' failed. The error was: error while evaluating conditional (ceph_osd_data_json[item.2]['encrypted'] | bool): 'dict object' has no attribute 'encrypted'
```
Closes: https://bugzilla.redhat.com/show_bug.cgi?id=1862416
Signed-off-by: Benoît Knecht <bknecht@protonmail.ch>
The task "remove old systemd unit file" under "switching from
non-containerized to containerized ceph rgw" only removes
the ceph-radosgw@.service file. The task should also remove
the ceph-radosgw.target file, like the "remove old systemd unit
files" tasks for the mons, mgrs, osds, etc, in order to clean up
all of the unused systemd unit files.
Signed-off-by: Kevin Coakley <kcoakley@sdsc.edu>
Otherwise it leaves an empty directory.
When shrinking and redeploying multiple OSDs you have no guarantee it
will reuse the same osd id.
Signed-off-by: Guillaume Abrioux <gabrioux@redhat.com>
In addition of 155e2a2, the active mds daemons isn't stop/start
correctly as opposed as the other services so that daemon doesn't come
back after the upgrade.
Closes: https://bugzilla.redhat.com/show_bug.cgi?id=1861688
Signed-off-by: Dimitri Savineau <dsavinea@redhat.com>
The dashboard upgrade workflow should do the same process than the ceph
upgrade otherwise any systemd unit modification won't be apply on the
monitoring/dashboard stack.
Closes: https://bugzilla.redhat.com/show_bug.cgi?id=1859173
Signed-off-by: Dimitri Savineau <dsavinea@redhat.com>
During the daemon upgrade we're
- stopping the service when it's not containerized
- running the daemon role
- start the service when it's not containerized
- restart the service when it's containerized
This implementation has multiple issue.
1/ We don't use the same service workflow when using containers
or baremetal.
2/ The explicity daemon start isn't required since we'are already
doing this in the daemon role.
3/ Any non backward changes in the systemd unit template (for
containerized deployment) won't work due to the restart usage.
This patch refacts the rolling_update playbook by using the same service
stop task for both containerized and baremetal deployment at the start
of the upgrade play.
It removes the explicit service start task because it's already included
in the dedicated role.
The service restart tasks for containerized deployment are also
removed.
Finally, this adds the missing service stop task for ceph crash upgrade
workflow.
Closes: https://bugzilla.redhat.com/show_bug.cgi?id=1859173
Signed-off-by: Dimitri Savineau <dsavinea@redhat.com>
This commit introduces a new role `ceph-crash` in order to deploy
everything needed for the ceph-crash daemon.
Signed-off-by: Guillaume Abrioux <gabrioux@redhat.com>
Set the cephadm cmd as a fact instead of rewriting the same command
over and over.
This also fix an issue when using docker as container engine because
the --docker cephadm parameter should be use before the subcommand
not after.
Signed-off-by: Dimitri Savineau <dsavinea@redhat.com>
This adds a new playbook for deploying ceph via cephadm.
This also adds a new dedicated tox file for CI purpose.
Signed-off-by: Dimitri Savineau <dsavinea@redhat.com>
This is a partial revert of b38019e because we don't want to execute
the whole play on the monitor otherwise if we have some empty group
like rgws or mdss then the orchestrator commands will still be
executed.
Instead we should keep the real target group name at play level and
delegate the orchestator commands to the monitor. The whole play
will be skipped is the group is empty.
Signed-off-by: Dimitri Savineau <dsavinea@redhat.com>
Print a message at the end of the playbook to inform users that they
don't have to user ceph-ansible playbooks anymore as everything else
need to be done via cephadm (day 2 operation).
Signed-off-by: Dimitri Savineau <dsavinea@redhat.com>
When reporting the orchestrator service/daemon list at the end of the
playbook, we can use the --refresh option otherwise we could have
an outdated output.
Signed-off-by: Dimitri Savineau <dsavinea@redhat.com>
After adopting a monitor we need to wait that monitor to join back
the quorum before moving to the next node.
Signed-off-by: Dimitri Savineau <dsavinea@redhat.com>
Like rolling_update or switch2container playbooks, we need to set/unset
some osd flags before and after the OSD daemons adoption.
This also adds a task for waiting for clean pgs at then of an OSd node.
Signed-off-by: Dimitri Savineau <dsavinea@redhat.com>
At the end of the playbook we can show the orchestrator status like
we do with the ceph status in initial deployment.
Signed-off-by: Dimitri Savineau <dsavinea@redhat.com>
It's better to use the --placement parameter when using ceph orch apply
commands to avoid confusion in the parameters.
Signed-off-by: Dimitri Savineau <dsavinea@redhat.com>
cephadm uses default value for dashboard container images which need to
be customized by ansible for upstream or downstream purpose.
This feature wasn't present when cephadm-adopt.yml has been designed.
Also set the container_image_base variable for upgrade purpose.
Signed-off-by: Dimitri Savineau <dsavinea@redhat.com>
It looks like we can't run the ceph orch apply commands on nodes other
than monitors even if it used to work in the past.
Signed-off-by: Dimitri Savineau <dsavinea@redhat.com>
The ceph config assimilate-conf command requires the client.admin
keyring which isn't present on all nodes most of the time.
Signed-off-by: Dimitri Savineau <dsavinea@redhat.com>
By default, ansible gathers facts from facter and ohai if installed on
the remote nodes, given we don't need them, let's exclude these facts
from our facts gathering
Signed-off-by: Dimitri Savineau <dsavinea@redhat.com>
When rgw and osd are collocated, the current workflow prevents from
scaling out the radosgw_num_instances parameter when rerunning the
playbook.
The environment file used in the rgw systemd template is rendered when
executing the `ceph-rgw` role but during a new run of the playbook (in
order to scale out rgw instances), handlers are triggered from `ceph-osd`
role which is run before `ceph-rgw`, therefore it tries to start the new
rgw daemon whereas its corresponding environment file hasn't been
rendered yet and fails like following:
```
ceph-radosgw@rgw.ceph4osd3.rgw1.service failed to run 'start-pre' task: No such file or directory
```
This commit moves the tasks generating this file in `ceph-config` role
so it is generated early.
Closes: https://bugzilla.redhat.com/show_bug.cgi?id=1851906
Signed-off-by: Guillaume Abrioux <gabrioux@redhat.com>
If a failure occurs in ceph-validate, the upgrade playbook keeps running
where we expect it to fail.
Signed-off-by: Guillaume Abrioux <gabrioux@redhat.com>
The commit adds a new playbook for converting an existing ceph cluster
deployed by ceph-ansible to the cephadm orchestrator.
Signed-off-by: Dimitri Savineau <dsavinea@redhat.com>
This commit makes the images pulling skipped if podman isn't installed
on the machine.
In OSP context, the podman installation is done later in the workflow,
it means all `podman pull` commands will fail.
Closes: https://bugzilla.redhat.com/show_bug.cgi?id=1849559
Signed-off-by: Guillaume Abrioux <gabrioux@redhat.com>
Since we only have one scenario since nautilus then we can just move
the container start command from ceph-osd-run.sh to the systemd unit
service.
As a result, the ceph-osd-run.sh.j2 template and the
ceph_osd_docker_run_script_path variable are removed.
Signed-off-by: Dimitri Savineau <dsavinea@redhat.com>
We shouldn't set this flag when running switch_to_containers playbook.
Otherwise the playbook fails waiting for pgs to be clean.
Closes: https://bugzilla.redhat.com/show_bug.cgi?id=1843569
Signed-off-by: Guillaume Abrioux <gabrioux@redhat.com>
The systemd LOAD and ACTIVE fileds could have more than one space between
both values.
This update the systemd regex the same way we're using it in different
part of the code.
Closes: https://bugzilla.redhat.com/show_bug.cgi?id=1843500
Signed-off-by: Dimitri Savineau <dsavinea@redhat.com>
The dashboard nodes (alertmanager, grafana, node-exporter, and prometheus)
were not manage during the docker to podman migration.
This adds the systemd container template of those services to a dedicated
file (systemd.yml) in order to include it in the docker2podman playbook.
This also adds the dashboard container images pull from docker to podman.
Closes: https://bugzilla.redhat.com/show_bug.cgi?id=1829389
Signed-off-by: Dimitri Savineau <dsavinea@redhat.com>
The docker2podman playbook only installs the podman package and updates
the systemd units with the right container_binary value.
We never pull the container image so if one service is restarted then
the container image will be pulled first before the service can start
which could cause longer downstream.
To avoid to download the container image from internet again we can just
pull it from the local docker daemon.
The container_{binding,package,service}_name variables are removed
because they are only used in the ceph-container-engine role which
isn't call in this playbook.
Signed-off-by: Dimitri Savineau <dsavinea@redhat.com>
When using skipped variables with from_json filter and python2 then we
need to have a default value otherwise the skipped task will fail.
Unexpected templating type error occurred on
({{ (ceph_volume_lvm_list.stdout | from_json) }}): expected string or
buffer
Closes: https://bugzilla.redhat.com/show_bug.cgi?id=1790472
Signed-off-by: Dimitri Savineau <dsavinea@redhat.com>
Since [1] we need to set pacific for the required OSD release during the
upgrade.
[1] https://github.com/ceph/ceph/commit/cc99c3bc
Signed-off-by: Dimitri Savineau <dsavinea@redhat.com>
The workflow in this playbook should be the same than in rolling_update,
we should first set noout and nodeep-scrub flags before migrating the
first osd and unset osd flags after the last osd is migrated.
Signed-off-by: Guillaume Abrioux <gabrioux@redhat.com>
When setting/unsetting osd flags, we can use `tasks_from` when importing
`ceph-facts` role to save some times given that we only need this role
for setting `container_binary`
Signed-off-by: Guillaume Abrioux <gabrioux@redhat.com>
We must call `ceph-osd` role from `container_options_facts.yml` because
ceph-osd-run.sh.j2 needs variables set in this file.
Closes: https://bugzilla.redhat.com/show_bug.cgi?id=1819681
Signed-off-by: Guillaume Abrioux <gabrioux@redhat.com>
Guillaume Abrioux
Dimitri Savineau
Francesco Pantano
Benoît Knecht
Kevin Coakley