This reverts commit 7348e9a253.
Since the nfs-ganesha rpm build for CentOS 8 has been fixed, and
the nfs-ganesha segfault caused by an issue in librgw has also been
fixed.
Signed-off-by: Dimitri Savineau <dsavinea@redhat.com>
this commit changes defaults value in default pool definitions.
there's no need to define `pg_num`, `pgp_num`, `size` and `min_size`,
`ceph_pool` module will use the current default if needed.
This also drops the 3 following `set_fact` in `ceph-facts`:
- osd_pool_default_pg_num,
- osd_pool_default_pgp_num,
- osd_pool_default_size_num
Signed-off-by: Guillaume Abrioux <gabrioux@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>
The `enable extras on centos` task just doesn't work when using the
variable ceph_docker_enable_centos_extra_repo to true.
fatal: [xxx]; FAILED! => {"changed": false, "msg": "Parameter
'baseurl', 'metalink' or 'mirrorlist' is required."}
The CentOS extras repository is enabled by default so it's pretty
safe to remove this task and the associated variable.
This also removes the ceph_docker_on_openstack variable as it's a
leftover and it is unused.
Signed-off-by: Dimitri Savineau <dsavinea@redhat.com>
Looks like nfs-ganesha 3.3 and 4.-dev doesn't work with recent changes
in librgw 16.0.0.
The nfs-ganesha daemon is segfaulting and restart in a loop.
See https://tracker.ceph.com/issues/47520
Signed-off-by: Dimitri Savineau <dsavinea@redhat.com>
This adds the ceph_dashboard_user ansible module for replacing the
command module usage with the ceph dashboard ac-user-xxx command.
Signed-off-by: Dimitri Savineau <dsavinea@redhat.com>
Even if the non containerized collocation scenario deploys ceph with
RPMs then we also deploy the dashboard/monitoring but with containers.
This requires to set the registry variable to ceph's quay.
Signed-off-by: Dimitri Savineau <dsavinea@redhat.com>
This changes the grafana container image regitry from docker.io to
quay.io to avoid rate limit.
This also adds the missing container image values for docker2podman
and podman scenarios.
Signed-off-by: Dimitri Savineau <dsavinea@redhat.com>
This commit diables nfs-ganesha testing on master for non-containerized
deployment because the dev repos are broken at the moment.
Signed-off-by: Guillaume Abrioux <gabrioux@redhat.com>
This re-adds the ceph-iscsi testing for both non containerized and
containerized deployment since the rados connection error on ceph
dev has been fixed [1].
[1] https://tracker.ceph.com/issues/47002
Signed-off-by: Dimitri Savineau <dsavinea@redhat.com>
This commit moves the erasure pool creation testing from `all_daemons`
to `lvm_osds` so we can decrease the number of osd nodes we spawn so the
OVH Jenkins slaves aren't less overwhelmed when a `all_daemons` based
scenario is being tested.
Signed-off-by: Guillaume Abrioux <gabrioux@redhat.com>
Since we've dropped ubuntu testing, we don't need these inventories
anymore. Let's remove this leftover.
Signed-off-by: Guillaume Abrioux <gabrioux@redhat.com>
Temporarily disable iscsigw testing for containerized deployments
because it's broken upstream on ceph@master.
non-containerized deployments use stable build for iscsigw to get around
this issue.
Signed-off-by: Guillaume Abrioux <gabrioux@redhat.com>
Since it is broken at the moment with dev repos, let's test against
stable builds so the CI is unlocked.
Signed-off-by: Guillaume Abrioux <gabrioux@redhat.com>
Otherwise we see some pytest warning.
PytestUnknownMarkWarning: Unknown pytest.mark.ceph_crash - is this a typo?
You can register custom marks to avoid this warning - for details,
see https://docs.pytest.org/en/latest/mark.html
Signed-off-by: Dimitri Savineau <dsavinea@redhat.com>
Due to [1], ceph-volume has now a dependency on pyyaml but it's not
installed by default via the package dependency.
This patch only add the required package on non containerized
deployment and as temporary workaround for the CI.
[1] https://tracker.ceph.com/issues/46759
Signed-off-by: Dimitri Savineau <dsavinea@redhat.com>
when rerunning lvm_setup.yml on existing cluster with OSDs already
deployed, it fails like following:
```
fatal: [osd0]: FAILED! => changed=false
msg: Sorry, no shrinking of data-lv2 to 0 permitted.
```
because we are asking `lvol` module to create a volume on an empty VG
with size extents = `100%FREE`.
The default behavior of `lvol` is to shrink the volume if the LV's current
size is greater than the requested size.
Given the requested size is calculated like this:
`size_requested = size_percent * this_vg['free'] / 100`
in our case, it is similar to:
`size_requested = 100 * 0 / 100` which basically means `0`
So the current LV size is well greater than the requested size which
leads the module to attempt to shrink it to 0 which isn't obviously now
allowed.
Adding `shrink: false` to the module calls fixes this issue.
Signed-off-by: Guillaume Abrioux <gabrioux@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>
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 commit enforces the pytest-rerunfailures installed so it's <9.0
This is to avoid the following error:
```
ERROR: pytest-rerunfailures 9.0 has requirement pytest>=5.0, but you'll have pytest 4.6.11 which is incompatible.
```
latest version of pytest-rerunfailures isn't compatible with the version
of pytest we are using.
Signed-off-by: Guillaume Abrioux <gabrioux@redhat.com>
This commit makes the zap function idempotent, especially when using
lvm_volumes variable.
Closes: https://bugzilla.redhat.com/show_bug.cgi?id=1845668
Signed-off-by: Guillaume Abrioux <gabrioux@redhat.com>
ansible 2.9.10 seems to have introduced a bug.
See https://github.com/ansible/ansible/issues/70168
This commit excludes this version from ceph-ansible requirements.
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>
This fixes a long standing fail in ceph-volumes lvm test suite.
Otherwise the default behaviour should not change.
Signed-off-by: Jan Fajerski <jfajerski@suse.com>
setting attributes with empty string is a bad user input.
Also, removing `rule_name` attribute when creating a code erasure pool.
(this rule isnt intended for code erasure pool type).
Signed-off-by: Guillaume Abrioux <gabrioux@redhat.com>
In containerized deployment, the ceph_volume module will always uses
the same container command prefix for all actions.
Instead of duplicate this code in all container tests we can define it
once.
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 CentOS 7 distribution could still be used be deploying ceph if
- it's a containerized deployment
- it's a non containerized deployment without the dashboard (due to
missing python3 libraries).
The ceph_stable_redhat_distro variable has been remove because we can
rely on the ansible_distribution_major_version fact instead.
The copr el8 repository configuration is only applied for CentOS 8.
The ceph-mgr-dashboard package is only installed when the
dashboard_enabled variable is set to true.
Signed-off-by: Dimitri Savineau <dsavinea@redhat.com>
The current centos/8 vagrant image (libvirt) is still using the
CentOS 8.0 release (1905) while the 8.1 release (1911) is already
available since few months.
Using an update CentOS 8 release fixes slow ceph-volume/lvm commands.
Signed-off-by: Dimitri Savineau <dsavinea@redhat.com>
With this change, the state `present` is enough to update a keyring.
If the keyring already exist, it will be updated if caps or secret
passed to the module are different.
If the keyring doen't exist, it will be created.
Closes: https://bugzilla.redhat.com/show_bug.cgi?id=1808367
Signed-off-by: Guillaume Abrioux <gabrioux@redhat.com>
Since Ceph Octopus is python3 only we don't need to specify the max open
files anymore with the container engine.
Signed-off-by: Dimitri Savineau <dsavinea@redhat.com>
When using the lvm batch ceph-volume subcommand with dedicated devices
for filestore (journal) or bluestore (db/wal) then the list of devices
is convert to a string instead of being extended via an iterable.
This was working with only one dedicated device but starting with more
then the ceph_volume module fails.
TASK [ceph-osd : use ceph-volume lvm batch to create bluestore osds] **
fatal: [xxxxxx]: FAILED! => changed=true
cmd:
- ceph-volume
- --cluster
- ceph
- lvm
- batch
- --bluestore
- --yes
- --prepare
- --osds-per-device
- '4'
- /dev/nvme2n1
- /dev/nvme3n1
- /dev/nvme4n1
- /dev/nvme5n1
- /dev/nvme6n1
- --db-devices
- /dev/nvme0n1 /dev/nvme1n1
- --report
- --format=json
msg: non-zero return code
rc: 2
stderr: |2-
stderr: lsblk: /dev/nvme0n1 /dev/nvme1n1: not a block device
stderr: error: /dev/nvme0n1 /dev/nvme1n1: No such file or directory
stderr: Unknown device, --name=, --path=, or absolute path in /dev/ or /sys expected.
usage: ceph-volume lvm batch [-h] [--db-devices [DB_DEVICES [DB_DEVICES ...]]]
[--wal-devices [WAL_DEVICES [WAL_DEVICES ...]]]
[--journal-devices [JOURNAL_DEVICES [JOURNAL_DEVICES ...]]]
[--no-auto] [--bluestore] [--filestore]
[--report] [--yes] [--format {json,pretty}]
[--dmcrypt]
[--crush-device-class CRUSH_DEVICE_CLASS]
[--no-systemd]
[--osds-per-device OSDS_PER_DEVICE]
[--block-db-size BLOCK_DB_SIZE]
[--block-wal-size BLOCK_WAL_SIZE]
[--journal-size JOURNAL_SIZE] [--prepare]
[--osd-ids [OSD_IDS [OSD_IDS ...]]]
[DEVICES [DEVICES ...]]
ceph-volume lvm batch: error: Unable to proceed with non-existing device: /dev/nvme0n1 /dev/nvme1n1
So the dedicated device list is considered as a single string.
This commit also adds the journal_devices, block_db_devices and
wal_devices documentation to the ceph_volume module.
Closes: https://bugzilla.redhat.com/show_bug.cgi?id=1816713
Signed-off-by: Dimitri Savineau <dsavinea@redhat.com>
Since 15ed9ee the ceph-mgr daemon binds on the IP address on the public
network instead of binding on all addresses.
This commit updates the testinfra code to reflect that change.
Signed-off-by: Dimitri Savineau <dsavinea@redhat.com>
Unregister marks generates warnings like:
PytestUnknownMarkWarning: Unknown pytest.mark.docker - is this a typo?
You can register custom marks to avoid this warning
https://docs.pytest.org/en/latest/mark.html
Signed-off-by: Dimitri Savineau <dsavinea@redhat.com>
This commit allows one to set the role for the admin user as read-only.
This can be controlled via the dashboard_admin_user_ro variable but the
default value is false for backward compatibility.
Closes: https://bugzilla.redhat.com/show_bug.cgi?id=1810176
Signed-off-by: Dimitri Savineau <dsavinea@redhat.com>
Since [1] we can't use osd pool without replicas (size: 1) by default.
We now need to set the mon_allow_pool_size_one flag to true in the ceph
configuration and add the --yes-i-really-mean-it flag to the osd pool
set size cli.
[1] https://github.com/ceph/ceph/commit/21508bd
Signed-off-by: Dimitri Savineau <dsavinea@redhat.com>
Make it so that more than one realm, zonegroup,
or zone can be created during a run of the rgw
multisite ansible playbooks.
The rgw hosts now need to be grouped into zones
and realms in the inventory.
.yml files need to be created in group_vars
for the realms and zones. Sample yaml files
are available.
Also remove multsite destroy playbook
and add --cluster before radosgw-admin commands
remove manually added rgw_zone_endpoints var
and have ceph-ansible automatically add the
correct endpoints of all the rgws in a rgw_zone
from the information provided in that rgws hostvars.
Signed-off-by: Ali Maredia <amaredia@redhat.com>
This commit changes the value passed for the attribute 'rule_name' in
openstack_pools definition. It doesn't make sense to have emptry string
as passed value here.
Signed-off-by: Guillaume Abrioux <gabrioux@redhat.com>
This commit makes the CI testing an OSD pool erasure creation due to the
recent refact of the OSD pool creation tasks in the playbook.
Signed-off-by: Guillaume Abrioux <gabrioux@redhat.com>
Looks like we are still seeing issue [1].
Let's increase this value to unlock the CI (however, it still needs to
be investigated).
Typical error (see [1] for further details) :
```
[root@osd2 ~]# ceph-volume --cluster ceph lvm batch --filestore --yes --journal-size '2048' /dev/sda /dev/sdb --journal-devices /dev/sdc
Running command: /sbin/vgcreate --force --yes ceph-journals-817ef90b-77ac-4f52-b8a9-30893849fb78 /dev/sdc
stdout: Physical volume "/dev/sdc" successfully created.
stdout: Volume group "ceph-journals-817ef90b-77ac-4f52-b8a9-30893849fb78" successfully created
--> Refusing to continue with configured size for journal
--> RuntimeError: journal sizes must be larger than 2GB, detected: 1024.00 MB
```
[1] https://tracker.ceph.com/issues/41374
Signed-off-by: Guillaume Abrioux <gabrioux@redhat.com>
Since [1] if a rgw user already exists then the radosgw-admin user create
command will return an error instead of modifying the current user.
We were already doing separated tasks for create and get operation but
only for multisite configuration but it's not enough.
Instead we should do the get task first and depending on the result
execute the create.
This commit also adds missing run_once and delegate_to statement.
[1] https://github.com/ceph/ceph/commit/269e9b9
Signed-off-by: Dimitri Savineau <dsavinea@redhat.com>
When osd_auto_discovery is set then we need to refresh the
ansible_devices fact between after the filestore OSD purge
otherwise the devices fact won't be populated.
Also remove the gpt header on ceph_disk_osds_devices because
the devices is empty at this point for osd_auto_discovery.
Adding the bool filter when needed.
Closes: https://bugzilla.redhat.com/show_bug.cgi?id=1729267
Signed-off-by: Dimitri Savineau <dsavinea@redhat.com>
We still need --destroy when using a raw device otherwise we won't be
able to recreate the lvm stack on that device with bluestore.
Running command: /usr/sbin/vgcreate -s 1G --force --yes ceph-bdc67a84-894a-4687-b43f-bcd76317580a /dev/sdd
stderr: Physical volume '/dev/sdd' is already in volume group 'ceph-b7801d50-e827-4857-95ec-3291ad6f0151'
Unable to add physical volume '/dev/sdd' to volume group 'ceph-b7801d50-e827-4857-95ec-3291ad6f0151'
/dev/sdd: physical volume not initialized.
--> Was unable to complete a new OSD, will rollback changes
Closes: https://bugzilla.redhat.com/show_bug.cgi?id=1792227
Signed-off-by: Dimitri Savineau <dsavinea@redhat.com>
The CentOS cloud infrastructure storing the vagrant CentOS 8 image
changed the directory path and remove the old 8.0 image so the vagrant
box add centos/8 fails returning a 404 http error.
As a workaround we can pull the image from CentOS instead of letting
vagrant doing the resolution.
Signed-off-by: Dimitri Savineau <dsavinea@redhat.com>
The nobarrier mount flag doesn't exist anymoer on XFS in the EL 8
kernel. That's why the task wasn't working on those systems.
We can still use the other options instead of skipping the task.
Signed-off-by: Dimitri Savineau <dsavinea@redhat.com>
Some docker commands were hardcoded in tests playbooks and some
conditions were not taking care of the containerized_deployment
variable but only the atomic fact.
Signed-off-by: Dimitri Savineau <dsavinea@redhat.com>
This commit adds a new job in order to test the
filestore-to-bluestore.yml infrastructure playbook.
Signed-off-by: Guillaume Abrioux <gabrioux@redhat.com>
Having max_mds value equals to the number of mds nodes generates a
warning in the ceph cluster status:
cluster:
id: 6d3e49a4-ab4d-4e03-a7d6-58913b8ec00a'
health: HEALTH_WARN'
insufficient standby MDS daemons available'
(...)
services:
mds: cephfs:3 {0=mds1=up:active,1=mds0=up:active,2=mds2=up:active}'
Let's use 2 active and 1 standby mds.
Signed-off-by: Dimitri Savineau <dsavinea@redhat.com>
The current ceph cluster health is in warning state:
health: HEALTH_WARN
13 pool(s) have no replicas configured
2 pool(s) have non-power-of-two pg_num
Because we're using only 1 replica then we need to disable the redundancy
check.
The pool pg num should be a power of two number (like 16).
Signed-off-by: Dimitri Savineau <dsavinea@redhat.com>
This adds device class support to crush rules when using the class key
in the rule dict via the create-replicated sub command.
If the class key isn't specified then we use the create-simple sub
command for backward compatibility.
Closes: https://bugzilla.redhat.com/show_bug.cgi?id=1636508
Signed-off-by: Dimitri Savineau <dsavinea@redhat.com>
This commit adds a playbook to be played before we run purge playbook,
it first creates an rbd image then map an rbd device on client0 so the
purge playbook will try to unmap it.
Signed-off-by: Guillaume Abrioux <gabrioux@redhat.com>
The ansible ssh connections are now using the ssh backend instead of
paramiko starting testinfra 3.1 and persistent connections too.
pytest 4.6 is the latest release to be supported by python 2.
Signed-off-by: Dimitri Savineau <dsavinea@redhat.com>
To avoid unnecessary ansible warnings during playbook execution we can
move the library and plugins test files under a different directory.
[WARNING]: Skipping plugin (plugins/filter/test_ipaddrs_in_ranges.py) as
it seems to be invalid:
cannot import name 'ipaddrs_in_ranges'
Closes: #4656
Signed-off-by: Dimitri Savineau <dsavinea@redhat.com>
This commit removes the backslash in allow command parameter, this was
needed before the ceph_key module integration.
Signed-off-by: Guillaume Abrioux <gabrioux@redhat.com>
Guillaume Abrioux
Dimitri Savineau
Jan Fajerski
Ali Maredia