Commit Graph

3601 Commits (43e23ffe4dc0ca303e5cd124fa2231a771f26dca)
 

Author SHA1 Message Date
Andrew Schoen 08f4875533 ceph_volume: refactor to not run ceph osd destroy
This changes state to action and gives the options 'create'
or 'zap'. The zap parameter is also removed.

Signed-off-by: Andrew Schoen <aschoen@redhat.com>
2018-04-10 14:19:21 +02:00
Andrew Schoen 36e71f6532 ceph_volume: perserve newlines in stdout and stderr when zapping
Because we have many commands we might need to run the
ANSIBLE_STDOUT_CALLBACK won't format these nicely because we're
not reporting these back at the root level of the json result.

Signed-off-by: Andrew Schoen <aschoen@redhat.com>
2018-04-10 14:19:21 +02:00
Andrew Schoen c6e8f8fb11 purge-cluster: no need to use objectstore for ceph_volume module
When zapping objectstore is not required.

Signed-off-by: Andrew Schoen <aschoen@redhat.com>
2018-04-10 14:19:21 +02:00
Andrew Schoen a8b0d3f045 ceph_volume: rc should be 0 on successful runs
Signed-off-by: Andrew Schoen <aschoen@redhat.com>
2018-04-10 14:19:21 +02:00
Andrew Schoen dbd527411c ceph_volume: defines the zap param in module_args
Signed-off-by: Andrew Schoen <aschoen@redhat.com>
2018-04-10 14:19:21 +02:00
Andrew Schoen a9b4c01b7c ceph_volume: make state not required so I can provide a default
I want a default value of 'present' for state, so it can not
be made required. Othewise it'll throw a 'Module alias error'
from ansible.

Signed-off-by: Andrew Schoen <aschoen@redhat.com>
2018-04-10 14:19:21 +02:00
Andrew Schoen 9f469e2c5a ceph_volume: objectstore is now optional except when state is present
Signed-off-by: Andrew Schoen <aschoen@redhat.com>
2018-04-10 14:19:21 +02:00
Andrew Schoen c29a75ac7f purge-cluster: use ceph_volume module to zap and destroy OSDs
Signed-off-by: Andrew Schoen <aschoen@redhat.com>
2018-04-10 14:19:21 +02:00
Andrew Schoen 98e237d234 tests: no need to remove partitions in lvm_setup.yml
Now that we are using ceph_volume_zap the partitions are
kept around and should be able to be reused.

Signed-off-by: Andrew Schoen <aschoen@redhat.com>
2018-04-10 14:19:21 +02:00
Andrew Schoen 59e700c9fe ceph_volume: adds a zap property and reworks to support state: absent
Signed-off-by: Andrew Schoen <aschoen@redhat.com>
2018-04-10 14:19:21 +02:00
Andrew Schoen b553a73443 ceph_volume: adds a state property
This can be either present or absent.

Signed-off-by: Andrew Schoen <aschoen@redhat.com>
2018-04-10 14:19:21 +02:00
Andrew Schoen 796caf7d8d ceph_volume: remove the subcommand argument
This really isn't needed currently and I don't believe is a good
mechanism for switching subcommands anwyay. The user of this module
should not have to be familar with all ceph-volume subcommands.

Signed-off-by: Andrew Schoen <aschoen@redhat.com>
2018-04-10 14:19:21 +02:00
Randy J. Martinez d1f2d64b15 purge-docker: added conditionals needed to successfully re-run purge
Added 'ignore_errors: true' to multiple lines which run docker commands; even in cases where docker is no longer installed. Because of this, certain tasks in the purge-docker-cluster.yml will cause the playbook to fail if re-run and stop the purge. This leaves behind a dirty environment, and a playbook which can no longer be run.
Fix Regex line 275: Sometimes 'list-units' will output 4 spaces between loaded+active. The update will account for both scenarios.
purge fetch_directory: in other roles fetch_directory is hard linked ex.: "{{ fetch_directory }}"/"{{ somedir }}". That being said, fetch_directory will never have a trailing slash in the all.yml so this task was never being run(causing failures when trying to re-deploy).

Signed-off-by: Randy J. Martinez <ramartin@redhat.com>
2018-04-10 13:39:14 +02:00
JohnHaan ad26d842f5 Fixed wrong path of ceph.conf in docs.
The path of ceph.conf sample template moved to ceph-config.
Therefore docs needs to be changed to the right directory.

Signed-off-by: JohnHaan <yongiman@gmail.com>
2018-04-10 10:33:25 +02:00
Guillaume Abrioux 66c4118dcd defaults: fix backward compatibility
backward compatibility with `ceph_mon_docker_interface` and
`ceph_mon_docker_subnet` was not working since there wasn't lookup on
`monitor_interface` and `public_network`

Signed-off-by: Guillaume Abrioux <gabrioux@redhat.com>
2018-04-10 00:19:11 +02:00
Ken Dreyer 3752cc6f38 common: upgrade/install ceph-test RPM first
Prior to this change, if a user had ceph-test-12.2.1 installed, and
upgraded to ceph v12.2.3 or newer, the RPM upgrade process would
fail.

The problem is that the ceph-test RPM did not depend on an exact version
of ceph-common until v12.2.3.

In Ceph v12.2.3, ceph-{osdomap,kvstore,monstore}-tool binaries moved
from ceph-test into ceph-base. When ceph-test is not yet up-to-date, Yum
encounters package conflicts between the older ceph-test and newer
ceph-base.

When all users have upgraded beyond Ceph < 12.2.3, this is no longer
relevant.
2018-04-09 18:09:52 +02:00
Sébastien Han bb60f2fea4 ceph-defaults: fix ceoh_uid for container image tag latest
According to our recent change, we now use "CentOS" as a latest
container image. We need to reflect this on the ceph_uid.

Signed-off-by: Sébastien Han <seb@redhat.com>
2018-04-09 13:54:55 +02:00
Sébastien Han 14eff6b571 tox: use container latest tag for upgrades
Currently tag-build-master-luminous-ubuntu-16.04 is not used anymore.
Also now, 'latest' points to CentOS so we need to make that switch here
too.

We know have latest tags for each stable release so let's use them and
point tox at them to deploy the right version.

Signed-off-by: Sébastien Han <seb@redhat.com>
2018-04-09 13:54:55 +02:00
Zack Cerza 0123d790cd Use the CentOS repo for Red Hat dev packages
No use even trying to use something that doesn't exist.

Signed-off-by: Zack Cerza <zack@redhat.com>
2018-04-09 10:05:57 +02:00
Guillaume Abrioux d981c6bd20 site-docker: followup on #2487
get a non empty array as default value for `groups.get('clients')`,
otherwise `| first` filter will complain because it can't work with
empty array.

Signed-off-by: Guillaume Abrioux <gabrioux@redhat.com>
2018-04-05 11:18:53 +02:00
Sébastien Han 3c4319ca4b add .vscode/ to gitignore
I personally dev on vscode and I have some preferences to save when it
comes to running the python unit tests. So escaping this directory is
actually useful.

Signed-off-by: Sébastien Han <seb@redhat.com>
2018-04-04 18:18:19 +02:00
Attila Fazekas ecd3563c21 Deploying without managed monitors failed
Tripleo deployment failed when the monitors not manged
by tripleo itself with:
    FAILED! => {"msg": "list object has no element 0"}

The failing play item was introduced by
 f46217b69a .

fixes: https://bugzilla.redhat.com/show_bug.cgi?id=1552327

Signed-off-by: Attila Fazekas <afazekas@redhat.com>
2018-04-04 18:16:46 +02:00
Guillaume Abrioux dcf6a246a4 defaults: remove `run_once: true` when creating fetch_directory
because of `serial: 1`, it can be an issue when the playbook is being
run on client nodes.
Since the refact of `ceph-client` we skip the role `ceph-defaults` on
every node except the first client node, it means that the task is not
going to be played because of `run_once: true`.

Signed-off-by: Guillaume Abrioux <gabrioux@redhat.com>
2018-04-04 10:51:17 +02:00
Guillaume Abrioux 18c0c7a508 config: use fact `ceph_uid`
Use fact `ceph_uid` in the task which ensures `/etc/ceph` exists in
containerized deployments.

Signed-off-by: Guillaume Abrioux <gabrioux@redhat.com>
2018-04-04 10:51:17 +02:00
Guillaume Abrioux 9c979c6390 clients: refact `ceph-clients` role
This commit refacts this role so we don't have to pull container image
on client nodes just to create pools and keys.

Fixes: https://bugzilla.redhat.com/show_bug.cgi?id=1550977

Signed-off-by: Guillaume Abrioux <gabrioux@redhat.com>
2018-04-04 10:51:17 +02:00
Guillaume Abrioux cefd471967 client: remove legacy code
This seems to be a leftover.
This commit removes an unnecessary 'set linux permissions' on
`/var/lib/ceph`

Signed-off-by: Guillaume Abrioux <gabrioux@redhat.com>
2018-04-04 10:51:17 +02:00
Guillaume Abrioux 9d3517c670 container: play docker-common only on first client node
This commit aims to set the default behavior to play
`ceph-docker-common` only on first node in clients group.

Currently, we play docker-common to pull container image so we can run
ceph commands in order to generate keys or create pools.
On a cluster with a large number of client nodes this can be time consuming
to proceed this way. An alternative would be to pull container image
only a first node and then copy keys on other nodes.

Signed-off-by: Guillaume Abrioux <gabrioux@redhat.com>
2018-04-04 10:51:17 +02:00
Guillaume Abrioux cf27c5e941 move selinux check to `ceph-defaults`
This check is alone in `ceph-docker-common` since a previous code
refactor.
Moving this check in `ceph-defaults` allows us to run `ceph-clients`
without having to run `ceph-docker-common` even in non-containerized
deployment.

Signed-off-by: Guillaume Abrioux <gabrioux@redhat.com>
2018-04-04 10:51:17 +02:00
Sébastien Han f3caee8460 ceph-iscsi: fix certificates generation and distribution
Prior to this patch, the certificates where being generated on a single
node only (because of the run_once: true). Thus certificates were not
distributed on all the gateway nodes.

This would require a second ansible run to work. This patches fix the
creation and keys's distribution on all the nodes.

Closes: https://bugzilla.redhat.com/show_bug.cgi?id=1540845
Signed-off-by: Sébastien Han <seb@redhat.com>
2018-04-04 09:27:39 +02:00
Guillaume Abrioux 5b73be254d do not delegate facts on client nodes
This commit is a workaround for
https://bugzilla.redhat.com/show_bug.cgi?id=1550977

We iterate over all nodes on each node and we delegate the facts gathering.
This is high memory consuming when having a large number of nodes in the
inventory.
That way of gathering is not necessary for clients node so we can simply
gather local facts for these nodes.

Signed-off-by: Guillaume Abrioux <gabrioux@redhat.com>
2018-04-04 09:23:03 +02:00
Guillaume Abrioux e32a177af8 purge-docker: remove redundant task
The `remove_packages` prompt is redundant to the `ireallymeanit` prompt
since it does exactly the same thing. I guess the only goal of this task
was to make a break to warn user about `--skip-tags=with_pkg` feature.
This warning should be part of the first prompt.

Signed-off-by: Guillaume Abrioux <gabrioux@redhat.com>
2018-04-03 11:54:42 +02:00
Randy J. Martinez ca572a11f1 ceph-mds: delete duplicate tasks which cause multimds container deployments to fail.
This update will resolve error['cephfs' is undefined.] in multimds container deployments.
See: roles/ceph-mon/tasks/create_mds_filesystems.yml. The same last two tasks are present there, and actully need to happen in that role since "{{ cephfs }}" gets defined in
roles/ceph-mon/defaults/main.yml, and not roles/ceph-mds/defaults/main.yml.

Signed-off-by: Randy J. Martinez <ramartin@redhat.com>
2018-03-29 09:32:40 +02:00
Alfredo Deza 3fcf966803 ceph-osd note that some scenarios use ceph-disk vs. ceph-volume
Signed-off-by: Alfredo Deza <adeza@redhat.com>
2018-03-29 09:11:33 +02:00
John Fulton e6e6bd078a Refer to expected-num-ojects as expected_num_objects, not size
Follow up patch to PR 2432 [1] which replaces "size" (sorry if
the original bug used that term, which can be confusing) with
expected_num_objects as is used in the Ceph documentation [2].

[1] https://github.com/ceph/ceph-ansible/pull/2432/files
[2] http://docs.ceph.com/docs/jewel/rados/operations/pools
2018-03-26 15:41:51 +02:00
Ning Yao 691ddf5349 cleanup osd.conf.j2 in ceph-osd
osd crush location is set by ceph_crush in the library,
osd.conf.j2 is not used any more.

Signed-off-by: Ning Yao <yaoning@unitedstack.com>
2018-03-26 15:57:37 +08:00
Patrick Donnelly 7f91547304 setup cephx keys when not nfs_obj_gw
Copy the admin key when configured nfs_file_gw (but not nfs_obj_gw). Also,
copy/setup RGW related directories only when configured as nfs_obj_gw.

Signed-off-by: Patrick Donnelly <pdonnell@redhat.com>
2018-03-22 14:01:08 +01:00
Andrew Schoen 6cffbd5409 ceph-defaults: set is_atomic variable
This variable is needed for containerized clusters and is required for
the ceph-docker-common role. Typically the is_atomic variable is set in
site-docker.yml.sample though so if ceph-docker-common is used outside
of that playbook it needs set in another way. Moving the creation of
the variable inside this role means playbooks don't need to worry
about setting it.

fixes: https://bugzilla.redhat.com/show_bug.cgi?id=1558252

Signed-off-by: Andrew Schoen <aschoen@redhat.com>
2018-03-21 19:16:11 +01:00
Andy McCrae fe4ba9d135 Fix config_template to consistently order sections
In ec042219e6 we added OrderedDict and
sorted to be able to preserve order for config_template k,v pairs inside
a section.

This patch adds a similar ordering for the sections themselves, which
could still change order and intiiate handler restarts.

OrderedDict isn't needed because we use .items() to return a list that
can then be sorted().
2018-03-16 23:24:28 +01:00
Andy McCrae 388562a4af Simplify ceph.conf generation
Since the approach to creating a ceph.conf file has changed, and now
no-longer relies on assembling config file fragments in /etc/ceph/ceph.d
we can avoid the conf_overrides rendering on the local host and skip out
the tasks related to that, instead using just the config_template task
to configure the file directly.
2018-03-15 15:47:41 +01:00
Sébastien Han e3275c1ca1 osd: add fs.aio-max-nr tuning
The number of osds per nodes is limited by aio-max-nr, default is low,
so we need to increase it.

Full story:
http://lists.ceph.com/pipermail/ceph-users-ceph.com/2017-August/020408.html

Fixes: https://bugzilla.redhat.com/show_bug.cgi?id=1553407
Signed-off-by: Sébastien Han <seb@redhat.com>
2018-03-15 14:06:26 +01:00
Sébastien Han f432819c1e osd: apply systcl right away
Without     sysctl_set: yes the sysctm tuning will only get applied on
the systctl.conf but not on the fly.

Signed-off-by: Sébastien Han <seb@redhat.com>
2018-03-15 14:06:26 +01:00
Sébastien Han 0f8a4251ba move system tuning to osd role
The changes from these tasks only apply to osd nodes so there is no
reason to have them in ceph-common.

Signed-off-by: Sébastien Han <seb@redhat.com>
2018-03-15 14:06:26 +01:00
Sébastien Han 3ab89ab48c ci: re-arrange group_vars files
We should stop putting everything in 'all'. This is too easy and this is
error prone as well for those who are separating variables into host
type, things that you should do.

Signed-off-by: Sébastien Han <seb@redhat.com>
2018-03-14 14:22:00 +01:00
Sébastien Han d5f8cac820 ci: remove left over iscsi_gws file
Wrong file that is not used, only iscsi-ggw that is present is correct.

Signed-off-by: Sébastien Han <seb@redhat.com>
2018-03-14 14:22:00 +01:00
Sébastien Han 8000ae342e remove unsed ceph_rgw_civetweb_port variable
Signed-off-by: Sébastien Han <seb@redhat.com>
2018-03-14 14:22:00 +01:00
Sébastien Han f119b25bbe client: implement proper pools creation
Just like we did for the monitor and openstack_config we now have the
ability to precisely create pools.

Signed-off-by: Sébastien Han <seb@redhat.com>
2018-03-14 14:22:00 +01:00
Sébastien Han e302c1baae mon: add support for erasure code pool
You can now specify type: erasure and   erasure_profile to use when
declaring the pool dictionnary.

Signed-off-by: Sébastien Han <seb@redhat.com>
2018-03-14 14:22:00 +01:00
Sébastien Han 277d885bc9 mon: add support for pgp, pool type and rule name
When creating pools, it's crucial to expose all the options available as
part of the pool creation command. As explained in:
http://docs.ceph.com/docs/jewel/rados/operations/pools/

Signed-off-by: Sébastien Han <seb@redhat.com>
2018-03-14 14:22:00 +01:00
Sébastien Han 4806ff4ff8 ci: test pool creation on container
On containerized scenario we also want to test pool creation.

Signed-off-by: Sébastien Han <seb@redhat.com>
2018-03-14 14:22:00 +01:00
Sébastien Han 26bc00fb74 mon: fail if pool creation fails
There is no reason to continue the deployment if these tasks fail.

Fixes: https://bugzilla.redhat.com/show_bug.cgi?id=1546185
Signed-off-by: Sébastien Han <seb@redhat.com>
2018-03-14 14:22:00 +01:00