]> git.apps.os.sepia.ceph.com Git - ceph-ansible.git/log
ceph-ansible.git
8 years agothe $? of pgrep ceph-rest-api command return is 1 when ceph-rest-api already run... 1381/head
yuliyang [Fri, 17 Mar 2017 09:40:45 +0000 (17:40 +0800)]
the $? of pgrep ceph-rest-api command return is 1 when ceph-rest-api already run.  use ps instead

8 years agoMerge pull request #1371 from ceph/yum-repo v2.2.0rc1
Sébastien Han [Thu, 16 Mar 2017 14:17:20 +0000 (15:17 +0100)]
Merge pull request #1371 from ceph/yum-repo

docker: change centos extra repo var name

8 years agodocker: change centos extra repo var name 1371/head
Sébastien Han [Thu, 16 Mar 2017 09:57:13 +0000 (10:57 +0100)]
docker: change centos extra repo var name

This is not only for monitors, but also mds, rgw and rbd mirror so
making the var name more generic:
ceph_docker_enable_centos_extra_repo

Signed-off-by: Sébastien Han <seb@redhat.com>
8 years agoMerge pull request #1331 from guits/enh_1321
Sébastien Han [Thu, 16 Mar 2017 12:28:45 +0000 (13:28 +0100)]
Merge pull request #1331 from guits/enh_1321

ceph-mon: create openstack pools and keys even for containerized deployments

8 years agoMerge pull request #1373 from ceph/docker-mon-wait
Sébastien Han [Thu, 16 Mar 2017 11:03:15 +0000 (12:03 +0100)]
Merge pull request #1373 from ceph/docker-mon-wait

mon: increase wait for monitor socket timeout

8 years agomon: increase wait for monitor socket timeout 1373/head
Sébastien Han [Thu, 16 Mar 2017 10:16:49 +0000 (11:16 +0100)]
mon: increase wait for monitor socket timeout

Sometimes the socket appears during the 5th attempt and sometimes not so
increasing the timeout a little bit.

Signed-off-by: Sébastien Han <seb@redhat.com>
8 years agoceph-mon: Create openstack pools and keys even for containerized deployments 1331/head
Guillaume Abrioux [Fri, 24 Feb 2017 09:22:16 +0000 (10:22 +0100)]
ceph-mon: Create openstack pools and keys even for containerized deployments

Add the possibility to create openstack pools and keys even for containerized deployments

Fix: #1321
Signed-off-by: Guillaume Abrioux <gabrioux@redhat.com>
8 years agoMerge pull request #1370 from ceph/template-rbd-mirror
Sébastien Han [Thu, 16 Mar 2017 09:19:26 +0000 (10:19 +0100)]
Merge pull request #1370 from ceph/template-rbd-mirror

rbd-mirror: fix spelling

8 years agorbd-mirror: fix spelling 1370/head
Sébastien Han [Thu, 16 Mar 2017 09:18:31 +0000 (10:18 +0100)]
rbd-mirror: fix spelling

It is RBD and NOT RDB.

Signed-off-by: Sébastien Han <seb@redhat.com>
8 years agoMerge pull request #1367 from D3N14L/fix_whitespaces_from_1336
Andrew Schoen [Wed, 15 Mar 2017 21:37:04 +0000 (16:37 -0500)]
Merge pull request #1367 from D3N14L/fix_whitespaces_from_1336

Fixing tabs that are breaking the syntax check

8 years agoFixing tabs that are breaking the syntax check 1367/head
Daniel Marks [Wed, 15 Mar 2017 13:15:15 +0000 (14:15 +0100)]
Fixing tabs that are breaking the syntax check

With the merge of PR #1336 the syntax check fails. This commit replaces
the tabs with proper indentation.

8 years agoMerge pull request #1336 from WingkaiHo/master
Sébastien Han [Wed, 15 Mar 2017 10:55:26 +0000 (11:55 +0100)]
Merge pull request #1336 from WingkaiHo/master

Load a variable file for devices partition

8 years agoMerge pull request #1363 from ceph/rgw-mds-container
Sébastien Han [Tue, 14 Mar 2017 21:50:05 +0000 (22:50 +0100)]
Merge pull request #1363 from ceph/rgw-mds-container

kv scenario: add KV_PORT option

8 years agokv scenario: add KV_PORT option 1363/head
Sébastien Han [Tue, 14 Mar 2017 21:46:09 +0000 (22:46 +0100)]
kv scenario: add KV_PORT option

This option was missing for rrgw, mds, rbd mirror and nfs making these
daemon impossible to run on a kv deployment with containers.

Signed-off-by: Sébastien Han <seb@redhat.com>
8 years agoMerge pull request #1317 from ibotty/harmonize-docker-names
Sébastien Han [Tue, 14 Mar 2017 17:20:20 +0000 (18:20 +0100)]
Merge pull request #1317 from ibotty/harmonize-docker-names

harmonize docker names

8 years agoMerge pull request #1358 from ceph/nfs-ganesha-fsal-deb
Sébastien Han [Tue, 14 Mar 2017 15:03:14 +0000 (16:03 +0100)]
Merge pull request #1358 from ceph/nfs-ganesha-fsal-deb

ceph-common: install nfs-ganesha FSALs on Debian

8 years agoceph-common: install nfs-ganesha FSALs on Debian 1358/head
Ken Dreyer [Mon, 13 Mar 2017 15:34:35 +0000 (09:34 -0600)]
ceph-common: install nfs-ganesha FSALs on Debian

Prior to this change, ceph-ansible would install the main NFS Ganesha
server daemon on Ubuntu, but it would skip the Ceph FSALs.

Running "apt-get install nfs-ganesha" will only install the main NFS Ganesha
server. It does *not* pull in the RGW FSAL
(/usr/lib/x86_64-linux-gnu/ganesha/libfsalrgw.so)

Running "apt-get install nfs-ganesha-fsal" will install the RGW FSAL as
well as the main NFS Ganesha server package.

Signed-off-by: Ken Dreyer <kdreyer@redhat.com>
8 years agoMerge pull request #1356 from ceph/wip-calamari-debug
Sébastien Han [Tue, 14 Mar 2017 09:55:52 +0000 (10:55 +0100)]
Merge pull request #1356 from ceph/wip-calamari-debug

calamari: Add debugging option

8 years agocalamari: Add debugging option 1356/head
Boris Ranto [Mon, 13 Mar 2017 12:46:38 +0000 (13:46 +0100)]
calamari: Add debugging option

This patch introduces calamari_debug option which will turn on debugging
for calamari before initializing and running it.

Signed-off-by: Boris Ranto <branto@redhat.com>
8 years agoMerge pull request #1355 from ceph/rolling-update-tests
Alfredo Deza [Thu, 9 Mar 2017 19:08:59 +0000 (14:08 -0500)]
Merge pull request #1355 from ceph/rolling-update-tests

tests: adds an update_cluster testing scenario

8 years agotests: adds an update_cluster testing scenario 1355/head
Andrew Schoen [Tue, 7 Mar 2017 20:05:03 +0000 (14:05 -0600)]
tests: adds an update_cluster testing scenario

This updates the centos 7 cluster scenario from jewel to kraken using
the rolling_update playbook.

Signed-off-by: Andrew Schoen <aschoen@redhat.com>
8 years agoMerge pull request #1349 from ceph/rhbz-1354700
Sébastien Han [Wed, 8 Mar 2017 17:02:30 +0000 (18:02 +0100)]
Merge pull request #1349 from ceph/rhbz-1354700

purge-cluster: do not set group name vars at playbook level

8 years agoMerge pull request #1347 from ceph/no-vfs-cache-pressure
Sébastien Han [Wed, 8 Mar 2017 17:02:06 +0000 (18:02 +0100)]
Merge pull request #1347 from ceph/no-vfs-cache-pressure

avoid setting vfs_cache_pressure

8 years agoswitch-to-containers: do not include group vars or role defaults 1349/head
Andrew Schoen [Tue, 7 Mar 2017 19:51:48 +0000 (13:51 -0600)]
switch-to-containers: do not include group vars or role defaults

Doing so will override any values set for these in the group_vars
directory relative to the users inventory.

Signed-off-by: Andrew Schoen <aschoen@redhat.com>
8 years agopurge-docker-cluster: do not include role defaults or group vars
Andrew Schoen [Tue, 7 Mar 2017 19:45:22 +0000 (13:45 -0600)]
purge-docker-cluster: do not include role defaults or group vars

Doing so at playbook level overrides whatever values might be set for
these in the user's group_vars directory that's relative to their
inventory.

Signed-off-by: Andrew Schoen <aschoen@redhat.com>
8 years agoswitch-to-containers: do not set group name vars at playbook level
Andrew Schoen [Tue, 7 Mar 2017 19:32:57 +0000 (13:32 -0600)]
switch-to-containers: do not set group name vars at playbook level

Signed-off-by: Andrew Schoen <aschoen@redhat.com>
8 years agopurge-docker-cluster: do not set group names vars at playbook level
Andrew Schoen [Tue, 7 Mar 2017 19:23:40 +0000 (13:23 -0600)]
purge-docker-cluster: do not set group names vars at playbook level

Signed-off-by: Andrew Schoen <aschoen@redhat.com>
8 years agorolling-update: do not set group name vars at playbook level
Andrew Schoen [Tue, 7 Mar 2017 19:16:12 +0000 (13:16 -0600)]
rolling-update: do not set group name vars at playbook level

Signed-off-by: Andrew Schoen <aschoen@redhat.com>
8 years agopurge-cluster: do not set group name vars at playbook level
Andrew Schoen [Tue, 7 Mar 2017 16:16:09 +0000 (10:16 -0600)]
purge-cluster: do not set group name vars at playbook level

This has the behavior of overriding custom values set in group_vars.
I've added defaults to the rest of the group names so that if they are
not overridden in group_vars then defaults will be used.

See: https://bugzilla.redhat.com/show_bug.cgi?id=1354700

Signed-off-by: Andrew Schoen <aschoen@redhat.com>
8 years agoUpdate make-osd-partitions.yml 1336/head
WingKai Ho [Wed, 8 Mar 2017 09:46:37 +0000 (17:46 +0800)]
Update make-osd-partitions.yml

change

8 years agoMerge pull request #1350 from ceph/make-handle-rc
Alfredo Deza [Tue, 7 Mar 2017 21:12:10 +0000 (16:12 -0500)]
Merge pull request #1350 from ceph/make-handle-rc

Makefile: handle "rc" version tags

8 years agoMakefile: handle "rc" version tags 1350/head
Ken Dreyer [Tue, 7 Mar 2017 18:01:15 +0000 (11:01 -0700)]
Makefile: handle "rc" version tags

When "git describe" returns an "rc" tag (eg "2.2.0rc1"), insert the RC
number into the RPM's "Release" field.

Add comments with some git-describe values and NVRs.

8 years agoMakefile: generate tagged RPMs with "-1" release value
Ken Dreyer [Tue, 7 Mar 2017 18:11:53 +0000 (11:11 -0700)]
Makefile: generate tagged RPMs with "-1" release value

Prior to this change, if we built an RPM from a tagged commit, the RPM
would have a release value of "0".

This causes some problems with RPM version comparisons when comparing
version numbers with "rc" tags/packages, which should also have a
less-than-one Release value.

8 years agoMerge pull request #1335 from batrick/mon-fs-cmd-1mon
Sébastien Han [Tue, 7 Mar 2017 15:45:16 +0000 (16:45 +0100)]
Merge pull request #1335 from batrick/mon-fs-cmd-1mon

send fs commands to one mon

8 years agoMerge pull request #1339 from ceph/rgw-remove-sudo
Sébastien Han [Tue, 7 Mar 2017 15:44:28 +0000 (16:44 +0100)]
Merge pull request #1339 from ceph/rgw-remove-sudo

rgw: remove sudoers file

8 years agoMerge pull request #1346 from ceph/fix-rhcs-rhel-supported-message
Sébastien Han [Mon, 6 Mar 2017 14:50:24 +0000 (15:50 +0100)]
Merge pull request #1346 from ceph/fix-rhcs-rhel-supported-message

check_system: improve RHCS support message and version comparison

8 years agoUpdate make-osd-partitions.yml
WingKai Ho [Mon, 6 Mar 2017 07:43:09 +0000 (15:43 +0800)]
Update make-osd-partitions.yml

When ansible do not load the file host_vars/{{ ansible_hostname }}.yml and host_vars/default.yml it will show syntactic, so keyword "skip" to fix it.
Exit the playbook if the user not define devices  in both  host_vars/{{ ansible_hostname }}.yml and host_vars/default.yml

8 years agoUpdate make-osd-partitions.yml
WingKai Ho [Mon, 6 Mar 2017 02:33:22 +0000 (10:33 +0800)]
Update make-osd-partitions.yml

When ansible do not load the file host_vars/{{ ansible_hostname }}.yml and host_vars/default.yml it will show syntactic err, so add keyword "skip" to fix it.

Exit the playbook if the user not define devices  in both  host_vars/{{ ansible_hostname }}.yml and host_vars/default.yml
host_vars/default.yml

8 years agoavoid setting vfs_cache_pressure 1347/head
Ken Dreyer [Fri, 3 Mar 2017 18:20:05 +0000 (11:20 -0700)]
avoid setting vfs_cache_pressure

From Josh Durgin, "I'd recommend not setting vfs_cache_pressure in
ceph-ansible. The syncfs issue is still there, and has caused real
problems in the past, whereas there hasn't been good data showing lower
vfs_cache_pressure is very helpful - the only cases I'm aware of have
shown it makes little difference to performance."

https://bugzilla.redhat.com/show_bug.cgi?id=1395451

8 years agocheck_system: improve RHCS support message and version comparison 1346/head
Ken Dreyer [Fri, 3 Mar 2017 16:56:30 +0000 (09:56 -0700)]
check_system: improve RHCS support message and version comparison

"red hat storage" -> "red hat ceph storage"

"RHEL 7.1" -> "RHEL 7"

and make the version number check verify the very latest version

https://bugzilla.redhat.com/show_bug.cgi?id=1354059

8 years agoMerge pull request #1328 from guits/fix_1303
Sébastien Han [Fri, 3 Mar 2017 15:33:48 +0000 (16:33 +0100)]
Merge pull request #1328 from guits/fix_1303

docker: Fix #1303

8 years agodocker: Fix #1303 1328/head
Guillaume Abrioux [Tue, 21 Feb 2017 16:58:24 +0000 (17:58 +0100)]
docker: Fix #1303

Install package from official repos rather than pip when using RHEL.

This commit fix https://bugzilla.redhat.com/show_bug.cgi?id=1420855

Also this commit Refact all `roles/ceph-*/tasks/docker/pre_requisite.yml`
to avoid a lot of duplicated code.

Fix: #1303
Signed-off-by: Guillaume Abrioux <gabrioux@redhat.com>
8 years agoUpdate make-osd-partitions.yml
WingKai Ho [Fri, 3 Mar 2017 09:26:53 +0000 (17:26 +0800)]
Update make-osd-partitions.yml

fix syntactic error

8 years agorgw: remove sudoers file 1339/head
Sébastien Han [Thu, 2 Mar 2017 13:59:21 +0000 (14:59 +0100)]
rgw: remove sudoers file

This was needed for Hammer and older version, not needed anymore since
we have a 'ceph' user to run ceph processes.

Signed-off-by: Sébastien Han <seb@redhat.com>
8 years agoUpdate ceph_keys.yml
WingKai Ho [Thu, 2 Mar 2017 06:09:08 +0000 (14:09 +0800)]
Update ceph_keys.yml

8 years agoUpdate ceph_keys.yml
WingKai Ho [Thu, 2 Mar 2017 06:07:52 +0000 (14:07 +0800)]
Update ceph_keys.yml

8 years agoUpdate make-osd-partitions.yml
WingKai Ho [Thu, 2 Mar 2017 06:06:41 +0000 (14:06 +0800)]
Update make-osd-partitions.yml

Remove capital `L`

8 years agoUpdate make-osd-partitions.yml
WingKai Ho [Thu, 2 Mar 2017 04:24:44 +0000 (12:24 +0800)]
Update make-osd-partitions.yml

there is an extra space between 'custom' and 'layout'

8 years agosend fs commands to one mon 1335/head
Patrick Donnelly [Tue, 28 Feb 2017 17:48:08 +0000 (12:48 -0500)]
send fs commands to one mon

Add-on to #1329.

Signed-off-by: Patrick Donnelly <pdonnell@redhat.com>
8 years agoUpdate ceph_keys.yml
WingKai Ho [Wed, 1 Mar 2017 09:39:53 +0000 (17:39 +0800)]
Update ceph_keys.yml

8 years agoLoad a variable file for devices parrition
WingKai Ho [Wed, 1 Mar 2017 09:27:57 +0000 (17:27 +0800)]
Load a variable file for devices parrition

load device partition file in directory host_vars

1) if the user define host_vars/hostname.yml load the devices  partition on this file.
2) otherwise load host_vars/default.yml for default

8 years agoMerge pull request #1334 from yangyimincn/yangyimincn-patch-1
Sébastien Han [Tue, 28 Feb 2017 09:48:26 +0000 (10:48 +0100)]
Merge pull request #1334 from yangyimincn/yangyimincn-patch-1

rolling_update check monitor quorum error

8 years agoUpdate rolling_update.yml 1334/head
yangyimincn [Tue, 28 Feb 2017 08:56:02 +0000 (16:56 +0800)]
Update rolling_update.yml

The task waiting for the monitor to join the quorum... , the result for ceph -s | grep monmap only contain monmap, not included quorum:

# ceph -s --cluster ceph | grep monmap
     monmap e1: 3 mons at {sh-office-ceph-1=10.12.10.34:6789/0,sh-office-ceph-2=10.12.10.35:6789/0,sh-office-ceph-3=10.12.10.36:6789/0}

If want to get monitor, should use this:

# ceph -s --cluster ceph | grep election
            election epoch 80, quorum 0,1 sh-office-ceph-1,sh-office-ceph-2

ceph verison: 10.2.5

8 years agoMerge pull request #1329 from guits/fix_1314
Sébastien Han [Mon, 27 Feb 2017 10:57:53 +0000 (11:57 +0100)]
Merge pull request #1329 from guits/fix_1314

ceph-mon: Check if ceph filesystem exists

8 years agoceph-mon: Check if ceph filesystem exists 1329/head
Guillaume Abrioux [Thu, 23 Feb 2017 15:06:04 +0000 (16:06 +0100)]
ceph-mon: Check if ceph filesystem exists

Check if ceph filesystem already exists before creating it.
If the ceph filesystem doesn't exist, execute the task only on one node.

Fix: #1314
Signed-off-by: Guillaume Abrioux <gabrioux@redhat.com>
8 years agoMerge pull request #1330 from ceph/fix-cluster-name-take-over
Sébastien Han [Thu, 23 Feb 2017 21:23:18 +0000 (16:23 -0500)]
Merge pull request #1330 from ceph/fix-cluster-name-take-over

infra: fix cluster name detection

8 years agoinfra: fix cluster name detection 1330/head
Sébastien Han [Thu, 23 Feb 2017 20:39:55 +0000 (15:39 -0500)]
infra: fix cluster name detection

The previous command was returning /etc/ceph/ceph.conf, we only need
'ceph' to be returned.

Signed-off-by: Sébastien Han <seb@redhat.com>
8 years agoharmonize docker names 1317/head
Tobias Florek [Mon, 20 Feb 2017 09:03:49 +0000 (10:03 +0100)]
harmonize docker names

Created containers now are named more or less in the form of

    <ansible role>-<ansible_hostname>

8 years agoMerge pull request #1287 from ceph/no-priviledge-mode
Andrew Schoen [Wed, 22 Feb 2017 21:24:18 +0000 (15:24 -0600)]
Merge pull request #1287 from ceph/no-priviledge-mode

No priviledge mode

8 years agotests: when using pytest mark decorators ensure all fixtures are defined 1287/head
Andrew Schoen [Wed, 22 Feb 2017 19:43:59 +0000 (13:43 -0600)]
tests: when using pytest mark decorators ensure all fixtures are defined

Decorating a test method directly with a pytest mark seems to break if
the test function does not explicitly define all pytest fixtures it
expects to recieve.

Signed-off-by: Andrew Schoen <aschoen@redhat.com>
8 years agoci: decorate the tests to not run on docker scenario
Sébastien Han [Wed, 22 Feb 2017 14:21:23 +0000 (09:21 -0500)]
ci: decorate the tests to not run on docker scenario

Certain scenario won't work on containerized deployment. So we decorate
them so they can be skipped.

Signed-off-by: Sébastien Han <seb@redhat.com>
8 years agoci: fix issue on ansible2.2-docker_dedicated_journal
Sébastien Han [Tue, 21 Feb 2017 21:14:39 +0000 (16:14 -0500)]
ci: fix issue on ansible2.2-docker_dedicated_journal

journal_collocation was enabled so the test suite was testing this
scenario and obviously failed since there is no second partition to
verify.

Signed-off-by: Sébastien Han <seb@redhat.com>
8 years agoci: do not use atomic host for ansible2.2-docker_dedicated_journal
Sébastien Han [Tue, 21 Feb 2017 20:39:42 +0000 (15:39 -0500)]
ci: do not use atomic host for ansible2.2-docker_dedicated_journal

Switch to CentOS since Atomic host does not have the right Docker
version.

Signed-off-by: Sébastien Han <seb@redhat.com>
8 years agoci: set a different directory for ceph osd docker run script
Sébastien Han [Tue, 21 Feb 2017 19:04:58 +0000 (14:04 -0500)]
ci: set a different directory for ceph osd docker run script

/usr/share is not writable on Atomic Host so we use /var/tmp instead.

Signed-off-by: Sébastien Han <seb@redhat.com>
8 years agoci: do not generate random hostname for ansible2.2-docker_dedicated_journal
Sébastien Han [Tue, 21 Feb 2017 18:39:47 +0000 (13:39 -0500)]
ci: do not generate random hostname for ansible2.2-docker_dedicated_journal

This fixes the error: Call to virDomainCreateWithFlags failed: internal
error: Monitor path
/var/lib/libvirt/qemu/domain-docker-cluster-dedicated-journal_osd0_1487692576_dbfc21d851071d3e2cd2/monitor.sock
too big for destination

Signed-off-by: Sébastien Han <seb@redhat.com>
8 years agomon: docker, ability to enable centos extra repo
Sébastien Han [Tue, 21 Feb 2017 15:33:40 +0000 (10:33 -0500)]
mon: docker, ability to enable centos extra repo

Signed-off-by: Sébastien Han <seb@redhat.com>
8 years agodocker: osd, clarify variable usage for scenarii
Sébastien Han [Sat, 18 Feb 2017 13:48:18 +0000 (14:48 +0100)]
docker: osd, clarify variable usage for scenarii

Signed-off-by: Sébastien Han <seb@redhat.com>
8 years agopurge-docker: re-implement zap devices
Sébastien Han [Sat, 18 Feb 2017 01:54:47 +0000 (02:54 +0100)]
purge-docker: re-implement zap devices

We now run the container and waits until it dies. Prior to this we were
stopping it before completion so not all the devices where zapped.

Signed-off-by: Sébastien Han <seb@redhat.com>
8 years agodocker: make ceph docker osd script path
Sébastien Han [Thu, 16 Feb 2017 21:13:26 +0000 (22:13 +0100)]
docker: make ceph docker osd script path

Since distro will not allow /usr/share to be writable (e.g: atomic) so
we let the operator decide where to put that script.

Signed-off-by: Sébastien Han <seb@redhat.com>
8 years agoci: add docker-cluster-dmcrypt-journal-collocation scenario
Sébastien Han [Thu, 16 Feb 2017 20:25:29 +0000 (21:25 +0100)]
ci: add docker-cluster-dmcrypt-journal-collocation scenario

Signed-off-by: Sébastien Han <seb@redhat.com>
8 years agotests: enable the docker_dedicated_journal scenario
Andrew Schoen [Fri, 10 Feb 2017 16:37:52 +0000 (10:37 -0600)]
tests: enable the docker_dedicated_journal scenario

Signed-off-by: Andrew Schoen <aschoen@redhat.com>
8 years agotests: add scenario for dedicated-journal on docker
Sébastien Han [Tue, 7 Feb 2017 21:08:47 +0000 (22:08 +0100)]
tests: add scenario for dedicated-journal on docker

Signed-off-by: Sébastien Han <seb@redhat.com>
8 years agodocker: osd, do not use priviledged container anymore
Sébastien Han [Tue, 7 Feb 2017 21:00:53 +0000 (22:00 +0100)]
docker: osd, do not use priviledged container anymore

Oh yeah! This patch adds more fine grained control on how we run the
activation osd container. We now use --device to give a read, write and
mknodaccess to a specific device to be consumed by Ceph. We also use
SYS_ADMIN cap to allow mount operations, ceph-disk needs to temporary
mount the osd data directory during the activation sequence.

This patch also enables the support of dedicated journal devices when
deploying ceph-docker with ceph-ansible.

Depends on https://github.com/ceph/ceph-docker/pull/478

Signed-off-by: Sébastien Han <seb@redhat.com>
8 years agopurge-docker: also purge journal devices
Sébastien Han [Tue, 7 Feb 2017 20:55:36 +0000 (21:55 +0100)]
purge-docker: also purge journal devices

Signed-off-by: Sébastien Han <seb@redhat.com>
8 years agodocker: osd, do not skip on failure
Sébastien Han [Tue, 7 Feb 2017 12:45:31 +0000 (13:45 +0100)]
docker: osd, do not skip on failure

If the systemd unit file can not be generated we should fail, same for
systemd enable and reload.

Signed-off-by: Sébastien Han <seb@redhat.com>
8 years agoMerge pull request #1315 from ceph/rolling-update-tests
Sébastien Han [Tue, 21 Feb 2017 20:53:57 +0000 (15:53 -0500)]
Merge pull request #1315 from ceph/rolling-update-tests

adds a testing scenario for rolling updates

8 years agoupdate group_vars sample for rolling_update variable 1315/head
Andrew Schoen [Tue, 21 Feb 2017 18:36:54 +0000 (12:36 -0600)]
update group_vars sample for rolling_update variable

Signed-off-by: Andrew Schoen <aschoen@redhat.com>
8 years agoceph-common: remove infernalis comment on radosgw_civetweb_port
Andrew Schoen [Tue, 21 Feb 2017 18:35:00 +0000 (12:35 -0600)]
ceph-common: remove infernalis comment on radosgw_civetweb_port

As of Infernalis, the Ceph daemons run as an unprivileged "ceph" UID,
and this is by design.

Commit f19b765 altered the default
civetweb port from 80 to 8080 with a comment in the commit log about
"until this gets solved"

Remove the comment about permissions on Infernalis, because this is
always going to be the case on the Ceph versions we support, and it
is just confusing.

If users want to expose civetweb to s3 clients using privileged TCP
ports, they can redirect traffic with iptables, or use a reverse proxy
application like HAproxy.

Signed-off-by: Andrew Schoen <aschoen@redhat.com>
8 years agotests: adds a new ansible2.2-update_dmcrypt scenario
Andrew Schoen [Fri, 17 Feb 2017 20:33:30 +0000 (14:33 -0600)]
tests: adds a new ansible2.2-update_dmcrypt scenario

This performs a rolling update on a cluster using dmcrypt dedicated
journals.

Signed-off-by: Andrew Schoen <aschoen@redhat.com>
8 years agorolling-update: do not use upstart to stop mons when using systemd
Andrew Schoen [Fri, 17 Feb 2017 20:31:25 +0000 (14:31 -0600)]
rolling-update: do not use upstart to stop mons when using systemd

Signed-off-by: Andrew Schoen <aschoen@redhat.com>
8 years agoceph-common: do not get current fsid when performing a rolling_update
Andrew Schoen [Fri, 17 Feb 2017 20:29:45 +0000 (14:29 -0600)]
ceph-common: do not get current fsid when performing a rolling_update

This avoids a situation where during a rolling_update we try to talk to
a mon to get the fsid and if that mon is down the playbook hangs
indefinitely.

Signed-off-by: Andrew Schoen <aschoen@redhat.com>
8 years agoceph-common: use yum_repository when adding the ceph_stable repo
Andrew Schoen [Fri, 17 Feb 2017 20:27:15 +0000 (14:27 -0600)]
ceph-common: use yum_repository when adding the ceph_stable repo

This gives us more flexibility than installing the ceph-release package
as we can easily use different mirrors. Also, I noticed an issue when
upgrading from jewel -> kraken as the ceph-release package for those
releases both have the same version number and yum doesn't know to
update anything.

Signed-off-by: Andrew Schoen <aschoen@redhat.com>
8 years agoMerge pull request #1324 from ktdreyer/rm-civetweb-port-comment
Sébastien Han [Tue, 21 Feb 2017 18:28:16 +0000 (13:28 -0500)]
Merge pull request #1324 from ktdreyer/rm-civetweb-port-comment

group_vars: rm Infernalis comment

8 years agogroup_vars: rm Infernalis comment 1324/head
Ken Dreyer [Tue, 21 Feb 2017 16:49:40 +0000 (09:49 -0700)]
group_vars: rm Infernalis comment

As of Infernalis, the Ceph daemons run as an unprivileged "ceph" UID,
and this is by design.

Commit f19b765f7995dab4aca46e1885320dc440485ead altered the default
civetweb port from 80 to 8080 with a comment in the commit log about
"until this gets solved"

Remove the comment about permissions on Infernalis, because this is
always going to be the case on the Ceph versions we support, and it
is just confusing.

If users want to expose civetweb to s3 clients using privileged TCP
ports, they can redirect traffic with iptables, or use a reverse proxy
application like HAproxy.

8 years agoMerge pull request #1325 from ceph/ipv6-rgw-doc
Sébastien Han [Tue, 21 Feb 2017 17:02:06 +0000 (12:02 -0500)]
Merge pull request #1325 from ceph/ipv6-rgw-doc

Ipv6 rgw doc

8 years agodocker-common: sync group_vars file 1325/head
Sébastien Han [Tue, 21 Feb 2017 17:01:08 +0000 (12:01 -0500)]
docker-common: sync group_vars file

Signed-off-by: Sébastien Han <seb@redhat.com>
8 years agocommon: add doc for rgw on ipv6
Sébastien Han [Tue, 21 Feb 2017 17:00:37 +0000 (12:00 -0500)]
common: add doc for rgw on ipv6

See: https://bugzilla.redhat.com/show_bug.cgi?id=1424799

Signed-off-by: Sébastien Han <seb@redhat.com>
8 years agoMerge pull request #1322 from ceph/fix-transparent-page
Sébastien Han [Tue, 21 Feb 2017 14:34:30 +0000 (09:34 -0500)]
Merge pull request #1322 from ceph/fix-transparent-page

common: fix "disable transparent hugepage"

8 years agoUpdate ceph_keys.yml 1323/head
WingKai Ho [Tue, 21 Feb 2017 01:49:52 +0000 (09:49 +0800)]
Update ceph_keys.yml

jewel version need to build the {{ cluster }}.client.admin.keyring exists key

8 years agocommon: fix "disable transparent hugepage" 1322/head
Sébastien Han [Mon, 20 Feb 2017 22:07:53 +0000 (17:07 -0500)]
common: fix "disable transparent hugepage"

To configure kernel the task is using "command" module which is not
respect operator ">". So this task just print to "stdout": "never >
/sys/kernel/mm/transparent_hugepage/enabled"

fix: #1319

Signed-off-by: Sébastien Han <seb@redhat.com>
8 years agoMerge pull request #1316 from zhsj/fix-osd-id
Sébastien Han [Mon, 20 Feb 2017 21:57:14 +0000 (16:57 -0500)]
Merge pull request #1316 from zhsj/fix-osd-id

fix grep match pattern for osd ids

8 years agofix grep match pattern for osd ids 1316/head
Shengjing Zhu [Mon, 20 Feb 2017 08:35:56 +0000 (16:35 +0800)]
fix grep match pattern for osd ids

Some playbooks use [0-9]*, others use \d+$
The latter is more correct since cluster name may contain numbers.

Signed-off-by: Shengjing Zhu <zsj950618@gmail.com>
8 years agoMerge pull request #1313 from ceph/docker-systemd-units
Sébastien Han [Thu, 16 Feb 2017 19:02:13 +0000 (20:02 +0100)]
Merge pull request #1313 from ceph/docker-systemd-units

docker: homogenise systemd file location

8 years agodocker: homogenise systemd file location 1313/head
Sébastien Han [Thu, 16 Feb 2017 17:31:32 +0000 (18:31 +0100)]
docker: homogenise systemd file location

So unit files were stored in /var/lib/ceph some where in
/etc/systemd/system. Now they are all under /etc/systemd/system.

closes: #1296

Signed-off-by: Sébastien Han <seb@redhat.com>
8 years agoMerge pull request #1311 from ceph/BZ-1414647
Andrew Schoen [Wed, 15 Feb 2017 17:57:09 +0000 (11:57 -0600)]
Merge pull request #1311 from ceph/BZ-1414647

purge-cluster: also purge dmcrypt dedicated journals

8 years agopurge-cluster: also purge dmcrypt dedicated journals 1311/head
Andrew Schoen [Wed, 15 Feb 2017 15:53:13 +0000 (09:53 -0600)]
purge-cluster: also purge dmcrypt dedicated journals

See: https://bugzilla.redhat.com/show_bug.cgi?id=1414647

Signed-off-by: Andrew Schoen <aschoen@redhat.com>
8 years agoMerge pull request #1307 from ceph/rgw-configure
Sébastien Han [Wed, 15 Feb 2017 09:52:38 +0000 (10:52 +0100)]
Merge pull request #1307 from ceph/rgw-configure

rgw-standalone.yml will also now collect keys

8 years agorgw-standalone: also fetch keys from mons 1307/head
Andrew Schoen [Tue, 14 Feb 2017 22:11:30 +0000 (16:11 -0600)]
rgw-standalone: also fetch keys from mons

This is to allow for ceph-installer usage of this playbook and
to ensure that you have the correct keys locally when bootstrapping.

Signed-off-by: Andrew Schoen <aschoen@redhat.com>
8 years agoMerge pull request #1306 from ceph/BZ-1422134
Andrew Schoen [Tue, 14 Feb 2017 18:29:54 +0000 (12:29 -0600)]
Merge pull request #1306 from ceph/BZ-1422134

purge-cluster: remove calamari-server package

8 years agopurge-cluster: remove calamari-server package 1306/head
Andrew Schoen [Tue, 14 Feb 2017 15:24:02 +0000 (09:24 -0600)]
purge-cluster: remove calamari-server package

See: https://bugzilla.redhat.com/show_bug.cgi?id=1422134

Signed-off-by: Andrew Schoen <aschoen@redhat.com>
Resolves rhbz#1422134