]> git-server-git.apps.pok.os.sepia.ceph.com Git - ceph-ansible.git/log
ceph-ansible.git
9 years agoceph-common: install ceph-common packages for clients on redhat for rhcs
Alfredo Deza [Tue, 31 May 2016 16:09:31 +0000 (12:09 -0400)]
ceph-common: install ceph-common packages for clients on redhat for rhcs

Signed-off-by: Alfredo Deza <adeza@redhat.com>
Resolves: rhbz#1339096

9 years agoceph-common: define client_group_name
Alfredo Deza [Tue, 31 May 2016 16:20:03 +0000 (12:20 -0400)]
ceph-common: define client_group_name

Signed-off-by: Alfredo Deza <adeza@redhat.com>
Resolves: rhbz#1339096

9 years agogroup_vars: define client_group_name
Alfredo Deza [Tue, 31 May 2016 16:18:50 +0000 (12:18 -0400)]
group_vars: define client_group_name

Signed-off-by: Alfredo Deza <adeza@redhat.com>
Resolves: rhbz#1339096

9 years agoceph-common: create a new install task for rh storage on redhat
Alfredo Deza [Tue, 24 May 2016 14:07:34 +0000 (10:07 -0400)]
ceph-common: create a new install task for rh storage on redhat

Signed-off-by: Alfredo Deza <adeza@redhat.com>
Resolves: rhbz#1339096

9 years agoceph-common: include the new install_rh_storage_on_redhat playbook when ceph_stable_r...
Alfredo Deza [Tue, 24 May 2016 14:10:37 +0000 (10:10 -0400)]
ceph-common: include the new install_rh_storage_on_redhat playbook when ceph_stable_rh_storage is on:

Signed-off-by: Alfredo Deza <adeza@redhat.com>
Resolves: rhbz#1339096

9 years agoceph-common: specify file path for repo key
Alfredo Deza [Tue, 24 May 2016 12:28:34 +0000 (08:28 -0400)]
ceph-common: specify file path for repo key

Signed-off-by: Alfredo Deza <adeza@redhat.com>
Resolves: rhbz#1339068

9 years agoceph-common: there is no Installer repo, it is now the Tools repo
Alfredo Deza [Mon, 23 May 2016 13:09:00 +0000 (09:09 -0400)]
ceph-common: there is no Installer repo, it is now the Tools repo

Signed-off-by: Alfredo Deza <adeza@redhat.com>
Resolves: rhbz#1338548

9 years agoceph-common: remove RedHat-only constraint for RH ISO install
Alfredo Deza [Wed, 18 May 2016 20:40:28 +0000 (16:40 -0400)]
ceph-common: remove RedHat-only constraint for RH ISO install

Signed-off-by: Alfredo Deza <adeza@redhat.com>
9 years agoceph-osd: fix typo when calling sgdisk second time around
Alfredo Deza [Thu, 19 May 2016 18:46:43 +0000 (14:46 -0400)]
ceph-osd: fix typo when calling sgdisk second time around

Signed-off-by: Alfredo Deza <adeza@redhat.com>
9 years agoinstall ceph-mds on debian systems when using red hat ceph storage
Andrew Schoen [Thu, 19 May 2016 14:56:06 +0000 (09:56 -0500)]
install ceph-mds on debian systems when using red hat ceph storage

Signed-off-by: Andrew Schoen <aschoen@redhat.com>
9 years agoinstall radosgw on debian systems when using red hat ceph storage
Andrew Schoen [Thu, 19 May 2016 14:52:18 +0000 (09:52 -0500)]
install radosgw on debian systems when using red hat ceph storage

Signed-off-by: Andrew Schoen <aschoen@redhat.com>
9 years agoceph-common: on rh storage on debian add the repositories and keys
Alfredo Deza [Thu, 19 May 2016 18:20:47 +0000 (14:20 -0400)]
ceph-common: on rh storage on debian add the repositories and keys

Signed-off-by: Alfredo Deza <adeza@redhat.com>
9 years agoceph-common: on rh storage on debian do not add the rh key by default
Alfredo Deza [Thu, 19 May 2016 19:16:18 +0000 (15:16 -0400)]
ceph-common: on rh storage on debian do not add the rh key by default

Signed-off-by: Alfredo Deza <adeza@redhat.com>
9 years agoceph-common: redhat repo template: remove the Calamari reference, no longer a distinc...
Alfredo Deza [Thu, 19 May 2016 20:17:02 +0000 (16:17 -0400)]
ceph-common: redhat repo template: remove the Calamari reference, no longer a distinct repo

Signed-off-by: Alfredo Deza <adeza@redhat.com>
9 years agoceph-osd: fail when ceph-disk fails to prepare an OSD
Alfredo Deza [Wed, 18 May 2016 12:09:26 +0000 (08:09 -0400)]
ceph-osd: fail when ceph-disk fails to prepare an OSD

Signed-off-by: Alfredo Deza <adeza@redhat.com>
9 years agoceph-common: install ceph-mds package
Sébastien Han [Sat, 7 May 2016 00:37:09 +0000 (02:37 +0200)]
ceph-common: install ceph-mds package

We kinda ommitted this package, now Jewel is out and the metadata server
is considered as stable.

Signed-off-by: Sébastien Han <seb@redhat.com>
(cherry picked from commit c4f3ab5649c0a79e708ea94a5556fe53acf99f20)

9 years agofix the /etc/default/ceph path on debian when setting cluster name
Andrew Schoen [Fri, 13 May 2016 20:01:00 +0000 (15:01 -0500)]
fix the /etc/default/ceph path on debian when setting cluster name

In ceph version 10.2.1 a bug was fixed that corrects the path:

https://github.com/ceph/ceph/commit/791eba81a5467dd5de4f1680ed0deb647eb3fb8b

Signed-off-by: Andrew Schoen <aschoen@redhat.com>
(cherry picked from commit 85fb03fc99e0b7cd14b27e369c8e952b79313d73)

9 years agoDo not create symlinks for systemd manually
Andrew Schoen [Fri, 13 May 2016 18:19:29 +0000 (13:19 -0500)]
Do not create symlinks for systemd manually

Instead use 'systemctl enable' to create the symlinks.

This fixes: https://github.com/ceph/ceph-ansible/issues/779

Signed-off-by: Andrew Schoen <aschoen@redhat.com>
(cherry picked from commit d2b94c2a4cda645a33f45f3ed042212a0f0a67b6)

9 years agoceph-common: purge ceph.conf file
Sébastien Han [Thu, 7 Apr 2016 08:46:42 +0000 (10:46 +0200)]
ceph-common: purge ceph.conf file

Since ##461 we have been having the ability to override ceph default
options. Previously we had to add a new line in the template and then
another variable as well. Doing a PR for one option was such a pain. As
a result, we now have tons of options that we need to maintain across
all the ceph version, yet another painful thing to do.
This commit removes all the ceph options so they are handled by ceph
directly. If you want to add a new option, feel free to to use the
`ceph_conf_overrides` variable of your `group_vars/all`.

Risks, for those who have been managing their ceph using ceph-ansible
this is not a trivial change as it will trigger a change in your
`ceph.conf` and then restart all your ceph services. Moreover if you did
some specific tweaks as well, prior to run ansible you should update the
`ceph_conf_overrides` variable to reflect your previous changes.

To avoid service restart, you need to know a bit of ansible for this,
but generally the idea would be to run ansible on a dummy host to
generate the ceph.conf, then scp this file to all your ceph hosts and
you should be good.

Closes: #693
Signed-off-by: Sébastien Han <seb@redhat.com>
(cherry picked from commit 47860a887fcc240d463a5b57a021f316c7533343)

9 years agoceph-mds: fix the handler
Sébastien Han [Sun, 8 May 2016 22:36:15 +0000 (00:36 +0200)]
ceph-mds: fix the handler

Moving the mds_name fact in the common role so the MDS can properly
be restarted.

Signed-off-by: Sébastien Han <seb@redhat.com>
(cherry picked from commit b3971354bf69e30fa63e400e4161f842e5d82379)

9 years agoceph-mds: fix MDS conditions
Sébastien Han [Sun, 8 May 2016 22:34:24 +0000 (00:34 +0200)]
ceph-mds: fix MDS conditions

We don't need to set is_systemd when we are applying a configuration for
a sysvinit init.

Signed-off-by: Sébastien Han <seb@redhat.com>
(cherry picked from commit 7949c2d5129ef13774757b3cc2ae0174d983f0b9)

9 years agoceph-mon: fix monitor startup
Sébastien Han [Sun, 8 May 2016 22:41:44 +0000 (00:41 +0200)]
ceph-mon: fix monitor startup

Somehow on CentOS 7.2 with Jewel, the service enablement by the Ansible service module
does not seem to work properly.

Signed-off-by: Sébastien Han <seb@redhat.com>
(cherry picked from commit c8e5e495524d515afa432b75fa06e7cb4032180e)

9 years agostrip any newlines from the init_system fact
Andrew Schoen [Fri, 6 May 2016 14:14:06 +0000 (09:14 -0500)]
strip any newlines from the init_system fact

Signed-off-by: Andrew Schoen <aschoen@redhat.com>
(cherry picked from commit 5139ed05926fa9b7502903e482cd29ac4ff943c9)

9 years agoadd a mdss group to the CLI testing inventory
Andrew Schoen [Thu, 5 May 2016 21:53:56 +0000 (16:53 -0500)]
add a mdss group to the CLI testing inventory

The ceph-mds role is being tested, but there was not group for it in the
inventory so ceph-mds was not being installed on the testing machine.

Signed-off-by: Andrew Schoen <aschoen@redhat.com>
(cherry picked from commit 8a09b9c6abf23ab5675428d72a958832692423d2)

9 years agoUse the use_systemd fact when starting rbd
Andrew Schoen [Thu, 5 May 2016 20:00:47 +0000 (15:00 -0500)]
Use the use_systemd fact when starting rbd

Signed-off-by: Andrew Schoen <aschoen@redhat.com>
(cherry picked from commit 8baf8830d6348c280c3a86a7d2f6d53c5f509c88)

9 years agoUse the use_systemd fact when starting mdss
Andrew Schoen [Thu, 5 May 2016 19:54:50 +0000 (14:54 -0500)]
Use the use_systemd fact when starting mdss

Signed-off-by: Andrew Schoen <aschoen@redhat.com>
(cherry picked from commit cb70d4975113e41f63b420665e9873711cf41074)

9 years agoUse the use_systemd fact when starting rgw
Andrew Schoen [Thu, 5 May 2016 19:31:12 +0000 (14:31 -0500)]
Use the use_systemd fact when starting rgw

Signed-off-by: Andrew Schoen <aschoen@redhat.com>
(cherry picked from commit 7436479c405f5054ac4a6b5535c299e7fceaad44)

9 years agouse the use_systemd fact when starting OSDs
Andrew Schoen [Thu, 5 May 2016 18:59:36 +0000 (13:59 -0500)]
use the use_systemd fact when starting OSDs

Signed-off-by: Andrew Schoen <aschoen@redhat.com>
(cherry picked from commit 5669b6ba7c2e021bd53088ea774508983dab11d3)

9 years agouse the use_systemd fact when starting monitors
Andrew Schoen [Thu, 5 May 2016 18:56:32 +0000 (13:56 -0500)]
use the use_systemd fact when starting monitors

Signed-off-by: Andrew Schoen <aschoen@redhat.com>
(cherry picked from commit 20d2d52ebd7d6cc426afe789a990d61ea61da0d2)

9 years agoAdds an use_systemd fact
Andrew Schoen [Wed, 4 May 2016 17:16:27 +0000 (12:16 -0500)]
Adds an use_systemd fact

This adds a helper fact that uses the ``init_system`` fact to determine if
we should be using systemd or not when controlling services.

Signed-off-by: Andrew Schoen <aschoen@redhat.com>
(cherry picked from commit 9eed0ea4e7aed94b0c354b62e54e0084ea7f0404)

9 years agoSet init_system fact and reference in ceph-osd role
Matt Thompson [Wed, 4 May 2016 10:47:24 +0000 (11:47 +0100)]
Set init_system fact and reference in ceph-osd role

The ceph-osd role currently uses ansible_service_mgr, which is a fact
only available on ansible 2.x and greater.  This commit sets a similar
fact called init_system which will store the contents of /proc/1/comm
(systemd, init, etc.) and then references it ceph-osd instead.

Closes #741

(cherry picked from commit ed2b7757d43ba7af96d97a9f345aba38e4410bd2)

9 years agotry to zap a device again if sgdisk fails on the first try
Alfredo Deza [Mon, 9 May 2016 13:12:28 +0000 (09:12 -0400)]
try to zap a device again if sgdisk fails on the first try

Signed-off-by: Alfredo Deza <adeza@redhat.com>
9 years agoceph-rgw: port start_radosgw task to use new ceph version facts
Alfredo Deza [Wed, 27 Apr 2016 14:51:15 +0000 (10:51 -0400)]
ceph-rgw: port start_radosgw task to use new ceph version facts

Signed-off-by: Alfredo Deza <adeza@redhat.com>
9 years agoceph-rbd-mirror: port start_rbd_mirror task to use new ceph version facts
Alfredo Deza [Wed, 27 Apr 2016 14:48:46 +0000 (10:48 -0400)]
ceph-rbd-mirror: port start_rbd_mirror task to use new ceph version facts

Signed-off-by: Alfredo Deza <adeza@redhat.com>
9 years agoceph-osd: port activate_osds task to use new ceph version facts
Alfredo Deza [Wed, 27 Apr 2016 14:43:09 +0000 (10:43 -0400)]
ceph-osd: port activate_osds  task to use new ceph version facts

Signed-off-by: Alfredo Deza <adeza@redhat.com>
9 years agoceph-mon: port start_monitor task to use new ceph version facts
Alfredo Deza [Wed, 27 Apr 2016 14:41:39 +0000 (10:41 -0400)]
ceph-mon: port start_monitor task to use new ceph version facts

Signed-off-by: Alfredo Deza <adeza@redhat.com>
9 years agoceph-mds: port pre_requisite task to use new ceph version facts
Alfredo Deza [Wed, 27 Apr 2016 14:39:57 +0000 (10:39 -0400)]
ceph-mds: port pre_requisite task to use new ceph version facts

Signed-off-by: Alfredo Deza <adeza@redhat.com>
9 years agoceph-common: port main handler to use new ceph version facts
Alfredo Deza [Wed, 27 Apr 2016 14:34:08 +0000 (10:34 -0400)]
ceph-common: port main handler to use new ceph version facts

Signed-off-by: Alfredo Deza <adeza@redhat.com>
9 years agoceph-mon: port deploy_monitors task to use new ceph version facts
Alfredo Deza [Wed, 27 Apr 2016 14:01:02 +0000 (10:01 -0400)]
ceph-mon: port deploy_monitors task to use new ceph version facts

Signed-off-by: Alfredo Deza <adeza@redhat.com>
9 years agoceph-common: port main task to use new ceph version facts
Alfredo Deza [Wed, 27 Apr 2016 13:50:10 +0000 (09:50 -0400)]
ceph-common: port main task to use new ceph version facts

Signed-off-by: Alfredo Deza <adeza@redhat.com>
9 years agocreate is_before_RELEASE vars as well
Alfredo Deza [Wed, 27 Apr 2016 13:28:17 +0000 (09:28 -0400)]
create is_before_RELEASE vars as well

Signed-off-by: Alfredo Deza <adeza@redhat.com>
9 years agoceph-common: include the facts task always
Alfredo Deza [Wed, 27 Apr 2016 13:19:32 +0000 (09:19 -0400)]
ceph-common: include the facts task always

Signed-off-by: Alfredo Deza <adeza@redhat.com>
9 years agoceph-common: create a facts task to define ceph versions
Alfredo Deza [Wed, 27 Apr 2016 13:17:56 +0000 (09:17 -0400)]
ceph-common: create a facts task to define ceph versions

Signed-off-by: Alfredo Deza <adeza@redhat.com>
9 years agoGet the correct OSD ID when the cluster name includes numbers
Andrew Schoen [Tue, 3 May 2016 19:47:47 +0000 (14:47 -0500)]
Get the correct OSD ID when the cluster name includes numbers

If the ceph cluster name includes numbers, the grep used to find the OSD
IDs from /var/lib/ceph/osd/ would also return the numbers that were in
the cluster name.

For example, if the cluster was named 'mine123' and there was only one
OSD on the node, then the task that finds the OSD IDs would return
'123' and '0'.

Signed-off-by: Andrew Schoen <aschoen@redhat.com>
(cherry picked from commit 4a181d420d7d0f8361d24f274c0e952b9d956945)

9 years agoAdds a use_server_package_split var
Andrew Schoen [Wed, 27 Apr 2016 21:10:39 +0000 (16:10 -0500)]
Adds a use_server_package_split var

This will allow a user to conditionally install the ceph package on rpm
based systems. Installing this package is not required or wanted in
versions passed infernalis.

Signed-off-by: Andrew Schoen <aschoen@redhat.com>
(cherry picked from commit f81140d21f5c77965c4438ba752bb174bf6bec70)

9 years agodefault rbd_client_directories to true
Andrew Schoen [Fri, 22 Apr 2016 15:56:30 +0000 (10:56 -0500)]
default rbd_client_directories to true

Signed-off-by: Andrew Schoen <aschoen@redhat.com>
(cherry picked from commit f297af23c5e0af3b86238b525987bc20cdbd4d38)

9 years agodefault rbd_client_admin_socket_path to /var/run/ceph
Andrew Schoen [Fri, 22 Apr 2016 15:59:00 +0000 (10:59 -0500)]
default rbd_client_admin_socket_path to /var/run/ceph

Signed-off-by: Andrew Schoen <aschoen@redhat.com>
(cherry picked from commit b649c4229ad9ea64579ae91e73b0ab400f1c5e1d)

9 years agodefault rbd_client_log_path to /var/log/ceph
Andrew Schoen [Fri, 22 Apr 2016 17:39:16 +0000 (12:39 -0500)]
default rbd_client_log_path to /var/log/ceph

Signed-off-by: Andrew Schoen <aschoen@redhat.com>
(cherry picked from commit 79a6e0522862fd116d1d0c5c08b339995759a8c9)

9 years agocorrectly set mon addr when monitor_interface is defined in group_vars
Andrew Schoen [Thu, 21 Apr 2016 14:09:16 +0000 (09:09 -0500)]
correctly set mon addr when monitor_interface is defined in group_vars

This fixes the ceph.conf template so that it will look for an inventory
defined value for monitor_interface or for monitor_interface defined in
a group_vars file.

Signed-off-by: Andrew Schoen <aschoen@redhat.com>
(cherry picked from commit 95d80ad3d30c90b9bb07ef46946779078ebec289)

9 years agoMerge pull request #715 from rootfs/jewel v1.0.5
Leseb [Fri, 15 Apr 2016 08:55:42 +0000 (10:55 +0200)]
Merge pull request #715 from rootfs/jewel

some fixes to run containers on Atomic

9 years agoset path to find mon and osd systemd unit file templates 715/head
Huamin Chen [Thu, 14 Apr 2016 21:57:27 +0000 (21:57 +0000)]
set path to find mon and osd systemd unit file templates

Signed-off-by: Huamin Chen <hchen@redhat.com>
9 years agofix deployment on Atomic
Huamin Chen [Thu, 14 Apr 2016 13:41:05 +0000 (13:41 +0000)]
fix deployment on Atomic

Signed-off-by: Huamin Chen <hchen@redhat.com>
9 years agoMerge pull request #713 from stpierre/fix-rbd-client-dirs
Leseb [Wed, 13 Apr 2016 20:25:59 +0000 (22:25 +0200)]
Merge pull request #713 from stpierre/fix-rbd-client-dirs

Fix creation of RBD client directories

9 years agoMerge pull request #712 from ceph/monitor-interface-not-defined
Leseb [Wed, 13 Apr 2016 20:19:33 +0000 (22:19 +0200)]
Merge pull request #712 from ceph/monitor-interface-not-defined

Check to see if hostvars[host]['monitor_interface'] is defined

9 years agoFix creation of RBD client directories 713/head
Chris St. Pierre [Wed, 13 Apr 2016 20:10:44 +0000 (15:10 -0500)]
Fix creation of RBD client directories

Need to use a template, not a bare variable name.

9 years agoCheck to see if hostvars[host]['monitor_interface'] is defined 712/head
Andrew Schoen [Wed, 13 Apr 2016 18:34:51 +0000 (13:34 -0500)]
Check to see if hostvars[host]['monitor_interface'] is defined

This fixes https://github.com/ceph/ceph-ansible/issues/711

Signed-off-by: Andrew Schoen <aschoen@redhat.com>
9 years agoMerge pull request #707 from cloudnull/ct_updates
Leseb [Wed, 13 Apr 2016 16:52:56 +0000 (18:52 +0200)]
Merge pull request #707 from cloudnull/ct_updates

Update the config_template plugin

9 years agoMerge pull request #710 from bengland2/limit-autodiscover-check v1.0.4 v1.04
Leseb [Tue, 12 Apr 2016 15:01:04 +0000 (17:01 +0200)]
Merge pull request #710 from bengland2/limit-autodiscover-check

purge-cluster - fix osd_auto_discovery check

9 years agodevices var is defined in context of osds role 710/head
Ben England [Tue, 12 Apr 2016 13:49:44 +0000 (09:49 -0400)]
devices var is defined in context of osds role

9 years agoMerge pull request #705 from ceph/no-agent-role-test
Leseb [Tue, 12 Apr 2016 12:05:33 +0000 (14:05 +0200)]
Merge pull request #705 from ceph/no-agent-role-test

remove agent role, no packages for it upstream yet

9 years agoremove rbd-mirror as well from testing. No packages are available yet 705/head
Alfredo Deza [Tue, 12 Apr 2016 11:38:13 +0000 (07:38 -0400)]
remove rbd-mirror as well from testing. No packages are available yet

Signed-off-by: Alfredo Deza <adeza@redhat.com>
9 years agoremove coreos from default testing as well
Alfredo Deza [Tue, 12 Apr 2016 11:17:35 +0000 (07:17 -0400)]
remove coreos from default testing as well

Signed-off-by: Alfredo Deza <adeza@redhat.com>
9 years agoMerge pull request #708 from ceph/docker-img
Leseb [Tue, 12 Apr 2016 10:08:35 +0000 (12:08 +0200)]
Merge pull request #708 from ceph/docker-img

docker: do not hardcode image and user name

9 years agoMerge pull request #706 from fultonj/patch-1
Leseb [Tue, 12 Apr 2016 10:07:45 +0000 (12:07 +0200)]
Merge pull request #706 from fultonj/patch-1

Balance tags on <your_host

9 years agoMerge pull request #703 from ceph/issue-670
Leseb [Tue, 12 Apr 2016 10:05:28 +0000 (12:05 +0200)]
Merge pull request #703 from ceph/issue-670

if ceph-disk fails to activate an OSD then bubble up the error

9 years agoMerge pull request #688 from bengland2/safer-disk-zap
Leseb [Tue, 12 Apr 2016 10:04:39 +0000 (12:04 +0200)]
Merge pull request #688 from bengland2/safer-disk-zap

low-priority - more robust, simpler, idempotent purge-cluster.yml

9 years agoMerge pull request #696 from stpierre/dedup-rbd-client-dirs
Leseb [Tue, 12 Apr 2016 10:03:56 +0000 (12:03 +0200)]
Merge pull request #696 from stpierre/dedup-rbd-client-dirs

Deduplicate RBD client directory creation

9 years agodocker: do not hardcode image and user name 708/head
Sébastien Han [Tue, 12 Apr 2016 10:01:53 +0000 (12:01 +0200)]
docker: do not hardcode image and user name

Signed-off-by: Sébastien Han <seb@redhat.com>
9 years agoUpdate the config_template plugin 707/head
Kevin Carter [Tue, 12 Apr 2016 04:28:40 +0000 (23:28 -0500)]
Update the config_template plugin

These changes address 2 issues:

* The config template plugin in v1 mode was not respecting a list
  of integers and would cause a stacktrace because the ConfigParese
  module expects everything written to be a string.
* The config template plugin in v2 mode was not properly handingly
  the multi-string-op type due to an oversight in the ``_option_write``
  method.

With these updates the config template is now able to better support
all of the potential options that could be thrown at it when writing
an INI config file.

Signed-off-by: Kevin Carter <kevin.carter@rackspace.com>
9 years agoBalance tags on <your_host 706/head
John Fulton [Tue, 12 Apr 2016 02:25:53 +0000 (22:25 -0400)]
Balance tags on <your_host

Third of occurrence of <your_host> didn't have a closing tag. This is a tiny change to add the he missing '>'.

This will let people quickly update this file with sed. For example: `sed s/\<your_host\>/192.168.1.41/g -i cluster-maintenance.yml`.

9 years agothis is how ceph-disk partition really looks 688/head
Ben England [Mon, 11 Apr 2016 20:53:45 +0000 (16:53 -0400)]
this is how ceph-disk partition really looks

9 years agoremove agent role, no packages for it upstream yet
Alfredo Deza [Mon, 11 Apr 2016 19:48:29 +0000 (15:48 -0400)]
remove agent role, no packages for it upstream yet

Signed-off-by: Alfredo Deza <adeza@redhat.com>
9 years agoif ceph-disk fails to activate an OSD then bubble up the error 703/head
Alfredo Deza [Mon, 11 Apr 2016 14:28:03 +0000 (10:28 -0400)]
if ceph-disk fails to activate an OSD then bubble up the error

Signed-off-by: Alfredo Deza <adeza@redhat.com>
9 years agoMerge pull request #701 from ceph/conf-fixes
Leseb [Fri, 8 Apr 2016 20:00:32 +0000 (22:00 +0200)]
Merge pull request #701 from ceph/conf-fixes

look at monitor_interface in hostvars when choosing interface or address

9 years agolook at monitor_interface in hostvars when choosing interface or address 701/head
Andrew Schoen [Fri, 8 Apr 2016 19:23:12 +0000 (14:23 -0500)]
look at monitor_interface in hostvars when choosing interface or address

This fixes a bug where monitor_interface might be set in your inventory
file and not by using group_vars or --extra-vars causing the template to
use the default address of 0.0.0.0 instead of the defined
monitor_interface.

Signed-off-by: Andrew Schoen <aschoen@redhat.com>
9 years agoDeduplicate RBD client directory creation 696/head
Chris St. Pierre [Thu, 7 Apr 2016 14:11:50 +0000 (09:11 -0500)]
Deduplicate RBD client directory creation

Instead of creating the RBD client socket path three different places
in three different ways, this creates it once. Ceph on OpenStack users
have the option to customize the permissions of the RBD client
directories.

Fixes #687

9 years agorebased
Ben England [Fri, 8 Apr 2016 17:01:11 +0000 (13:01 -0400)]
rebased

9 years agoMerge pull request #700 from ceph/fix-readme-rbdmirror
Leseb [Fri, 8 Apr 2016 10:32:30 +0000 (12:32 +0200)]
Merge pull request #700 from ceph/fix-readme-rbdmirror

rbd-mirror: fix readme title

9 years agorbd-mirror: fix readme title 700/head
Sébastien Han [Fri, 8 Apr 2016 10:32:05 +0000 (12:32 +0200)]
rbd-mirror: fix readme title

Signed-off-by: Sébastien Han <seb@redhat.com>
9 years agoMerge pull request #699 from ceph/contrib-rbdmirror
Leseb [Fri, 8 Apr 2016 10:31:11 +0000 (12:31 +0200)]
Merge pull request #699 from ceph/contrib-rbdmirror

contrib: update with ceph-rbd-mirror

9 years agocontrib: update with ceph-rbd-mirror 699/head
Sébastien Han [Fri, 8 Apr 2016 10:30:14 +0000 (12:30 +0200)]
contrib: update with ceph-rbd-mirror

Signed-off-by: Sébastien Han <seb@redhat.com>
9 years agoMerge pull request #698 from ceph/rbdmirror-license
Leseb [Fri, 8 Apr 2016 10:18:35 +0000 (12:18 +0200)]
Merge pull request #698 from ceph/rbdmirror-license

ceph-rbd-mirorr: add license file

9 years agoceph-rbd-mirorr: add license file 698/head
Sébastien Han [Fri, 8 Apr 2016 10:17:46 +0000 (12:17 +0200)]
ceph-rbd-mirorr: add license file

Signed-off-by: Sébastien Han <seb@redhat.com>
9 years agoMerge pull request #678 from ceph/rbd-mirror
Leseb [Fri, 8 Apr 2016 10:12:30 +0000 (12:12 +0200)]
Merge pull request #678 from ceph/rbd-mirror

ceph-rbd-mirror: add new role

9 years agoceph-rbd-mirror: add new role 678/head
Sébastien Han [Sun, 3 Apr 2016 22:35:09 +0000 (00:35 +0200)]
ceph-rbd-mirror: add new role

Signed-off-by: Sébastien Han <seb@redhat.com>
9 years agoMerge pull request #697 from fultonj/rolling_update_sysvinit
Leseb [Fri, 8 Apr 2016 09:33:17 +0000 (11:33 +0200)]
Merge pull request #697 from fultonj/rolling_update_sysvinit

sysvinit changes to rolling update playbook

9 years agoThis update has the following changes for the rolling update playbook. 697/head
John Fulton [Fri, 8 Apr 2016 03:55:15 +0000 (23:55 -0400)]
This update has the following changes for the rolling update playbook.

1. Change how sysvinit ceph is determined to be enabled

For mons, the playbook checks if sysvinit is enabled by trying to
stat /var/lib/ceph/mon/ceph-{{ ansible_hostname }}/sysvinit [1].
However, that file is not created when a monitor is configured to
use sysvinit, instead, Ansible's service module is used [2]. Ansible
2.0 can verify if a service is enabled and does this by checking for
a glob, in this context it would be '/etc/rc?.d/S??ceph' [3]. Because
Ansible 1.9 does not support this feature, this change updates
rolling_update.yml by checking if sysvinit is enabled by having stat
glob for the same pattern and following the symlink. This is done only
for the mons.

2. Change how sysvinit ceph is restarted

The playbook passes the argument "mon" to the sysv init script
but, the init script does not necessarily take that argument and
it failed when tested on a RHEL7 system. However, dropping the
argument and just using Ansible's service module for state=restarted
worked so this change does not have this line for when
monsysvinit.stat.exists. A similar change is in this pull request
for the OSD restart (removing args=osd).

A second ceph mon restart command is run regardless of any conditions
being met. I am not sure why the service is restarted in the case of
upstart or sysvinit and then restarted again. I am going to assume
there is a subtle reason for this and not touch this second run but I
added a condition so that when ansible_os_family is Red Hat, then mon
is not passed as an argument, otherwise it is restarted as was already
in place.

[1] https://github.com/ceph/ceph-ansible/blob/v1.0.3/rolling_update.yml#L32-L33
[2] https://github.com/ceph/ceph-ansible/blob/v1.0.3/roles/ceph-mon/tasks/start_monitor.yml#L42-L45
[3] https://github.com/ansible/ansible-modules-core/blob/stable-2.0/system/service.py#L492-L493

9 years agomore robust, simpler, idempotent
Ben England [Wed, 6 Apr 2016 19:58:17 +0000 (15:58 -0400)]
more robust, simpler, idempotent

9 years agoMerge pull request #684 from bengland2/when-hammer-try2
Leseb [Wed, 6 Apr 2016 17:38:05 +0000 (19:38 +0200)]
Merge pull request #684 from bengland2/when-hammer-try2

fix when clauses for installation of ceph RPM

9 years agoSebastien's suggestion - add infernalis to list of releases with no separate ceph... 684/head
Ben England [Wed, 6 Apr 2016 17:14:24 +0000 (13:14 -0400)]
Sebastien's suggestion - add infernalis to list of releases with no separate ceph-mon or ceph-osd

9 years agoMerge pull request #685 from bengland2/create-sysconfig-ceph
Leseb [Wed, 6 Apr 2016 11:21:00 +0000 (13:21 +0200)]
Merge pull request #685 from bengland2/create-sysconfig-ceph

create /etc/sysconfig/ceph if it doesn't already exist

9 years agoMerge pull request #683 from stpierre/fix-config-template-docker
Leseb [Wed, 6 Apr 2016 11:15:47 +0000 (13:15 +0200)]
Merge pull request #683 from stpierre/fix-config-template-docker

Fix mis-use of config_template in docker tasks

9 years agocreate /etc/sysconfig/ceph if it doesn't already exist 685/head
Ben England [Wed, 6 Apr 2016 01:33:05 +0000 (21:33 -0400)]
create /etc/sysconfig/ceph if it doesn't already exist

9 years agofix when clauses for installation of ceph RPM
Ben England [Wed, 6 Apr 2016 01:11:37 +0000 (21:11 -0400)]
fix when clauses for installation of ceph RPM

9 years agoFix mis-use of config_template in docker tasks 683/head
Chris St. Pierre [Tue, 5 Apr 2016 17:09:09 +0000 (12:09 -0500)]
Fix mis-use of config_template in docker tasks

As written, generating the config file for ceph-mon in Docker yielded:

ERROR: config_template is not a legal parameter in an Ansible task or
handler

This fixes that error condition.

9 years agoMerge pull request #682 from ceph/change-pool-min-size
Leseb [Tue, 5 Apr 2016 15:03:25 +0000 (17:03 +0200)]
Merge pull request #682 from ceph/change-pool-min-size

ceph-common: let ceph handle pool_default_min_size

9 years agoceph-common: let ceph handle pool_default_min_size 682/head
Sébastien Han [Tue, 5 Apr 2016 15:01:49 +0000 (17:01 +0200)]
ceph-common: let ceph handle pool_default_min_size

We now set the value to 0, which means no specific default; ceph will
use size-size/2

Signed-off-by: Sébastien Han <seb@redhat.com>
9 years agoMerge pull request #681 from ceph/change-pool-default-size
Leseb [Tue, 5 Apr 2016 14:26:17 +0000 (16:26 +0200)]
Merge pull request #681 from ceph/change-pool-default-size

Change pool default size

9 years agovagrant use pool_default_size to 2 681/head
Sébastien Han [Tue, 5 Apr 2016 14:24:21 +0000 (16:24 +0200)]
vagrant use pool_default_size to 2

For dev envs it is easir to use a replica count of 2 instead of 3.

Signed-off-by: Sébastien Han <seb@redhat.com>
9 years agoceph-common: change the pool_default_size to 3
Sébastien Han [Tue, 5 Apr 2016 14:22:29 +0000 (16:22 +0200)]
ceph-common: change the pool_default_size to 3

3 has been the default since Firefly I believe so we should stick with
this value.

Signed-off-by: Sébastien Han <seb@redhat.com>