]> git.apps.os.sepia.ceph.com Git - ceph.git/log
ceph.git
12 years agoDisplay error message when locking a vpm fails due to downburst errors.
Warren Usui [Wed, 14 Aug 2013 00:16:00 +0000 (17:16 -0700)]
Display error message when locking a vpm fails due to downburst errors.
When doing a lock-many, do not lock any of the vpms when downburst errors
occur.   Do not display message on unlock with downburst error, because
slot is still freed (and can be locked by someone else).

Fix: 5957
Signed-off-by: Warren Usui <warren.usui@inktank.com>
12 years agoMerge pull request #30 from ceph/wip-5952
Sage Weil [Tue, 13 Aug 2013 18:45:35 +0000 (11:45 -0700)]
Merge pull request #30 from ceph/wip-5952

Fix a case where _get_config_value_for_remote could return None mistakenly

12 years agoFix a case where _get_config_value_for_remote could return None mistakenly
Zack Cerza [Tue, 13 Aug 2013 18:30:22 +0000 (13:30 -0500)]
Fix a case where _get_config_value_for_remote could return None mistakenly

12 years agoMerge remote-tracking branch 'gh/next'
Sage Weil [Tue, 13 Aug 2013 18:03:00 +0000 (11:03 -0700)]
Merge remote-tracking branch 'gh/next'

12 years agosamba: wait longer for smbd to start
Sage Weil [Tue, 13 Aug 2013 18:02:32 +0000 (11:02 -0700)]
samba: wait longer for smbd to start

sleep(1) is occasionally not long enough, it seems!

12 years agoMerge pull request #27 from dachary/wip-5946
Zack Cerza [Tue, 13 Aug 2013 17:46:30 +0000 (10:46 -0700)]
Merge pull request #27 from dachary/wip-5946

do not check the jobid if check-locks is False

12 years agodo not check the jobid if check-locks is False
Loic Dachary [Tue, 13 Aug 2013 10:43:21 +0000 (12:43 +0200)]
do not check the jobid if check-locks is False

Because it relies on the lock server which is presumably not available
since check-locks was set to False. It matters when using teuthology
on a minimal installation.

http://tracker.ceph.com/issues/5946 fixes #5946

Signed-off-by: Loic Dachary <loic@dachary.org>
12 years agoMerge pull request #28 from alfredodeza/5887
Zack Cerza [Tue, 13 Aug 2013 16:10:45 +0000 (09:10 -0700)]
Merge pull request #28 from alfredodeza/5887

Allow a user to define a ceph-deploy branch to use (instead of master)

12 years agoMerge pull request #24 from ceph/wip-5937
Alfredo Deza [Tue, 13 Aug 2013 15:57:18 +0000 (08:57 -0700)]
Merge pull request #24 from ceph/wip-5937

Don't hit an AttributeError if ctx.config['overrides'] is set to None

12 years agoremove fallback of the fallback. so silly
Alfredo Deza [Tue, 13 Aug 2013 15:56:37 +0000 (11:56 -0400)]
remove fallback of the fallback. so silly

Signed-off-by: Alfredo Deza <alfredo@deza.pe>
12 years agoRemove three characters just for Alfredo
Zack Cerza [Tue, 13 Aug 2013 15:45:53 +0000 (10:45 -0500)]
Remove three characters just for Alfredo

12 years agosafer fallback. Updates docstring
Alfredo Deza [Tue, 13 Aug 2013 15:32:15 +0000 (11:32 -0400)]
safer fallback. Updates docstring

Signed-off-by: Alfredo Deza <alfredo@deza.pe>
12 years agoradosgw-agent.py: refactor, enable overrides
Joe Buck [Fri, 9 Aug 2013 04:14:56 +0000 (21:14 -0700)]
radosgw-agent.py: refactor, enable overrides

Refactored the radosgw-agent.py code so that it
is structured more like existing teuthology
tasks.
Additionally, added code to enable:
using the override field in YAML files,
specifying which radosgw-agent github branch
to use checkout and for the YAML file to
specify one of the following: a full sync,
an incremental sync and the starting of the
test radosgw-agent server (previously the
server was always started by this task).

Signed-off-by: Joe Buck <jbbuck@gmail.com>
Reviewed-by: Josh Durgin <josh.durgin@inktank.com>
12 years agorgw.py: fix example config
Joe Buck [Fri, 9 Aug 2013 04:04:42 +0000 (21:04 -0700)]
rgw.py: fix example config

In the example config, the region root and
zone root where pointing to the same pool,
which is not a best practice. Updated the
example to show them pointing to different pools.

Signed-off-by: Joe Buck <jbbuck@gmail.com>
12 years agorgw.py: add a better error message
Joe Buck [Fri, 9 Aug 2013 04:03:27 +0000 (21:03 -0700)]
rgw.py: add a better error message

When pulling region info from the config
structure, if the region info isn't there,
log a more helpful message.

Signed-off-by: Joe Buck <jbbuck@gmail.com>
12 years agos3readwrite.py: make user creation optional
Joe Buck [Sat, 3 Aug 2013 23:55:19 +0000 (16:55 -0700)]
s3readwrite.py: make user creation optional

Extend the s3readwrite.py task to enable the
creation and deletion of users for the s3readwrite
tests to be independently specified with the default
assumption to both being true.
This is needed for tests that will create a user and
data in one execution and read it in another.

Signed-off-by: Joe Buck <jbbuck@gmail.com>
12 years agorgw.py: refactor configure
Joe Buck [Thu, 1 Aug 2013 06:19:57 +0000 (23:19 -0700)]
rgw.py: refactor configure

Extend the rados pool configuration options to
specify all pools (if desired).
Also, reordered zone and region configuration
so that they're configure (per client) in
this order: zone, region, set default region

Signed-off-by: Joe Buck <jbbuck@gmail.com>
12 years agorgy.py: make log_(data|metadata) configurable
Joe Buck [Thu, 1 Aug 2013 21:03:59 +0000 (14:03 -0700)]
rgy.py: make log_(data|metadata) configurable

The log_data and log_metadata are made configurable
via the YAML file and default to false
(meaning neither data nor metadata operations are
logged).

Signed-off-by: Joe Buck <jbbuck@gmail.com>
12 years agoagent name should also contain port number
Yehuda Sadeh [Mon, 5 Aug 2013 03:57:25 +0000 (20:57 -0700)]
agent name should also contain port number

otherwise multiple agents per machines will have the same id

Signed-off-by: Yehuda Sadeh <yehuda@inktank.com>
12 years agoradosgw-admin: various multiregion related fixes and changes
Yehuda Sadeh [Sat, 3 Aug 2013 03:28:46 +0000 (20:28 -0700)]
radosgw-admin: various multiregion related fixes and changes

Go to the master zone in the master region for radosgw-admin
operations. Trigger metadata sync. Other fixes.

Signed-off-by: Yehuda Sadeh <yehuda@inktank.com>
12 years agoMerge pull request #25 from dmick/master
Zack Cerza [Mon, 12 Aug 2013 23:10:26 +0000 (16:10 -0700)]
Merge pull request #25 from dmick/master

tasks/ceph-deploy: allow configurations with no mds

12 years agoChange "Exception" to "RuntimeError" as advised by zackc
Dan Mick [Mon, 12 Aug 2013 22:43:02 +0000 (15:43 -0700)]
Change "Exception" to "RuntimeError" as advised by zackc

Signed-off-by: Dan Mick <dan.mick@inktank.com>
12 years agotasks/ceph-deploy: allow configurations with no mds
Dan Mick [Mon, 12 Aug 2013 21:55:14 +0000 (14:55 -0700)]
tasks/ceph-deploy: allow configurations with no mds

Signed-off-by: Dan Mick <dan.mick@inktank.com>
12 years agoDon't hit an AttributeError if ctx.config['overrides'] is set to None
Zack Cerza [Mon, 12 Aug 2013 21:18:42 +0000 (16:18 -0500)]
Don't hit an AttributeError if ctx.config['overrides'] is set to None

12 years agoMerge pull request #23 from ceph/pep8
Alfredo Deza [Mon, 12 Aug 2013 17:40:58 +0000 (10:40 -0700)]
Merge pull request #23 from ceph/pep8

Make this module pep8-clean.

12 years agoMake this module pep8-clean.
Zack Cerza [Mon, 12 Aug 2013 16:40:04 +0000 (11:40 -0500)]
Make this module pep8-clean.

12 years agoMerge pull request #22 from ceph/rpm-upgrades
Sage Weil [Sat, 10 Aug 2013 18:35:05 +0000 (11:35 -0700)]
Merge pull request #22 from ceph/rpm-upgrades

Fix rpm upgrades for nested branch configs

12 years agoMake _get_baseurlinfo_and_dist() provide the correct uri
Zack Cerza [Sat, 10 Aug 2013 17:45:21 +0000 (12:45 -0500)]
Make _get_baseurlinfo_and_dist() provide the correct uri

12 years agoMerge pull request #21 from ceph/rpm-upgrades
Zack Cerza [Fri, 9 Aug 2013 23:26:39 +0000 (16:26 -0700)]
Merge pull request #21 from ceph/rpm-upgrades

Make install.upgrade work on RPM-based systems.

12 years agoMake install.upgrade work with CentOS/RHEL/Fedora.
Zack Cerza [Fri, 9 Aug 2013 22:01:36 +0000 (17:01 -0500)]
Make install.upgrade work with CentOS/RHEL/Fedora.

12 years agoAdd Remote.system_type property. Also, clean up whitespace.
Zack Cerza [Fri, 9 Aug 2013 18:28:38 +0000 (13:28 -0500)]
Add Remote.system_type property. Also, clean up whitespace.

12 years agoadmin_socket: split command into multiple arguments
Sage Weil [Fri, 9 Aug 2013 16:04:25 +0000 (09:04 -0700)]
admin_socket: split command into multiple arguments

We were passing the command as a single string, but that does not work the
same way it used to.

Signed-off-by: Sage Weil <sage@inktank.com>
12 years agoceph.conf: open osd classes on start
Sage Weil [Wed, 7 Aug 2013 21:01:00 +0000 (14:01 -0700)]
ceph.conf: open osd classes on start

This is the default for master, but needs to be explicitly enabled for
cuttlefish.

12 years agoMerge remote-tracking branch 'gh/next'
Sage Weil [Wed, 7 Aug 2013 16:03:51 +0000 (09:03 -0700)]
Merge remote-tracking branch 'gh/next'

12 years agoadd a key for ceph-deploy-branch detection
Alfredo Deza [Tue, 6 Aug 2013 22:17:51 +0000 (15:17 -0700)]
add a key for ceph-deploy-branch detection

Signed-off-by: Alfredo Deza <alfredo@deza.pe>
12 years agoAdd a useful error message in case of an invalid hostname
Zack Cerza [Mon, 5 Aug 2013 23:48:27 +0000 (18:48 -0500)]
Add a useful error message in case of an invalid hostname

12 years agos3readwrite.py: enable overrides
Joe Buck [Sun, 4 Aug 2013 19:36:04 +0000 (12:36 -0700)]
s3readwrite.py: enable overrides

Enable s3readwrite task to have the branch to
download specified and for overrides to be
incorporated into the config at run-time.
Code based on the s3tests.py task.

Signed-off-by: Joe Buck <jbbuck@gmail.com>
Reviewed-by: Josh Durgin <josh.durgin@inktank.com>
12 years agoMerge remote-tracking branch 'gh/next'
Sage Weil [Sun, 4 Aug 2013 21:54:26 +0000 (14:54 -0700)]
Merge remote-tracking branch 'gh/next'

12 years agosamba: wait for smbd process to exit
Sage Weil [Sun, 4 Aug 2013 21:54:20 +0000 (14:54 -0700)]
samba: wait for smbd process to exit

I saw

2013-08-03T12:56:26.641 DEBUG:teuthology.orchestra.run:Running [10.214.131.28]: 'sudo killall -9 smbd'
2013-08-03T12:56:26.727 DEBUG:teuthology.orchestra.run:Running [10.214.131.28]: 'sudo lsof /home/ubuntu/cephtest/93695/mnt.0'
2013-08-03T12:56:26.830 INFO:teuthology.orchestra.run.out:[10.214.131.28]: COMMAND   PID USER   FD   TYPE DEVICE SIZE/OFF NODE NAME
2013-08-03T12:56:26.830 INFO:teuthology.orchestra.run.out:[10.214.131.28]: smbd    12381 root  cwd    DIR    0,0        0    1 /home/ubuntu/cephtest/93695/mnt.0

which makes me think we just need to wait a moment before
attempting the umount?

Signed-off-by: Sage Weil <sage@inktank.com>
12 years agorgw.py: add None object check when parsing info
Joe Buck [Sun, 4 Aug 2013 00:16:52 +0000 (17:16 -0700)]
rgw.py: add None object check when parsing info

The rgw task was failing to check for a None object
when parsing user info in the case where there were
config options set for the client that did not include
user info (e.g. valgrind: ).

12 years agorgw: do not ignore leaks
Sage Weil [Fri, 2 Aug 2013 18:57:14 +0000 (11:57 -0700)]
rgw: do not ignore leaks

Signed-off-by: Sage Weil <sage@inktank.com>
12 years agorgw.py: adding a safety check
Joe Buck [Fri, 2 Aug 2013 03:54:38 +0000 (20:54 -0700)]
rgw.py: adding a safety check

Correcting a bug where specifying
a rgw server for a client but not specifying
a system user would throw an exception.

Signed-off-by: Joe Buck <jbbuck@gmail.com>
Reviewed-by: Josh Durgin <josh.durgin@inktank.com>
12 years agorgw: add log settings to example config
Josh Durgin [Fri, 2 Aug 2013 02:11:28 +0000 (19:11 -0700)]
rgw: add log settings to example config

Signed-off-by: Josh Durgin <josh.durgin@inktank.com>
12 years agoradosgw-agent: get verbose logs from the agent
Josh Durgin [Fri, 2 Aug 2013 01:56:48 +0000 (18:56 -0700)]
radosgw-agent: get verbose logs from the agent

This makes it a lot easier to debug.

Signed-off-by: Josh Durgin <josh.durgin@inktank.com>
12 years agoradosgw-agent: move zone extraction to helper
Josh Durgin [Fri, 2 Aug 2013 01:56:08 +0000 (18:56 -0700)]
radosgw-agent: move zone extraction to helper

This will be useful for other tasks.

Signed-off-by: Josh Durgin <josh.durgin@inktank.com>
12 years agorgw: don't make domain_root pool the same as zone root
Josh Durgin [Fri, 2 Aug 2013 01:54:52 +0000 (18:54 -0700)]
rgw: don't make domain_root pool the same as zone root

These are supposed to be separate pools.

Signed-off-by: Josh Durgin <josh.durgin@inktank.com>
12 years agorgw: create system users with the --system flag
Josh Durgin [Fri, 2 Aug 2013 01:50:11 +0000 (18:50 -0700)]
rgw: create system users with the --system flag

without this they have no special permissions

Signed-off-by: Josh Durgin <josh.durgin@inktank.com>
12 years agorgy.py: make log_(data|metadata) configurable
Joe Buck [Thu, 1 Aug 2013 21:03:59 +0000 (14:03 -0700)]
rgy.py: make log_(data|metadata) configurable

The log_data and log_metadata are made configurable
via the YAML file and default to false
(meaning neither data nor metadata operations are
logged).

Signed-off-by: Joe Buck <jbbuck@gmail.com>
Reviewed-by: Josh Durgin <josh.durgin@inktank.com>
12 years agoradosgw-agent: rename task to match tool name
Josh Durgin [Fri, 2 Aug 2013 00:23:40 +0000 (17:23 -0700)]
radosgw-agent: rename task to match tool name

Signed-off-by: Josh Durgin <josh.durgin@inktank.com>
12 years agorgw_sync_agent: fix a bunch of issues
Josh Durgin [Fri, 2 Aug 2013 00:18:00 +0000 (17:18 -0700)]
rgw_sync_agent: fix a bunch of issues

- Read ceph.conf from stored copy that includes overrides
- Get system users and keys from cluster instead of reading other
  tasks' yaml, which may not be complete.
- Put zone info extraction from the cluster into utility functions,
  since it'll be useful for other tests later.
- Work with more than one agent on a single host
- Accept more than one client to run, like almost every other task
- Rename target to dest for consistency with radosgw-agent
- Don't make everything one large function

Signed-off-by: Josh Durgin <josh.durgin@inktank.com>
12 years agorgw: ignore exit status when removing base apache dir
Josh Durgin [Fri, 2 Aug 2013 00:08:01 +0000 (17:08 -0700)]
rgw: ignore exit status when removing base apache dir

subdirs have already been removed, and this only needs to succeed once
per host.

Signed-off-by: Josh Durgin <josh.durgin@inktank.com>
12 years agorgw_sync_agent: add new rgw_sync_agent task
Greg Farnum [Wed, 31 Jul 2013 06:14:32 +0000 (23:14 -0700)]
rgw_sync_agent: add new rgw_sync_agent task

This pulls access data out of the rgw task and off disk,
and then downloads, sets up, and runs an rgw sync agent
in test mode.

Signed-off-by: Greg Farnum <greg@inktank.com>
12 years agoFix for #5836 (--lock-many with vms)
Sandon Van Ness [Fri, 2 Aug 2013 00:04:04 +0000 (17:04 -0700)]
Fix for #5836 (--lock-many with vms)

This makes --lock-many work when --machine-type vps is passed.
Before it wasn't handled correctly and guests were not created.
Now it creates and gives the back the user the list-targets for
said guests.

teuthology-lock --lock-many 4 --machine-type vps --os-type centos

This fixes issue #5836

Signed-off-by: Sandon Van Ness <sandon@inktank.com>
Reviewed-by: Alfredo Deza <alfredo@deza.pe>
12 years agoFix for Debian wheezy (remove vda from block device list)
Sandon Van Ness [Thu, 1 Aug 2013 19:33:11 +0000 (12:33 -0700)]
Fix for Debian wheezy (remove vda from block device list)

On debian wheezy its mount output uses device-by-label and makes
our normal method of checking if a device is mounted not work.
Since vm's will always be vda for their boot device we will just
remove it from devs if its in there so it doesn't attempt to zap
vda.

I also added a strip() to remove the last blank entry that was
always getting added to the devs list on all machines. Example:

devs=['/dev/sda', '/dev/sdb', '/dev/sdc', '/dev/sdd', '']

Signed-off-by: Sandon Van Ness <sandon@inktank.com>
Reviewed-by: Alfredo Deza <alfredo@deza.pe>
12 years agorgw.py: check for empty client config
Joe Buck [Thu, 1 Aug 2013 17:07:56 +0000 (10:07 -0700)]
rgw.py: check for empty client config

Fixes a bug where an rgw client without
a system user specified would cause teuthology
to error out.

Signed-off-by: Joe Buck <jbbuck@gmail.com>
Reviewed-by: Yehuda Sadeh <yehuda@inktank.com>
12 years agorgw.py: refactoring to separate user creation
Joe Buck [Wed, 31 Jul 2013 22:14:48 +0000 (15:14 -0700)]
rgw.py: refactoring to separate user creation

By separating out the user creation from
generating the region/zone info, we can generate
users for RGW tests that run against the default
pools.

Signed-off-by: Joe Buck <jbbuck@gmail.com>
Reviewed-by: Josh Durgin <josh.durgin@inktank.com>
12 years agorgw.py: change --secret-key to --secret
Joe Buck [Wed, 31 Jul 2013 21:05:49 +0000 (14:05 -0700)]
rgw.py: change --secret-key to --secret

A 'user create' call was being passed to radosgw-admin
with '--secret-key' instead of the valid '--secret'
which was causing a random secret to be generated,
which was causing subsequent tests to fail.

Signed-off-by: Joe Buck <jbbuck@gmail.com>
12 years agos3/swift tests: call radosgw-admin as the right client
Josh Durgin [Wed, 31 Jul 2013 20:32:58 +0000 (13:32 -0700)]
s3/swift tests: call radosgw-admin as the right client

This allows the right region and zone info to be read from ceph.conf

Signed-off-by: Josh Durgin <josh.durgin@inktank.com>
12 years agorgw: fix dir creation and keyring
Josh Durgin [Wed, 31 Jul 2013 19:47:48 +0000 (12:47 -0700)]
rgw: fix dir creation and keyring

fastcgi_sock dir needs to exist before radosgw starts, and apache-execed radosgw needs an explicit keyring argument.

Signed-off-by: Josh Durgin <josh.durgin@inktank.com>
12 years agoReconnect after running chef task.
Sandon Van Ness [Wed, 31 Jul 2013 00:26:59 +0000 (17:26 -0700)]
Reconnect after running chef task.

Just a simple change to reconnect to SSH after running
ceph-qa-chef to get around things like ulimit changes.

Signed-off-by: Sandon Van Ness <sandon@inktank.com>
12 years agoFix RHEL/centos ceph-deploy installs.
Sandon Van Ness [Tue, 30 Jul 2013 19:36:12 +0000 (12:36 -0700)]
Fix RHEL/centos ceph-deploy installs.

Use a hiphen as a diliminator instead of a dot.

Signed-off-by: Sandon Van Ness <sandon@inktank.com>
12 years agoceph_manager: wait 20s before wait_for_clean in test_map_discontinuity
Samuel Just [Tue, 30 Jul 2013 17:44:53 +0000 (10:44 -0700)]
ceph_manager: wait 20s before wait_for_clean in test_map_discontinuity

Fixes: #5806
Signed-off-by: Samuel Just <sam.just@inktank.com>
12 years agoRemove now unnecessary try/except/raise that came about as a result of
Warren Usui [Tue, 30 Jul 2013 16:50:20 +0000 (09:50 -0700)]
Remove now unnecessary try/except/raise that came about as a result of
the previous check-in.

Fixes: #5803
Signed-off-by: Warren Usui <warren.usui@inktank.com>
12 years agoAlways raise exception if yum install fails. This avoids later confusion when
Warren Usui [Tue, 30 Jul 2013 02:20:43 +0000 (19:20 -0700)]
Always raise exception if yum install fails.  This avoids later confusion when
packages are missing (the old code skipped 'Nothing to do' messages, but these
cases are still errors).

Fixes #5803
Signed-off-by: Warren Usui <warren.usui@inktank.com>
Reviewed by: Sandon Van Ness

12 years agoMore changes for creating vms manually with lock (no config)
Sandon Van Ness [Tue, 30 Jul 2013 00:03:30 +0000 (17:03 -0700)]
More changes for creating vms manually with lock (no config)

Needed some more changes to allow for the case of creating vm's
manually with teuthology-lock instead of letting teuthology handle
it in internal.py with lock_machines(). Just some additional checks
to go to defaults when ctx.config is non-existent (causes an
attributeerror).

Signed-off-by: Sandon Van Ness <sandon@inktank.com>
Reviewed-by: Warren Usui <warren.usui@inktank.com>
12 years agoFixing teuthology-lock for os-type instead of vm-type.
Sandon Van Ness [Mon, 29 Jul 2013 23:30:40 +0000 (16:30 -0700)]
Fixing teuthology-lock for os-type instead of vm-type.

Teuthology got updated to use --os-type and os_type in yaml
instead of --vm-type. I added this to teuthology but forgot
to update tuthology-lock as well for manually creating vms.

Signed-off-by: Sandon Van Ness <sandon@inktank.com>
12 years agorados_util: adjust-ulimits, not enable-coredump
Sage Weil [Mon, 29 Jul 2013 21:06:34 +0000 (14:06 -0700)]
rados_util: adjust-ulimits, not enable-coredump

This was responsible for the lost_unfound test breakage.

Signed-off-by: Sage Weil <sage@inktank.com>
12 years agoceph_manager: wait for all_up in test_map_discontinuity
Samuel Just [Mon, 29 Jul 2013 19:58:28 +0000 (12:58 -0700)]
ceph_manager: wait for all_up in test_map_discontinuity

Signed-off-by: Samuel Just <sam.just@inktank.com>
12 years agoceph_manager: allow-experimental-feature now causes an EINVAL
Samuel Just [Mon, 29 Jul 2013 19:12:24 +0000 (12:12 -0700)]
ceph_manager: allow-experimental-feature now causes an EINVAL

Signed-off-by: Samuel Just <sam.just@inktank.com>
12 years agos3tests: clone correct branch
Yehuda Sadeh [Thu, 25 Jul 2013 23:47:34 +0000 (16:47 -0700)]
s3tests: clone correct branch

Signed-off-by: Yehuda Sadeh <yehuda@inktank.com>
12 years agoMerge pull request #20 from ceph/double-requirements
Zack Cerza [Mon, 29 Jul 2013 15:15:43 +0000 (08:15 -0700)]
Merge pull request #20 from ceph/double-requirements

fix double requirements issue

12 years agofix double requirements issue
Alfredo Deza [Mon, 29 Jul 2013 15:09:41 +0000 (11:09 -0400)]
fix double requirements issue

Signed-off-by: Alfredo Deza <alfredo@deza.pe>
12 years agoadjust-ulimits: 16k open files
Sage Weil [Sun, 28 Jul 2013 22:09:43 +0000 (15:09 -0700)]
adjust-ulimits: 16k open files

Signed-off-by: Sage Weil <sage@inktank.com>
12 years agoceph_manager, dump_stuck: fix injectargs args
Sage Weil [Sun, 28 Jul 2013 00:41:51 +0000 (17:41 -0700)]
ceph_manager, dump_stuck: fix injectargs args

Signed-off-by: Sage Weil <sage@inktank.com>
12 years agoradosgw-admin: fix typo in import
Josh Durgin [Sat, 27 Jul 2013 19:30:18 +0000 (12:30 -0700)]
radosgw-admin: fix typo in import

Signed-off-by: Josh Durgin <josh.durgin@inktank.com>
12 years agovalgrind: suppress curl, gnutlss, libfcgi leaks
Sage Weil [Sat, 27 Jul 2013 14:49:11 +0000 (07:49 -0700)]
valgrind: suppress curl, gnutlss, libfcgi leaks

Signed-off-by: Sage Weil <sage@inktank.com>
12 years agoceph_manager: don't mark out an out osd on kill_osd(..., mark_out=True)
Sage Weil [Sat, 27 Jul 2013 14:48:55 +0000 (07:48 -0700)]
ceph_manager: don't mark out an out osd on kill_osd(..., mark_out=True)

Fixes

2013-07-27T02:02:23.879 INFO:teuthology.task.thrashosds.thrasher:Recovered, killing an osd
2013-07-27T02:02:23.879 INFO:teuthology.task.thrashosds.thrasher:Killing osd 2, live_osds are [3, 4, 0, 1, 5, 2]
2013-07-27T02:02:24.547 INFO:teuthology.task.thrashosds.thrasher:Removing osd 2, in_osds are: [4, 0, 1, 5]
...
ValueError: list.remove(x): x not in list

Signed-off-by: Sage Weil <sage@inktank.com>
12 years agorest-api: rename
Sage Weil [Sat, 27 Jul 2013 01:21:21 +0000 (18:21 -0700)]
rest-api: rename

Signed-off-by: Sage Weil <sage@inktank.com>
12 years agoAdd rest_api, a new task that starts up /usr/bin/ceph-rest-api running
Warren Usui [Sat, 27 Jul 2013 00:52:02 +0000 (17:52 -0700)]
Add rest_api, a new task that starts up /usr/bin/ceph-rest-api running
as a daemon.

Fixes: 5613
Signed-off-by: Warren Usui <warren.usui@inktank.com>
12 years agoAllow OS version over-ride (distro version)
Sandon Van Ness [Fri, 26 Jul 2013 23:15:19 +0000 (16:15 -0700)]
Allow OS version over-ride (distro version)

Read os_version from yaml to over-ride teuthology defaults.

Signed-off-by: Sandon Van Ness <sandon@inktank.com>
12 years agoMerge pull request #19 from alfredodeza/master
Sage Weil [Fri, 26 Jul 2013 22:09:31 +0000 (15:09 -0700)]
Merge pull request #19 from alfredodeza/master

Top Level Cleanup

12 years agorgw: correct socket option name
Josh Durgin [Fri, 26 Jul 2013 21:18:24 +0000 (14:18 -0700)]
rgw: correct socket option name

Signed-off-by: Josh Durgin <josh.durgin@inktank.com>
12 years agorgw: add rgw log socket to daemonized radosgw too
Josh Durgin [Fri, 26 Jul 2013 21:15:28 +0000 (14:15 -0700)]
rgw: add rgw log socket to daemonized radosgw too

Signed-off-by: Josh Durgin <josh.durgin@inktank.com>
12 years agorgw: pass socket path directly to radosgw
Josh Durgin [Fri, 26 Jul 2013 21:11:01 +0000 (14:11 -0700)]
rgw: pass socket path directly to radosgw

Only radosgw needs this option, and each one will be different, so
remove it from the ceph.conf template.

Signed-off-by: Josh Durgin <josh.durgin@inktank.com>
12 years agoceph_manager: add test_map_discontinuity to thrasher
Samuel Just [Fri, 26 Jul 2013 02:43:08 +0000 (19:43 -0700)]
ceph_manager: add test_map_discontinuity to thrasher

Signed-off-by: Samuel Just <sam.just@inktank.com>
12 years agosamba: run lsof and fuser after shutdown
Greg Farnum [Thu, 25 Jul 2013 18:04:50 +0000 (11:04 -0700)]
samba: run lsof and fuser after shutdown

The clients are pretty regularly reporting busy on unmount when
samba runs above them. This will hopefully give us some info about why.

Signed-off-by: Greg Farnum <greg@inktank.com>
12 years agoMerge branch 'master' of github.com:ceph/teuthology
Sandon Van Ness [Fri, 26 Jul 2013 02:50:39 +0000 (19:50 -0700)]
Merge branch 'master' of github.com:ceph/teuthology

12 years agoMerge remote-tracking branch 'origin/wip-sandon-vm'
Sandon Van Ness [Fri, 26 Jul 2013 02:50:02 +0000 (19:50 -0700)]
Merge remote-tracking branch 'origin/wip-sandon-vm'

Conflicts:
teuthology/lock.py
teuthology/misc.py
teuthology/task/install.py

12 years agoadded test_mon_destroy option to ceph-deploy task
tamil [Fri, 26 Jul 2013 00:12:52 +0000 (17:12 -0700)]
added test_mon_destroy option to ceph-deploy task

Signed-off-by: tamil <tamil.muthamizhan@inktank.com>
12 years agoAdded --os-type argument to run.py
Sandon Van Ness [Thu, 25 Jul 2013 22:33:11 +0000 (15:33 -0700)]
Added --os-type argument to run.py

Because for some reason it was never there before?

Signed-off-by: Sandon Van Ness <sandon@inktank.com>
12 years agoAdded get_distro() to misc.py
Sandon Van Ness [Thu, 25 Jul 2013 21:45:02 +0000 (14:45 -0700)]
Added get_distro() to misc.py

Since getting the ostype is used multiple places I made a
function for it and modified the existing code to use
said function. I also added tests for the function.

Signed-off-by: Sandon Van Ness <sandon@inktank.com>
12 years agoFix priority so it only is added once.
Sandon Van Ness [Thu, 25 Jul 2013 21:24:43 +0000 (14:24 -0700)]
Fix priority so it only is added once.

Needed for baremetal so the sed doesn't keep adding
the priority line in the repos file.

Signed-off-by: Sandon Van Ness <sandon@inktank.com>
12 years agoceph_manager: try both new and old tell mon.* syntax
Sage Weil [Thu, 25 Jul 2013 20:54:20 +0000 (13:54 -0700)]
ceph_manager: try both new and old tell mon.* syntax

So we can test old code with new teuthology (e.g., upgrade testing)

Signed-off-by: Sage Weil <sage@inktank.com>
12 years agoschedule_suite: less ceph-deploy client debugging
Sage Weil [Thu, 25 Jul 2013 19:46:04 +0000 (12:46 -0700)]
schedule_suite: less ceph-deploy client debugging

12 years agoschedule_suite.sh: select ceph-dpeloy branch
Sage Weil [Thu, 25 Jul 2013 19:45:52 +0000 (12:45 -0700)]
schedule_suite.sh: select ceph-dpeloy branch

12 years agoschedule_suite.sh: don't need full ms debug on mons
Sage Weil [Thu, 25 Jul 2013 16:10:47 +0000 (09:10 -0700)]
schedule_suite.sh: don't need full ms debug on mons

Signed-off-by: Sage Weil <sage@inktank.com>
12 years agoceph: make restart take both a list of daemons and other arguments
Sage Weil [Wed, 24 Jul 2013 20:14:02 +0000 (13:14 -0700)]
ceph: make restart take both a list of daemons and other arguments

Signed-off-by: Sage Weil <sage@inktank.com>
Reviewed-by: Tamil Muthamizhan <tamil@inktank.com>
12 years agoteuth: rgw task example config update
Joe Buck [Wed, 24 Jul 2013 18:44:53 +0000 (11:44 -0700)]
teuth: rgw task example config update

Due to bug #5716, pools need to start with a '.' at present.
Updating the examples to follow this convention.

Signed-off-by: Joe Buck <jbbuck@gmail.com>
Reviewed-by: Josh Durgin <josh.durgin@inktank.com>
12 years agoteuth: fix issue in cleanup code
Joe Buck [Wed, 24 Jul 2013 19:24:33 +0000 (12:24 -0700)]
teuth: fix issue in cleanup code

The post-yield code in create_dirs needed to
be tweaked to correctly delete the {tdir}/apache
directory (if it exists) on each client.

Signed-off-by: Joe Buck <jbbuck@gmail.com>
Reviewed-by: Josh Durgin <josh.durgin@inktank.com>
12 years agoceph: wait-for-osds-up option for restart
Sage Weil [Tue, 23 Jul 2013 23:55:43 +0000 (16:55 -0700)]
ceph: wait-for-osds-up option for restart

Signed-off-by: Sage Weil <sage@inktank.com>