suites/marginal:  Add backtrace restart test

Signed-off-by: Sam Lang <sam.lang@inktank.com>

Merge remote-tracking branch 'gh/last'

Conflicts:
suites/fs/singleton/all/cfuse_workunit_suites_blogbench_ceph_deploy.yaml

add fs collection ceph-deploy blogbench test in new singleton suite

Signed-off-by: Greg Farnum <greg@inktank.com>

add fs collection ceph-deploy blogbench test in new singleton suite

Signed-off-by: Greg Farnum <greg@inktank.com>

apparently this should never work on our current configs

Signed-off-by: Greg Farnum <greg@inktank.com>

apparently this should never work on our current configs

Signed-off-by: Greg Farnum <greg@inktank.com>

upgrade: fix up rgw tests a bit

nfs: debug mds

I've seen a run hang on rmdir on shutdown, and want to see why the MDS didn't
reply.

upgrade/rgw: run first s3tests pass using bobtail tests

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

upgrade: fix client ids

upgrade rgw: increase client mount timeout

This should let radosgw restart and connect once the cluster is finished upgrading

Signed-off-by: Josh Durgin <josh.durgin@inktank.com>

Revert "Revert "Specify xfs for osd powercycle testing""

Pushed a fix to teuthology, should work now.

This reverts commit 853e8fdb731e863e4703d86d8852ecdb4160275f.

Revert "Specify xfs for osd powercycle testing"

This is currently broken.

This reverts commit 79abc44205f53d4b3a12e9a0cc9430e91225a564.

rgw: restart radosgw too

Signed-off-by: Josh Durgin <josh.durgin@inktank.com>

Specify xfs for osd powercycle testing

Signed-off-by: Samuel Just <sam.just@inktank.com>

upgrade: fs: ignore 'wrongly marked down'

rbd: dont' test python on bobtail

The workunit will pull the latest and fail

upgrade: restructure rbd tests

- expand matrix
- include branch: bobtail in first set of tests so that we run the right
  version of the test

nfs: debug mds

I've seen a run hang on rmdir on shutdown, and want to see why the MDS didn't
reply.

Merge remote-tracking branch 'gh/next'

upgrade: debug fs jobs

These are hanging; crank up logs to see why.

upgrade: dbench instead of blogbench

blogbench hangs bobtail ceph-fuse in some
cases, it seems.

upgrade: reorganize the basic/rados suite

Use facets instead of duplicating the test content each time.

rgw asdf

upgrade: rgw: restructure collection

- use separate facets
- make final swift use client.1 so it doesn't
collide with a previous run

upgrade: do not start second radosgw

Use the first one.  This verifies bobtail radosgw works against cuttlefish
osds.

upgrade: mount fs with ceph-fuse for fs tests

upgrade: run blogbench against ceph-fuse

Otherwise this runs on the local disk, not touching the ceph cluster.

upgrade: run rados python test on bobtail to avoid polluting cluster with pools

Extra pools from test.sh will make this fail:

2013-04-26T11:06:45.631 INFO:teuthology.task.workunit.client.0.err:test_rados.TestRados.test_list_pools ... FAIL
2013-04-26T11:06:45.633 INFO:teuthology.task.workunit.client.0.err:
2013-04-26T11:06:45.633 INFO:teuthology.task.workunit.client.0.err:======================================================================
2013-04-26T11:06:45.633 INFO:teuthology.task.workunit.client.0.err:FAIL: test_rados.TestRados.test_list_pools
2013-04-26T11:06:45.634 INFO:teuthology.task.workunit.client.0.err:----------------------------------------------------------------------
2013-04-26T11:06:45.634 INFO:teuthology.task.workunit.client.0.err:Traceback (most recent call last):
2013-04-26T11:06:45.634 INFO:teuthology.task.workunit.client.0.err:  File "/usr/lib/python2.7/dist-packages/nose/case.py", line 197, in runTest
2013-04-26T11:06:45.634 INFO:teuthology.task.workunit.client.0.err:    self.test(*self.arg)
2013-04-26T11:06:45.634 INFO:teuthology.task.workunit.client.0.err:  File "/home/ubuntu/cephtest/1527/mnt.0/client.0/tmp/test_rados.py", line 37, in test_list_pools
2013-04-26T11:06:45.634 INFO:teuthology.task.workunit.client.0.err:    eq(set(), self.list_non_default_pools())
2013-04-26T11:06:45.635 INFO:teuthology.task.workunit.client.0.err:  File "/usr/lib/python2.7/dist-packages/nose/tools.py", line 31, in eq_
2013-04-26T11:06:45.635 INFO:teuthology.task.workunit.client.0.err:    assert a == b, msg or "%r != %r" % (a, b)
2013-04-26T11:06:45.635 INFO:teuthology.task.workunit.client.0.err:AssertionError: set([]) != set(['foo.24250.', 'test-rados-api-plana05-22578-27', 'foo.24250', 'foo.23821'])

ceph config data goes in conf, not config

Signed-off-by: Greg Farnum <greg@inktank.com>

ceph config data goes in conf, not config

Signed-off-by: Greg Farnum <greg@inktank.com>

temporarily add cephfs debugging to overrides

Signed-off-by: Greg Farnum <greg@inktank.com>

Revert "turn on debugging for MDS and Client in FS runs"

We want to apply debugging and whitelists, not one or the
other -- whoops!

This reverts commit 60e7fb4152a7f42594d864239d04a084ac576075.

Signed-off-by: Greg Farnum <greg@inktank.com>

temporarily add cephfs debugging to overrides

Signed-off-by: Greg Farnum <greg@inktank.com>

Revert "turn on debugging for MDS and Client in FS runs"

We want to apply debugging and whitelists, not one or the
other -- whoops!
This reverts commit cb1e8ed954c41840f28f5d88fd3fb58961cf309b.

Signed-off-by: Greg Farnum <greg@inktank.com>

remove ext4 from rados thrashing for now

Signed-off-by: Samuel Just <sam.just@inktank.com>

Merge branch 'next'

set 'filestore flush min = 0' for all ffsb jobs

Until we fix #4579

turn on debugging for MDS and Client in FS runs

Signed-off-by: Greg Farnum <greg@inktank.com>

turn on debugging for MDS and Client in FS runs

Signed-off-by: Greg Farnum <greg@inktank.com>

ior-cfuse: remove the binary/ dir that make install creates

Signed-off-by: Greg Farnum <greg@inktank.com>

rgw suite for upgrade task

Signed-off-by: tamil <tamil.muthamizhan@inktank.com>

ceph-deploy qa suites. moved a level up to make it run in the nightlies.

Signed-off-by: tamil <tamil.muthamizhan@inktank.com>

rbd: add qemu-iotests

Put these in the singleton collection since they have to be run as
client.admin, which is possible when the client and monitor are on the
same host.

Signed-off-by: Josh Durgin <josh.durgin@inktank.com>

rbd: add qemu-iotests

Put these in the singleton collection since they have to be run as
client.admin, which is possible when the client and monitor are on the
same host.

Signed-off-by: Josh Durgin <josh.durgin@inktank.com>

prevent osd warnings from failing CephFS tests

Signed-off-by: Greg Farnum <greg@inktank.com>

prevent osd warnings from failing CephFS tests

Signed-off-by: Greg Farnum <greg@inktank.com>

changed "master" to "next"
Signed-off-by: tamil <tamil.muthamizhan@inktank.com>

task yaml for upgrade task

Signed-off-by: tamil <tamil.muthamizhan@inktank.com>

rados: whitelist quota warning in cluster log for pool quota test

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

rados: test mon config keys

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

rados: test pool quotas when testing api

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

marginal: remove incomplete librbd collection

Move disabled qemu + iozone task into rbd suite.

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

marginal: remove partial collection thrash

Incomplete and broken.  The ffsb test is part of the kernel suite now.

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

marginal: add multimds collection

Lots of stuff here, and several things fail with the export thrashing on,
but usually the others pass.

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

rados test suite for upgrade task

Signed-off-by: tamil <tamil.muthamizhan@inktank.com>

rbd test suites for upgrade task

Signed-off-by: tamil <tamil.muthamizhan@inktank.com>

fs test suite for upgrade task

Signed-off-by: tamil <tamil.muthamizhan@inktank.com>

suites for upgrade

Signed-off-by: tamil <tamil.muthamizhan@inktank.com>

Merge branch 'master' of github.com:ceph/ceph-qa-suite

qa suites for upgrade task

Signed-off-by: tamil <tamil.muthamizhan@inktank.com>

use ceph.com/git instead of github

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

radosgw-admin: Adjust garbage collection settings.

Signed-off-by: caleb miles <caleb.miles@inktank.com>

Merge branch 'master' of github.com:ceph/ceph-qa-suite

ceph-deploy qa suite for fs tests

Signed-off-by: tamil <tamil.muthamizhan@inktank.com>

ceph-deploy qa suite for rbd tests

Signed-off-by: tamil <tamil.muthamizhan@inktank.com>

ceph-deploy qa suite for rados

Signed-off-by: tamil <tamil.muthamizhan@inktank.com>

move osd powercycling tests to rados suite

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

fs: drop fuse_use_invalidate_cb: true, since that is now the default

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

task yaml for basic upgrade

Signed-off-by: tamil <tamil.muthamizhan@inktank.com>

task yaml for upgrade tests

Signed-off-by: tamil <tamil.muthamizhan@inktank.com>

task yaml for upgrade test

Signed-off-by: tamil <tamil.muthamizhan@inktank.com>

rados/thrashers: enable split testing

Signed-off-by: Samuel Just <sam.just@inktank.com>

ceph-deploy suites task yaml with roles

Signed-off-by: tamil <tamil.muthamizhan@inktank.com>

fs: add osd and mds message delays to thrash suite

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

fs: move mds thrashing from marginal, add pjd task

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

radosgw-admin: Adjust garbage collection settings.

Signed-off-by: caleb miles <caleb.miles@inktank.com>

marginal: fix mds thrasher tasks

Make the ceph cluster get installed and set up before we start thrashing
the mds.

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

regression: add fs/traceless

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

fs: test traceless replies from mds

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

task for ceph-deploy sample run

Signed-off-by: tamil <tamil.muthamizhan@inktank.com>

marginal/multiclient: /tmp/cephtest -> $TESTDIR

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

marginal/osd_powercycle: restructure collection

Move install and ceph into the powercycle facet, so that it comes before
the osdthrasher.  Remove it from all of the tasks.

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

fs:  Add fuse_use_invalidate_cb option

Enables the fuse_use_invalidate_cb for the
ceph fuse client in the nightlies.  Once these
tests pass reliably, we can flip the boolean
in the config to true.

Signed-off-by: Sam Lang <sam.lang@inktank.com>

test: add hadoop-internal test

Add a test for our branch of the Hadoop code.

Signed-off-by: Joe Buck <jbbuck@gmail.com>
Reviewed-by: Sam Lang <sam.lang@inktank.com>

test: update java tests

Add additional packages to be installed
for the java tests. This is required
  due to recent changes to teuthology.

Signed-off-by: Joe Buck <jbbuck@gmail.com>
Reviewed-by: Sam Lang <sam.lang@inktank.com>

Add rbd locking/fencing test

Signed-off-by: Josh Durgin <josh.durgin@inktank.com>

rgw: no lockdep on radosgw

This currently crashes because of linking problems, see #4253.

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

rados: crank up mon debugging for the mon thrashing tests

At least for now!

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

specify deterministic admin socket path for admin_socket.py tests

Otherwise we don't know the path because the default now includes $pid.

rbd_xfstests: re-enable test 049

Test 049 is now passing, so add it back to the set that runs
nightly.  It only takes 10-15 seconds to run, so it shouldn't
adversely affect the balance of tests run across the three clients.

This resolves:
    http://tracker.ceph.com/issues/4244

Signed-off-by: Alex Elder <elder@inktank.com>

clusters: add mds to fixed-1.yaml

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

rbd/librbd: fix the cache mode facet

This appears to have been broken, which means we weren't actually testing
any of the non-default cache modes (writethrough, none).  Whoops!  We were
getting jobs like

task:
- install: null
- ceph:
    conf:
      client:
        rbd cache: false
tasks:
- chef: null
- clock: null
- install: null
- ceph: null
- rbd_fsx:
    clients:
    - client.0
    ops: 5000

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

add explicit install task before ceph task

rados: fix ceph_test_filejournal test

Assume it is in the $PATH.

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

rbd: use 2 node cluster, mostly.

Not for the collection that includes KVM tests.

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

rgw: use 2 node cluster

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

rados: use 2 instead of 3 machines

Put the client on the same node as some of the osds.  This shouldn't cause
any problems.  It will make timing a bit different, but if anything that
is probably a good thing.

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

add fixed-2 cluster

rados: don't pass ceph.conf path to workloadgen

The workunit tasks sets CEPH_CONF as needed, which the binaries already
understand.