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/5946fixes#5946
Signed-off-by: Loic Dachary <loic@dachary.org>
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>
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>
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>
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>
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>
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>
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>
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>
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>
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: ).
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>
