* refs/pull/32282/head:
cephadm: clean up shell containers on exit
cephadm: do not force container removal on unit stop
Reviewed-by: Sebastian Wagner <swagner@suse.com>
Reviewed-by: Michael Fritch <mfritch@suse.com>
By definition in the orchestrator's docs, we're using `host` insted of `node`
Fixes wrong usage of PlacementSpec in mgr/ansible
Signed-off-by: Sebastian Wagner <sebastian.wagner@suse.com>
Don't use site-packages, since the host might have the same version, and
it won't have the ditaa and breathe.
Require Cython to make the venv sufficient.
Signed-off-by: Sage Weil <sage@redhat.com>
* use the latest Sphinx
* use the latest breathe. also, since there is no need to be compatible
with python2. we can move to 4.14.0 which is python3 only.
Signed-off-by: Kefu Chai <kchai@redhat.com>
The 'rbd bench' command was recently modified to print IEC units
instead of bytes/sec. This broke the handling for QoS throughput
tests since it was incorrectly evaluating the available RBD
throughput. Additionally, the QoS tests should use a "<="
comparison operator since the QoS is the upper-bound limit.
Signed-off-by: Jason Dillaman <dillaman@redhat.com>
stop.sh: do not try to contact mon unless cluster is up
Reviewed-by: Tatjana Dehler <tdehler@suse.com>
Reviewed-by: Kotresh HR <khiremat@redhat.com>
Reviewed-by: David Disseldorp <ddiss@suse.de>
The run_cmd() was already passing --rm; do it for shell_cmd too so that
we don't leave behind a mess from shell commands, ceph-volume, etc.
Signed-off-by: Sage Weil <sage@redhat.com>
The -f arg for 'podman rm' will remove the container without actually
stopping it, leaving the running containers and processes (conmon and
daemon) running and generally making a mess of things.
Doing 'podman rm' without the -f on a running container will complain
that the container is still running.
So, first stop, then remove.
Fixes: f1de706076
Signed-off-by: Sage Weil <sage@redhat.com>
otherwise "ceph mon metadata" will hang, when we try to stop an
already-stopped cluster. and `vstart_runner.py` always calls `stop.sh`
before setting up a cluster.
Signed-off-by: Kefu Chai <kchai@redhat.com>
mgr/dashboard: Fix OSD IDs are not displayed when using cephadm backend
Reviewed-by: Lenz Grimmer <lgrimmer@suse.com>
Reviewed-by: Ernesto Puertat <epuertat@redhat.com>
Reviewed-by: Laura Paduano <lpaduano@suse.com>
To create Ganesha daemons with vstart following packages needs to be installed:
nfs-ganesha nfs-ganesha-ceph nfs-ganesha-rados-grace
Version used for these packages 2.7.6-2 on fedora 30 system.
This is an experimental option, it just deploys specified number of ganesha
daemons.
Signed-off-by: Varsha Rao <varao@redhat.com>
* refs/pull/31884/head:
qa/tasks: Fix the volume ls in test_volume_rm
qa/tasks: tests for 'fs volume create' and 'fs volume ls'
Reviewed-by: Patrick Donnelly <pdonnell@redhat.com>
* refs/pull/29059/head:
mds: complete all the replay op when mds is restarted ,no matter the session is closed or not.
Reviewed-by: Zheng Yan <zyan@redhat.com>
Reviewed-by: Patrick Donnelly <pdonnell@redhat.com>
* refs/pull/31662/head:
mds: fix revoking caps after after stale->resume circle
client: fix incorrect debug message
Reviewed-by: Patrick Donnelly <pdonnell@redhat.com>
* refs/pull/31123/head:
mds : optimization functions,get_dirfrags_under, to speed up processing directories with tens of millions of files
Reviewed-by: Zheng Yan <zyan@redhat.com>
Reviewed-by: Patrick Donnelly <pdonnell@redhat.com>
* refs/pull/31595/head:
mds: optimize function, fragset_t::simplify, to improve the efficiency of merging fragment
Reviewed-by: Zheng Yan <zyan@redhat.com>
Some tests set this to a dynamic value, it'd be helpful to know how long
a test is planning to wait.
Signed-off-by: Patrick Donnelly <pdonnell@redhat.com>
