* refs/pull/25052/head:
mgr: lock pg_map too for osd_pool_stats and notify_osdmap
mgr/DaemonServer: use with_osdmap_and_pgmap instead of nested calls
mgr/ClusterState: add with_osdmap_and_pgmap
Reviewed-by: Kefu Chai <kchai@redhat.com>
* add "common/ceph_context.h" header file to CompressionPluginBrotli.cc
* set the brotli v1.0.7 stable version for compressor plugin use.
Signed-off-by: BI SHUN KE <aionshun@livemail.tw>
The dashboard will need to manipulate the remote cluster mon_host and
key attributes. This logic should be re-used between the rbd CLI and
the dashboard.
Signed-off-by: Jason Dillaman <dillaman@redhat.com>
This flag is used in the inventory reporting and available is deemed more
appropriate. Furthermore this fixes a bug where rejected_reasons
accumulated duplicate entries.
Fixes: http://tracker.ceph.com/issues/36701
Signed-off-by: Jan Fajerski <jfajerski@suse.com>
Separate diskprediction local cloud from the diskprediction plugin.
Devicehealth invoke device prediction function related on the global
configuration "device_failure_prediction_mode".
Signed-off-by: Rick Chen <rick.chen@prophetstor.com>
The ServiceDescription object just holds a list of ServiceLocation
objects. Eliminate ServiceLocation and merge it into the
ServiceDescription class. Have describe_service return a list of
ServiceDescriptions.
Suggested-by: Tim Serong <tserong@suse.com>
Signed-off-by: Jeff Layton <jlayton@redhat.com>
There's currently no way to enumerate the services that the
orchestrator backend knows about. Add a "service ls" command
that will list them with some basic info, subject to filters
given on the command line.
Signed-off-by: Jeff Layton <jlayton@redhat.com>
Rework the describe_service operation for rook to allow listing of
services, with optional filters for service type, id and nodename.
Signed-off-by: Jeff Layton <jlayton@redhat.com>
log is not very visible from user's perspective. we'd better return the
error message using the input parameter, if the technique is
unsupported.
also, remove the trailing newline, as `ss` will be sent back to user,
it's the the front-end's responsibility to format the error message.
Signed-off-by: Kefu Chai <kchai@redhat.com>
This reverts commit c5bafae6fd.
in 37f73607 and 3ec41030, we detect invalid "k". and we should always
detect invalid combinations of parameters in
ErasureCodeInterface::init(), and return -EINVAL instead of "autotune" them
as put in the commit message of c5bafae6, if the profile is invalid for
given erasure plugin. so ErasureCodeBench::encode() and
ErasureCodeBench::decode() will bail out with the error message returned by
ErasureCodeInterface::init() if the parameters are not valid.
also, i checked all ErasureCodeInterface::init() implementations, none
of them is overwriting the settings without returning a non-zero return
code. so we are not overriding or autotuning the settings specified by
user.
Fixes: http://tracker.ceph.com/issues/16500
Signed-off-by: Kefu Chai <kchai@redhat.com>
ceph-mgr: hold lock while accessing the request list and submitting request
Reviewed-by: Boris Ranto <branto@redhat.com>
Reviewed-by: Kefu Chai <kchai@redhat.com>
