mgr/dashboard: do not fail on user creation

Accordingly to other Ceph commands do not fail on user
creation with a non-zero error code if the user already
exists.
Instead succeed and return the message 'User <username>
already exists'.

Fixes: https://tracker.ceph.com/issues/44502
Signed-off-by: Tatjana Dehler <tdehler@suse.com>
(cherry picked from commit 9992faa9479b62f05f36352daf89f6e7541f29c9)

Merge PR #34192 into octopus

* refs/pull/34192/head:
mgr/cephadm: add watch_url to NFS bootstrap config
mgr/cephadm: check for an existing NFS config object
mgr/cephadm: combine `get_daemon_name` and `get_rados_user` funcs
mon: allow `auth caps` for `mgr` profile
mgr/cephadm: add deployment example to NFS docs
mgr/cephadm: add utils.py
mgr/cephadm: create/update keyring during nfs config
mgr/cephadm: pass NFSServiceSpec instead of pool/namespace args
mgr/cephadm: add nfs apply test
mgr/cephadm: fix nfs redeploy/reconfig
mgr/cephadm: ensure nfs uses the correct entity
mgr/cephadm: fix service name for nfs type
mgr/cephadm: apply nfs service spec
mgr/cephadm: don't create a unique nfs daemon name
mgr/cephadm: update docs
mgr/cephadm: add minimal caps for nfs client keyring
mgr/cephadm: add logging to NFSGanesha class
mgr/cephadm: create empty nfs rados config obj
mgr/cephadm: add nfs tests
mgr/cephadm: add a minimal ganesha.conf
mgr/cephadm: deploy nfs-ganesha service
mgr/cephadm: split out `config-and-keyrings` json
mgr/orchestrator: rearrange interface to match impl. order

Reviewed-by: Michael Fritch <mfritch@suse.com>

Merge PR #34258 into octopus

* refs/pull/34258/head:
cephadm: Fix trucated output of `mgr dump`

Reviewed-by: Sage Weil <sage@redhat.com>

Merge PR #34199 into octopus

* refs/pull/34199/head:
mgr/rook: Blinking lights

Reviewed-by: Sage Weil <sage@redhat.com>

Merge PR #34204 into octopus

* refs/pull/34204/head:
osd/PeeringState: drop mimic assert

Reviewed-by: Sage Weil <sage@redhat.com>
Reviewed-by: Neha Ojha <nojha@redhat.com>

Merge PR #34178 into octopus

* refs/pull/34178/head:
pybind/mgr/*: fix config_notify handling of default values

Reviewed-by: Kefu Chai <kchai@redhat.com>

Merge PR #34123 into octopus

* refs/pull/34123/head:
osd/PeeringState: fix pending want_acting vs osd offline race

Reviewed-by: Samuel Just <sjust@redhat.com>
Reviewed-by: xie xingguo <xie.xingguo@zte.com.cn>

cephadm: Fix trucated output of `mgr dump`

In case the process has terminated, we
have to make sure to read till EOF, otherwise
we might skip some bytes.

Fixes: https://tracker.ceph.com/issues/44642
Co-authored-by: Tim Serong <tserong@suse.com>
Signed-off-by: Sebastian Wagner <sebastian.wagner@suse.com>
(cherry picked from commit 6a6759eef06c32efe1a10190925da6bc7e5ec4b0)

Merge pull request #34214 from neha-ojha/wip-34183-octopus

octopus: doc/releases/octopus.rst: fix min_mon_release string

Reviewed-by: Josh Durgin <jdurgin@redhat.com>

Merge pull request #34207 from liewegas/bp-34188-octopus

octopus: doc: do not encourage readers to run multiple clusters on same HW

Reviewed-by: Nathan Cutler <ncutler@suse.com>

Merge pull request #34237 from smithfarm/wip-sle-15.2-octopus

octopus: qa/distros: add SLE-15-SP2

Reviewed-by: Thomas Bechtold <tbechtold@suse.com>

doc: fix cephadm link

Fixes: 8fe460e7562b28e007179eeb533ef68a3e99b4c8
Signed-off-by: Nathan Cutler <ncutler@suse.com>
(cherry picked from commit 7dd012252ce3be13d8d44a5c3829198fc72b5754)

qa/distros: add SLE-15-SP2

Ceph octopus is known to run on SLE-15-SP2 so add it to
qa/distros/all.

Signed-off-by: Thomas Bechtold <tbechtold@suse.com>
(cherry picked from commit 3261e24636474a3e274c65b859e9325299470f11)

doc/releases/octopus.rst: fix min_mon_release string

Signed-off-by: Neha Ojha <nojha@redhat.com>
(cherry picked from commit c1c6f8f8c191d206659116bbc04695999478b450)

Merge PR #34194 into octopus

* refs/pull/34194/head:
CMakeLists: 15.2.0
ceph_release: stable, not rc

Reviewed-by: Neha Ojha <nojha@redhat.com>

Merge PR #33939 into octopus

* refs/pull/33939/head:
mgr/dashboard: NoRebalance flag is added to the Dashboard

Reviewed-by: Lenz Grimmer <lgrimmer@suse.com>
Reviewed-by: Ernesto Puerta <epuertat@redhat.com>
Reviewed-by: Volker Theile <vtheile@suse.com>
Reviewed-by: Stephan Müller <smueller@suse.com>

Merge PR #34163 into octopus

* refs/pull/34163/head:
mgr: ceph_pg_* metrics contains last value instead of sum all of them

Reviewed-by: Kefu Chai <kchai@redhat.com>

Merge PR #34185 into octopus

* refs/pull/34185/head:
mgr/cephadm: DriveGroupSpec needs to know about unmanaged attribute
mgr/cephadm: DriveGroupSpec needs to support/ignore _unmanaged_

Reviewed-by: Sebastian Wagner <swagner@suse.com>
Reviewed-by: Joshua Schmid <jschmid@suse.de>

Merge PR #34186 into octopus

* refs/pull/34186/head:
mgr/cephadm: use docker.io in default container image name
mgr/cephadm: fix 'upgrade start' message when specifying a version

Reviewed-by: Michael Fritch <mfritch@suse.com>

Merge PR #34190 into octopus

* refs/pull/34190/head:
mgr/cephadm: fix 'orch ps --refresh'

Reviewed-by: Michael Fritch <mfritch@suse.com>

Merge PR #34191 into octopus

* refs/pull/34191/head:
mgr/cephadm: bind grafana to all interfaces

Reviewed-by: Michael Fritch <mfritch@suse.com>

doc: do not encourage readers to run multiple clusters on same HW

This material was first added on December 22, 2012 by
46d0334456c14da2bc8e012f121c545b1b33b95f

Signed-off-by: Nathan Cutler <ncutler@suse.com>
(cherry picked from commit 8fe460e7562b28e007179eeb533ef68a3e99b4c8)

Merge pull request #34176 from theanalyst/backports/oct-distro-fix

octopus: doc: releases: add a note on debian packages & those built by the community

Reviewed-By: Sage Weil <sweil@redhat.com>

osd/PeeringState: drop mimic assert

This code can still execute for a cluster that has require_osd_release >=
nautilus because a PG is processing an older OSDMap.  Drop this assert
since the osd boot checks on require_osd_release are sufficient.

Fixes: https://tracker.ceph.com/issues/44759
Signed-off-by: Sage Weil <sage@redhat.com>
(cherry picked from commit 29b7469fb7196f0ce7b2ae5d0a21226ab15ba306)

mgr/rook: Blinking lights

Blinking lights implementation

Signed-off-by: Juan Miguel Olmo Martínez <jolmomar@redhat.com>
(cherry picked from commit a35c1b25933ba5583e9c6b8752d2949462f92025)

CMakeLists: 15.2.0

I'm not actually sure if this is used for much of anything...

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

ceph_release: stable, not rc

This should have happened before v15.2.0.. oops!

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

mgr/cephadm: add watch_url to NFS bootstrap config

watch for changes to the per-daemon RADOS conf file

Signed-off-by: Michael Fritch <mfritch@suse.com>
(cherry picked from commit ff9265104f69c702630e5069b1e1a2105d1f74fa)

mgr/cephadm: check for an existing NFS config object

assume an existing NFS config if the named rados object already exists

Signed-off-by: Michael Fritch <mfritch@suse.com>
(cherry picked from commit 75b988f314ab8792cccad70bb9894ff1a50ae3f3)

mgr/cephadm: combine `get_daemon_name` and `get_rados_user` funcs

Signed-off-by: Michael Fritch <mfritch@suse.com>
(cherry picked from commit fdf90bf530edbb9608fb8a5d273f7ce38147f534)

mon: allow `auth caps` for `mgr` profile

cephadm orchestrator needs to update the caps during reconfig/redeploy

Signed-off-by: Michael Fritch <mfritch@suse.com>
(cherry picked from commit ef8a46a3b13ea6753b8c03121da4c6daff370fd2)

mgr/cephadm: add deployment example to NFS docs

Signed-off-by: Michael Fritch <mfritch@suse.com>
(cherry picked from commit 77440d91c1be7c872fc91042a0d723b9198f224f)

mgr/cephadm: add utils.py

move `name_to_config_section` and `assert_valid_host` into utils.py

Signed-off-by: Michael Fritch <mfritch@suse.com>
(cherry picked from commit 322eea8099616b25a091ac66a1253d1991a0da2c)

mgr/cephadm: create/update keyring during nfs config

the keyring might already exist from a prior config/reconfig/redeploy.

attempt to get_or_create the keyring first and than update the keyring
caps afterward

Signed-off-by: Michael Fritch <mfritch@suse.com>
(cherry picked from commit 084fd4a91ae781ab7bed9e06ceca17b1a5ca5be2)

mgr/cephadm: pass NFSServiceSpec instead of pool/namespace args

Signed-off-by: Michael Fritch <mfritch@suse.com>
(cherry picked from commit 8bb79a5f275ecf79428dc35a8fc1a76c36569df3)

mgr/cephadm: add nfs apply test

Signed-off-by: Michael Fritch <mfritch@suse.com>
(cherry picked from commit 78ce6d431806c1f3e8e55e946eff64d450821db1)

mgr/cephadm: fix nfs redeploy/reconfig

search for the `NFSServiceSpec` instead of passing empty
`cephadm_args` during `_create_daemon`

Signed-off-by: Michael Fritch <mfritch@suse.com>
(cherry picked from commit 107885801bb5d7401dead52cbc2abadaa2865a91)

mgr/cephadm: ensure nfs uses the correct entity

mon entity needs to start with `client.`

Signed-off-by: Michael Fritch <mfritch@suse.com>
(cherry picked from commit fd6322f7ec51f3dc57168bbd08ffab3c832a4598)

mgr/cephadm: fix service name for nfs type

the `nfs` service was incorrectly reported as `mds`

Signed-off-by: Michael Fritch <mfritch@suse.com>
(cherry picked from commit d9f04ec1d73948e915dda0f51592306542314238)

mgr/cephadm: apply nfs service spec

orch apply nfs <svc_id> [<placement>]

Signed-off-by: Michael Fritch <mfritch@suse.com>
(cherry picked from commit 76f27894caba2b3bc146c79d86c4335c9c33dcde)

mgr/cephadm: don't create a unique nfs daemon name

only one nfs daemon will be deployed per host

Signed-off-by: Michael Fritch <mfritch@suse.com>
(cherry picked from commit 81823176af101864524fc611fff346bf9798f0e0)

mgr/cephadm: update docs

Signed-off-by: Michael Fritch <mfritch@suse.com>
(cherry picked from commit 530c62d76d79c4dd90a966302cd07d1017261686)

mgr/cephadm: add minimal caps for nfs client keyring

Signed-off-by: Michael Fritch <mfritch@suse.com>
(cherry picked from commit 94600c69e2f547c02017204c909d11d626649066)

mgr/cephadm: add logging to NFSGanesha class

Signed-off-by: Michael Fritch <mfritch@suse.com>
(cherry picked from commit 26ea33ad4b0e624345d34526dcf04e35176d2c8e)

mgr/cephadm: create empty nfs rados config obj

Signed-off-by: Michael Fritch <mfritch@suse.com>
(cherry picked from commit 9103f4758beb7e0f0dba523f776f9918c0dbfe6d)

mgr/cephadm: add nfs tests

Signed-off-by: Michael Fritch <mfritch@suse.com>
(cherry picked from commit b2b6e016d190d4a5dacfc1a4f27adf5d2d7a7aa0)

mgr/cephadm: add a minimal ganesha.conf

we only need to specify the rados pool and config object url

Signed-off-by: Michael Fritch <mfritch@suse.com>
(cherry picked from commit 2606cc3543bba8afc67e8f78185286123fd71744)

mgr/cephadm: deploy nfs-ganesha service

orch daemon add nfs <svc_arg> <pool> [<namespace>] [<placement>]

Signed-off-by: Michael Fritch <mfritch@suse.com>
(cherry picked from commit 770816dc9f5b440fd1fc6cc507a06cd8573f6ea1)

mgr/cephadm: split out `config-and-keyrings` json

Signed-off-by: Michael Fritch <mfritch@suse.com>
(cherry picked from commit 3be608102834bbc7ba5de927ccc868b5b5fb3cad)

mgr/orchestrator: rearrange interface to match impl. order

rgw impl. happens before rbd impl

Signed-off-by: Michael Fritch <mfritch@suse.com>
(cherry picked from commit 7976ac745aa83ac664f2628fb9fe7bbec0c07100)

mgr/cephadm: bind grafana to all interfaces

Fixes: https://tracker.ceph.com/issues/44608
Signed-off-by: Sage Weil <sage@redhat.com>
(cherry picked from commit 5f4fedbd5beeccc4792241bcfd2b81bdeb4c7d57)

mgr/cephadm: fix 'orch ps --refresh'

The use of host in the refresh loop was clobbering the argument value.

Fixes: https://tracker.ceph.com/issues/44513
Signed-off-by: Sage Weil <sage@redhat.com>
(cherry picked from commit 764ce71bd202315fdf4403d9e1e8ae55ef849a90)

Merge PR #34154 into octopus

* refs/pull/34154/head:
mgr/cephadm: point alertmanager at all mgr/dashboard URLs

Reviewed-by: Michael Fritch <mfritch@suse.com>

Merge PR #34155 into octopus

* refs/pull/34155/head:
debian: add ceph-grafana-dashboards package
ceph.spec: put prometheus alerts in vendor-neutral location
mgr/cephadm: include prom alerts, if present in the container

Reviewed-by: Michael Fritch <mfritch@suse.com>

Merge PR #34157 into octopus

* refs/pull/34157/head:
Add alertmanager adopt

Reviewed-by: Michael Fritch <mfritch@suse.com>

Merge PR #34158 into octopus

* refs/pull/34158/head:
Add alertmanager sample

Reviewed-by: Michael Fritch <mfritch@suse.com>

Merge PR #34160 into octopus

* refs/pull/34160/head:
mgr/orch: allow list daemons by service_name

Reviewed-by: Michael Fritch <mfritch@suse.com>

Merge PR #34153 into octopus

* refs/pull/34153/head:
mgr/cephadm: Add example to run when debugging ssh failures

Reviewed-by: Sebastian Wagner <swagner@suse.com>

Merge PR #34156 into octopus

* refs/pull/34156/head:
cephadm: init config and keyring with None
cephadm: allow config and keyring args for ceph-volume
qa/workunits/cephadm: create lvs before calling cephadm
qa/workunits/cephadm/test_cephadm.sh: move osd test to ceph-volume
qa/workunits/cephadm/test_cephadm.sh: fix mon.b failure

Reviewed-by: Michael Fritch <mfritch@suse.com>

mgr/cephadm: use docker.io in default container image name

Not all hosts have docker.io configured as a default registry.

Signed-off-by: Sage Weil <sage@redhat.com>
(cherry picked from commit d4f367a7ea96c97c7aa437f79a1c2dabb2c7caf8)

mgr/cephadm: fix 'upgrade start' message when specifying a version

Signed-off-by: Sage Weil <sage@redhat.com>
(cherry picked from commit c6b0c28d748f7eec1455d00be15ab7f0b985ecf5)

mgr/cephadm: DriveGroupSpec needs to know about unmanaged attribute

Signed-off-by: Joshua Schmid <jschmid@suse.de>
(cherry picked from commit f7bb0236aa601e5439875a97c2250be20956a4e7)

mgr/cephadm: DriveGroupSpec needs to support/ignore _unmanaged_

Signed-off-by: Joshua Schmid <jschmid@suse.de>
(cherry picked from commit bda6a47aac0d5db6a5b07c50e7ac45c4682a9d0d)

cephadm: init config and keyring with None

and we should not assume that both `config` and `keying` are specified
when calling this method. because, for instance, `create_daemon_dirs()`
does handle the case where `config` and/or `keyring` is not specified.

this is a follow-up fix of 245d6a5cec9cc0f299613b8cc0415e494a4c3ac5

Signed-off-by: Kefu Chai <kchai@redhat.com>
Signed-off-by: Michael Fritch <mfritch@suse.com>
(cherry picked from commit 86de901d89b26920c1493e3867ddfafa9f9c95cd)

cephadm: allow config and keyring args for ceph-volume

easier to script via the cephadm tests

Signed-off-by: Michael Fritch <mfritch@suse.com>
(cherry picked from commit bac530f06a60d96259376cd19224bdbffce1bdc4)

pybind/mgr/*: fix config_notify handling of default values

After 5ace82e65c72847fb875fc01c419937a26a59d70 was merged, I found
three more instances of the code being patched.

The commit message of 5ace82e65c72847fb875fc01c419937a26a59d70 was/is:

"The default values are handled by mgr_module.py's _get_module_option();
the or here means that we break any non-true (0, false, none) value and
override it with the default."

Fixes: 5ace82e65c72847fb875fc01c419937a26a59d70
Fixes: https://tracker.ceph.com/issues/43746
Signed-off-by: Nathan Cutler <ncutler@suse.com>
(cherry picked from commit 30585f44de6147c824ed2df1a477ad17f7b51fff)

doc: releases: add a note on debian packages & those built by the community

Signed-off-by: Abhishek Lekshmanan <abhishek@suse.com>
(cherry picked from commit 2cdca1ff6def0da2b6e6c67b276eaf964b02a010)

mgr: ceph_pg_* metrics contains last value instead of sum all of them

During evaluation of pool stats metrics contains last reported value instead of sum

Fixes: https://tracker.ceph.com/issues/44590
Signed-off-by: Jacek Suchenia <jacek.suchenia@gmail.com>
(cherry picked from commit ad4d790a35af483f3a240f247cda5754301a9199)

mgr/orch: allow list daemons by service_name

Services like rgw and mds are differentiated by service_name. For
example: mds.xyz vs. mds.abc. With current interface, we can't list all
daemons belonged to mds.xyz only. Add service_name as a new argument to
filter daemons by it.

Fixes: https://tracker.ceph.com/issues/44541
Signed-off-by: Kiefer Chang <kiefer.chang@suse.com>
(cherry picked from commit 079c9fd7718ceb17660da61d46019c53012e0f0c)

Add alertmanager sample

Default contents of alertmanager.yml with additional peers key

Signed-off-by: Eric Jackson <swiftgist@gmail.com>
(cherry picked from commit 5cbe0c4f8fc9b7ca24a140643237e13df6851463)

Add alertmanager adopt

Signed-off-by: Eric Jackson <swiftgist@gmail.com>
(cherry picked from commit 354d9d2d58b3991c73eb0fb9b58fcfaf9436afba)

qa/workunits/cephadm: create lvs before calling cephadm

Fixes: https://tracker.ceph.com/issues/44682
Signed-off-by: Jan Fajerski <jfajerski@suse.com>
(cherry picked from commit ed0db2445f0de91b2ce53fd133cc8381cb234d1e)

qa/workunits/cephadm/test_cephadm.sh: move osd test to ceph-volume

instead of via the orchestrator

Signed-off-by: Michael Fritch <mfritch@suse.com>
(cherry picked from commit 26b7a3f8192abbd20563bd8280f17744b64f79e8)

qa/workunits/cephadm/test_cephadm.sh: fix mon.b failure

and is_available check to ensure quorum

Signed-off-by: Michael Fritch <mfritch@suse.com>
(cherry picked from commit 90cb2feec3165817f5041a6bdc26aa0073a90e00)

debian: add ceph-grafana-dashboards package

Signed-off-by: Sage Weil <sage@redhat.com>
(cherry picked from commit 01015f8ca320e2eab089741e5a8871a4d78c2fc2)

ceph.spec: put prometheus alerts in vendor-neutral location

Signed-off-by: Sage Weil <sage@redhat.com>
(cherry picked from commit f1b516d1bf44f2236c22d5a36f0363f4768e4a6b)

mgr/cephadm: include prom alerts, if present in the container

The prometheus config already looks in/etc/prometheus/alerting/*, so drop
this file into position.

Note that the file is currently only present in SUSE builds, so that needs
to be fixed still.

Signed-off-by: Sage Weil <sage@redhat.com>
(cherry picked from commit b1463998bddb67973b59930cc73901143d0961f8)

mgr/cephadm: point alertmanager at all mgr/dashboard URLs

Signed-off-by: Sage Weil <sage@redhat.com>
(cherry picked from commit 41d0966b72b52a78aa63cb8776b07d23668526d9)

mgr/cephadm: Add example to run when debugging ssh failures

```
$ ceph orch host add foobar
Error ENOENT: Failed to connect to foobar (foobar).  Check that the host is reachable and accepts connections using the cephadm SSH key
you may want to run:
> ssh -F =(ceph cephadm get-ssh-config) -i =(ceph config-key get mgr/cephadm/ssh_identity_key) rook@foobar
$  ssh -F =(ceph cephadm get-ssh-config) -i =(ceph config-key get mgr/cephadm/ssh_identity_key) rook@foobar
ssh: Could not resolve hostname foobar: Temporary failure in name resolution
```

Signed-off-by: Sebastian Wagner <sebastian.wagner@suse.com>
(cherry picked from commit c840dd2204e9a419c28875ace18d33ad6f87500f)

Merge pull request #34125 from smithfarm/wip-cbs-deprecate-octopus

octopus: ceph-backport.sh: add deprecation warning

Reviewed-by: Jan Fajerski <jfajerski@suse.com>
Reviewed-by: Mykola Golub <mgolub@suse.com>

Merge pull request #34137 from theanalyst/octopus-changelog-1

octopus: doc: add changelog for octopus release

Reviewed-by: Sage Weil <sage@redhat.com>
Reviewed-by: Nathan Cutler <ncutler@suse.com>

doc: add changelog for octopus release

Signed-off-by: Abhishek Lekshmanan <abhishek@suse.com>
(cherry picked from commit dc7cf3fe8986de7b3e245ea04084eef51c7a0cdf)

Merge PR #34100 into octopus

* refs/pull/34100/head:
doc/cephadm: notes on status/stability
doc/cephadm: some troubleshooting tips

Reviewed-by: Sebastian Wagner <swagner@suse.com>

Merge PR #34115 into octopus

* refs/pull/34115/head:
doc/releases/octopus: drop stray line
doc/releases/octopus: note about repository locations
doc/releases: include octopus in index
doc/install/get-packages: update package install instructions
doc/releases/octopus: final notes

Reviewed-by: Abhishek Lekshmanan <abhishek.lekshmanan@gmail.com>

doc/releases/octopus: drop stray line

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

doc/releases/octopus: note about repository locations

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

doc/releases: include octopus in index

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

15.2.0

doc/install/get-packages: update package install instructions

- include cephadm method
- update to include new version URLs

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

Merge PR #34119 into octopus

* refs/pull/34119/head:
doc: update octopus release notes for rgw

Reviewed-by: Abhishek Lekshmanan <abhishek.lekshmanan@gmail.com>

Merge PR #34122 into octopus

* refs/pull/34122/head:
cephadm: default to v15 tag for octopus

Reviewed-by: Michael Fritch <mfritch@suse.com>

ceph-backport.sh: add deprecation warning

The script is not maintained in the stable versions of Ceph.

This commit is not a cherry-pick from master, because the script is maintained
in master. It is unmaintained only in the stable versions, like
"octopus".

Fixes: https://tracker.ceph.com/issues/42194
Signed-off-by: Nathan Cutler <ncutler@suse.com>
(manual cherry pick of commit 365a93bd3fe62369d33d07993dd246e9b4dd63fd)

Conflicts:
    src/script/ceph-backport.sh

osd/PeeringState: fix pending want_acting vs osd offline race

In general there are two scenarios we might call choose_acting to
post a pending want_change change:

1) we are in the middle of peering, and we decide to select some
  peers other than current acting set in order to continue serving
  client reads and writes.
  In this case, when any OSD from the pending want_acting set goes down,
  primary will restart peering process and tidy want_acting up properly
  (see PeeringState::Primary::exit()).

2) PG is active, and we want to transit all successfully backfilled
  (or async-recovered) peers back into up set.
  In this case, any want_acting member is deemed to be either coming from
  the current up set or acting set (as we pass restrict_to_up_acting == true
  when calling down into choose_acting).

From 1, we know we'd never leak a want_acting set that might
contain stray peers into Active state. From 2, we know that assert would
effectively catch any potential bad Active choose_acting callers without
setting restrict_to_up_acting properly.

However, in 023524a I did introduce a third scenario that might be
against rule 2 — we now call choose_acting with restrict_to_up_acting
option off on any stray peer coming back to life when PG is active,
and if that peer is down (again) and the corresponding pg_temp change
is still in-flight, then we would reliably fire the assert.

Fix by calling choose_acting again whenever Active sees a new map
that marks down an stray osd in want_acting, so we don't leave
a dirty want_acting (and pg_temp) there.

Fixes: https://tracker.ceph.com/issues/44507
Signed-off-by: xie xingguo <xie.xingguo@zte.com.cn>
(cherry picked from commit 155c44bfe068f749dfb186df47d954f402bf903e)

cephadm: default to v15 tag for octopus

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

doc: update octopus release notes for rgw

Signed-off-by: Casey Bodley <cbodley@redhat.com>
(cherry picked from commit 984f3ff7e0fe10f1b54da4271a2a0879ffea0ddb)

doc/releases/octopus: final notes

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

doc/cephadm: notes on status/stability

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

Merge PR #34042 into octopus

* refs/pull/34042/head:
mgr/rook: list rgw services
mgr/rook: tolerate timestamps that are None
mgr/orch: add 'subcluster' property to RGWSpec
mgr/rook: do not create radosgw pools
mgr/rook: refactor apply/add for rgw
mgr/cephadm: configure rgw_frontends for rgw service
mgr/orch: accept port and ssl flags to 'apply rgw'
python-common/ceph/deployment/service_spec: add ssl to RGWSpec
mgr/rook: fix 'orch ps' for osds

Reviewed-by: Yehuda Sadeh <yehuda@redhat.com>

Merge PR #33959 into octopus

* refs/pull/33959/head:
qa: fix all the fsx.sh-invoking yaml files to install dependencies

Reviewed-by: Sage Weil <sage@redhat.com>

Merge PR #34067 into octopus

* refs/pull/34067/head:
mgr/DaemonServer: add explicit check that acting matches for merge

Reviewed-by: Neha Ojha <nojha@redhat.com>
Reviewed-by: xie xingguo <xie.xingguo@zte.com.cn>