rbd-nbd: fix ebusy when do map

When doing rbd-nbd map, if the Ceph service is not available,
the codes will wait on rados.connect(), unless killing the process.
In that case, the close_nbd logic is skipped with NBD_CLEAR_SOCK ioctl
not called. On the CentOS 7 kernel, it leaves nbd->file not cleared, which
causes the subsequent map requests return EBUSY, this patch fixes it
by connecting Ceph first prior to calling NBD_SET_SOCK ioctl

Fixes: http://tracker.ceph.com/issues/23528
Signed-off-by: Li Wang <laurence.liwang@gmail.com>
(cherry picked from commit ab77dcc0170c0d63795fe0d50427cda630bfd593)

Merge pull request #21177 from tchaikov/wip-luminous-pr-21159

luminous: src: s/--use-wheel//

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

src: s/--use-wheel//

--use-wheel was deprecated in favor of --only-binary in pip v7.0.0. and
--use-wheel was removed in a recent release of pip. but some packages
are source packages, so we cannot simply replace use-wheel with
only-binary. so a simpler approach is to drop --use-wheel option, as pip
respects --find-links, and will find the required package from the
wheelhouse.

Signed-off-by: Kefu Chai <kchai@redhat.com>
(cherry picked from commit a80cce4ba6afee921a7c6adce802c91e30f8c0a9)

Conflicts:
src/ceph-detect-init/CMakeLists.txt
src/ceph-disk/CMakeLists.txt: trivial resolution
src/pybind/mgr/dashboard/CMakeLists.txt: dashboard2 is not
in luminous, so drop this change.

Merge pull request #20254 from k0ste/luminous_backports

luminous: mgr: prometheus: set metadata metrics value to '1' (#22717).

Reviewed-by: John Spray <john.spray@redhat.com>

Merge pull request #20334 from tchaikov/wip-pr-20255-luminous

luminous: config: Change bluestore_cache_kv_max to type INT64

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

Merge pull request #20358 from pdvian/wip-22940-luminous

luminous: Double free in rados_getxattrs_next

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

Merge pull request #20359 from xiexingguo/wip-luminous-balancer-fixes

luminous: mgr/balancer: pool-specific optimization support and bug fixes

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

Merge pull request #20399 from smithfarm/wip-22942-luminous

luminous: ceph osd force-create-pg cause all ceph-mon to crash and unable to come up again

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

Merge pull request #20433 from pdvian/wip-22971-luminous

mon: allow removal of tier of ec overwritable pool

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

Merge pull request #20495 from pdvian/wip-23024-luminous

luminous: osd: Fix assert when checking missing version

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

Merge pull request #20554 from pdvian/wip-23074-luminous

luminous: bluestore: statfs available can go negative

Reviewed-by: Igor Fedotov <ifedotov@mirantis.com>

Merge pull request #20585 from pdvian/wip-23114-luminous

luminous: librados: make OPERATION_FULL_FORCE the default for rados_remove()

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

Merge pull request #20609 from pdvian/wip-23075-luminous

luminous: osd: objecter sends out of sync with pg epochs for proxied ops

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

Merge pull request #20988 from pdvian/wip-23410-luminous

luminous: legal: remove doc license ambiguity

Reviewed-by: Abhishek Lekshmanan <abhishek.lekshmanan@gmail.com>
Reviewed-by: Nathan Cutler <ncutler@suse.com>

Merge pull request #20825 from rzarzynski/wip-22534-luminous

luminous: rocksdb: incorporate the fix in RocksDB for no fast CRC32 path.

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

Merge pull request #20604 from pdvian/wip-23101-luminous

luminous: systemd: Wait 10 seconds before restarting ceph-mgr

Reviewed-by: Wido den Hollander <wido@widodh.nl>
Reviewed-by: Nathan Cutler <ncutler@suse.com>

Merge pull request #20651 from gregsfortytwo/luminous-23141-rotational

luminous: bluestore: correctly check all block devices to decide if journal is_…

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

Merge pull request #20664 from tchaikov/wip-luminous-22344

luminous: ceph.in: print all matched commands if arg missing

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

Merge pull request #20771 from tchaikov/wip-23212-luminous

luminous: os/bluestore: recalc_allocated() when decoding bluefs_fnode_t

Reviewed-by: Igor Fedotov <ifedotov@mirantis.com>

Merge pull request #20781 from wido/luminous-zabbix

mgr/zabbix: Backports to Luminous

Reviewed-by: John Spray <john.spray@redhat.com>

Merge pull request #20785 from badone/wip-luminous-osd-removing

luminous: osd: add numpg_removing metric

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

Merge pull request #20957 from pdvian/wip-23351-luminous

luminous: os/filestore: fix do_copy_range replay bug

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

Merge pull request #20964 from ifed01/wip-ifed-reap-col-luminous

luminous: os/bluestore: backport fixes around _reap_collection

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

Merge pull request #20725 from smithfarm/wip-23230-luminous

luminous: Update mgr/restful documentation

Reviewed-By: Boris Ranto <branto@redhat.com>

Merge PR #20527 into luminous

* refs/pull/20527/head:
mds: various fixes for backport
mds: convert to boost::string_view
test/encoding: refactor to avoid escaping shell magic
mds: minor refactor of SimpleLock
mds: track Capability in mempool
mds: move CInode container members to mempool
mds: move CDentry container members to mempool
mds: move CDir container members to mempool
mds: put MDSCacheObject compact_map in mempool
common: use size_t for object size
mds: convert to allocator agnostic string_view
mds: simplify initialization
compact_*: support mempool allocated containers

Reviewed-by: Patrick Donnelly <pdonnell@redhat.com>

Merge pull request #20251 from pdvian/wip-22862-luminous

luminous: cephfs-journal-tool: Fix Dumper destroyed before shutdown

Reviewed-by: Patrick Donnelly <pdonnell@redhat.com>

Merge pull request #20302 from pdvian/wip-22891-luminous

luminous: qa: synchronize kcephfs suites with fs/multimds

Reviewed-by: Patrick Donnelly <pdonnell@redhat.com>

Merge pull request #20342 from batrick/i22688

luminous: client: release revoking Fc after invalidate cache

Reviewed-by: Yan, Zheng <zyan@redhat.com>

Merge pull request #20354 from pdvian/wip-22935-luminous

luminous: Client: setattr should drop "Fs" rather than "As" for mtime and size

Reviewed-by: Patrick Donnelly <pdonnell@redhat.com>

Merge pull request #20356 from pdvian/wip-22936-luminous

luminous : Client:Fix readdir bug

Reviewed-by: Patrick Donnelly <pdonnell@redhat.com>

Merge pull request #20417 from pdvian/wip-22966-luminous

luminous : qa: adjust cephfs full test for kclient

Reviewed-by: Patrick Donnelly <pdonnell@redhat.com>

Merge PR #20522 into luminous

* refs/pull/20522/head:
qa: ignore full cluster warnings for mds-full

Reviewed-by: Patrick Donnelly <pdonnell@redhat.com>

Merge PR #20457 into luminous

* refs/pull/20457/head:
qa: ignore bad backtrace cluster wrn
qa/cephfs: Add tests to validate scrub functionality
cephfs: Add option to load invalid metadata from disk
cephfs: Reset scrub data when inodes move

Reviewed-by: Patrick Donnelly <pdonnell@redhat.com>

Merge pull request #20432 from pdvian/wip-22969-luminous

luminous: mds: fix session reference leak

Reviewed-by: Patrick Donnelly <pdonnell@redhat.com>

Merge pull request #20455 from pdvian/wip-23013-luminous

luminous: mds: optimize getattr file size

Reviewed-by: Patrick Donnelly <pdonnell@redhat.com>

Merge pull request #20523 from pdvian/wip-23061-luminous

luminous:qa: ignore MON_DOWN while thrashing mons

Reviewed-by: Patrick Donnelly <pdonnell@redhat.com>

Merge pull request #20525 from pdvian/wip-23062-luminous

luminous: qa: ignore io pause warnings in mds-full test

Reviewed-by: Patrick Donnelly <pdonnell@redhat.com>

Merge pull request #20626 from pdvian/wip-23150-luminous

luminous: mds: add uptime to MDS status

Reviewed-by: Patrick Donnelly <pdonnell@redhat.com>

Merge pull request #20654 from pdvian/wip-23156-luminous

luminous: common: omit short option for id in help for clients

Reviewed-by: Patrick Donnelly <pdonnell@redhat.com>

Merge pull request #20904 from pdvian/wip-23314-luminous

luminous: client: allow client to use caps that are revoked but not yet returned

Reviewed-by: Patrick Donnelly <pdonnell@redhat.com>
Reviewed-by: Jeff Layton <jlayton@redhat.com>

Merge pull request #20960 from pdvian/wip-23355-luminous

luminous: client: use either dentry_invalidate_cb or remount_cb to invalidate k…

Reviewed-by: Patrick Donnelly <pdonnell@redhat.com>

Merge pull request #21072 from ceph/backport-cv-vagrant-reload

luminous ceph-volume improve robustness when reloading vms in tests

Reviewed-by: Andrew Schoen <aschoen@redhat.com>

ceph-volume tests create a retry script for vagrant reload

Signed-off-by: Alfredo Deza <adeza@redhat.com>
(cherry picked from commit 3be723d4d9816e6b487681d686f87cf41f67abd0)

ceph-volume tests use a reload script in simple functional tests

Signed-off-by: Alfredo Deza <adeza@redhat.com>
(cherry picked from commit b1b9c7a4162dcaa919c58c106915e77bab5eb40a)

ceph-volume tests use a reload script in lvm functional tests

Signed-off-by: Alfredo Deza <adeza@redhat.com>
(cherry picked from commit a54f5037861063d985b8ac1d158b919f119637b2)

Merge pull request #20253 from pdvian/wip-22857-luminous

luminous: librbd: potential crash if object map check encounters error

Reviewed-by: Jason Dillaman <dillaman@redhat.com>

Merge pull request #20416 from pdvian/wip-22964-luminous

luminous: rbd-mirror: fix potential infinite loop when formatting status message

Reviewed-by: Jason Dillaman <dillaman@redhat.com>

Merge pull request #20454 from pdvian/wip-23011-luminous

luminous: [journal] allocating a new tag after acquiring the lock should use on-disk committed position

Reviewed-by: Jason Dillaman <dillaman@redhat.com>

Merge pull request #20550 from pdvian/wip-23064-luminous

luminous: test/librbd: utilize unique pool for cache tier testing

Reviewed-by: Jason Dillaman <dillaman@redhat.com>

Merge pull request #20617 from smithfarm/wip-perms-ocf-rbd-luminous

luminous: cmake: rbd resource agent needs to be executable.

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

Merge pull request #20628 from pdvian/wip-23152-luminous

luminous: TestLibRBD.RenameViaLockOwner may still fail with -ENOENT

Reviewed-by: Jason Dillaman <dillaman@redhat.com>

Merge pull request #20715 from pdvian/wip-23177-luminous

luminous: qa/workunits/rbd: switch devstack tempest to 17.2.0 tag

Reviewed-by: Jason Dillaman <dillaman@redhat.com>

Merge pull request #20860 from pdvian/wip-23304-luminous

luminous: parent blocks are still seen after a whole-object discard

Reviewed-by: Ilya Dryomov <idryomov@redhat.com>

Merge pull request #20967 from pdvian/wip-23407-luminous

luminous: cls/rbd: group_image_list incorrectly flagged as RW

Reviewed-by: Jason Dillaman <dillaman@redhat.com>

Merge pull request #20991 from trociny/luminous-23423

luminous: librados/snap_set_diff: don't assert on empty snapset

Reviewed-by: Jason Dillaman <dillaman@redhat.com>

Merge pull request #21001 from pdvian/wip-23414-luminous

luminous: PendingReleaseNotes: add note about upgrading MDS

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

PendingReleaseNotes: add note about upgrading MDS

Fixes: https://tracker.ceph.com/issues/23172
Signed-off-by: Patrick Donnelly <pdonnell@redhat.com>
(cherry picked from commit 9c809d3c58126694e577ade6d017e61a6c283411)

Conflicts:
PendingReleaseNotes: added section 12.2.5 with cephfs release notes

Merge pull request #21031 from ceph/wip-yuriw-fix-j-x

qa/tests - Change machine type from 'vps' to 'ovh' as 'vps' does not …

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

qa/tests - Change machine type from 'vps' to 'ovh' as 'vps' does not exist anymore

Signed-off-by: Yuri Weinstein <yweinste@redhat.com>

Merge pull request #21015 from dzafman/wip-remove-bc-luminous

luminous: test: Replace bc command with printf command

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

test: Replace bc command with printf command

Signed-off-by: David Zafman <dzafman@redhat.com>
(cherry picked from commit 293ac9895f86b1258434d0dc539a7d4b7046cb6d)

Merge pull request #20736 from pdvian/wip-23176-luminous

luminous: qa/rgw: remove some civetweb overrides for beast testing

Reviewed-by: Casey Bodley <cbodley@redhat.com>

Merge pull request #20859 from pdvian/wip-23302-luminous

luminous: rgw: add radosgw-admin sync error trim to trim sync error log

Reviewed-by: Casey Bodley <cbodley@redhat.com>

Merge pull request #20889 from pdvian/wip-23310-luminous

luminous: rgw: s3website error handler uses original object name

Reviewed-by: Casey Bodley <cbodley@redhat.com>

Merge pull request #20932 from pdvian/wip-23318-luminous

luminous: rgw: fix crash with rgw_run_sync_thread false

Reviewed-by: Casey Bodley <cbodley@redhat.com>

librados/snap_set_diff: don't assert on empty snapset

Instead treat the diff as a full-object delta.

Signed-off-by: Mykola Golub <mgolub@suse.com>
(cherry picked from commit 2be4840afd4fc54c2ad06ccc2aac903e45bd37f5)

Conflicts:
src/librbd/deep_copy/ObjectCopyRequest.cc
src/librbd/deep_copy/ObjectCopyRequest.h
            deep_copy::ObjectCopyRequest does not exist in luminous,
            fixed rbd_mirror::image_sync::ObjectCopyRequest instead

Merge pull request #20937 from pdvian/wip-23346-luminous

luminous: rgw: reject encrypted object COPY before supported

Reviewed-by: Casey Bodley <cbodley@redhat.com>

legal: remove doc license ambiguity

The license applicable to the files under doc/ was originally declared (in the
top-level file COPYING) to be "Creative Commons Attribution-ShareAlike (CC
BY-SA)" by ed0653b493a3f919a3abc37a0aa9b5aa29ae0b0e

This license declaration omitted a version number.

Some time later ef7418421b3748c712019c8aedd02b8005c1e1ea was merged, mentioning
CC-BY-SA-1.0 as one of the Ceph source code licenses. Although the purpose of that
commit was only to summarize the canonical license information from COPYING, it
unintentionally became the only place in the source code where the doc license
version was specified.

In March 2018, the Debian project warned that CC-BY-SA-1.0 does not meet its
criteria for inclusion in the "free" section of the Debian distribution. (For
that, at least version 3.0 must be used.)

This new commit removes the doc license ambiguity by setting it to CC-BY-SA-3.0
in all places where the license is mentioned.

The exact spelling of the license name is taken from https://spdx.org/licenses/

Fixes: http://tracker.ceph.com/issues/23336
Signed-off-by: Nathan Cutler <ncutler@suse.com>
(cherry picked from commit e294dd59100d962753f4dee8f7156be5a7fba875)

Merge pull request #20672 from linuxbox2/luminous-21942

luminous: rgw_log, rgw_file: account for new required envvars

Reviewed-by: Casey Bodley <cbodley@redhat.com>
Reviewed-by: Ali Maredia <amaredia@redhat.com>

Merge pull request #20741 from pdvian/wip-23179-luminous

luminous: rgw: can't download object with range when compression enabled

Reviewed-by: Casey Bodley <cbodley@redhat.com>

Merge pull request #20157 from pdvian/wip-22812-luminous

luminous: rgw: log the right http status code in civetweb frontend's access log

Reviewed-by: Casey Bodley <cbodley@redhat.com>

Merge pull request #20252 from pdvian/wip-22858-luminous

luminous: rgw: allow beast frontend to listen on specific IP address

Reviewed-by: Casey Bodley <cbodley@redhat.com>

Merge pull request #20345 from pdvian/wip-22889-luminous

luminous: rgw_file: avoid evaluating nullptr for readdir offset

Reviewed-by: Matt Benjamin <mbenjami@redhat.com>

Merge pull request #20476 from pdvian/wip-23020-luminous

luminous: rgw: fix the max-uploads parameter not work

Reviewed-by: Casey Bodley <cbodley@redhat.com>

Merge pull request #20477 from pdvian/wip-23022-luminous

luminous: rgw: upldate the max-buckets when the quota is uploaded

Reviewed-by: Casey Bodley <cbodley@redhat.com>

Merge pull request #20586 from pdvian/wip-23102-luminous

luminous: rgw: parse old rgw_obj with namespace correctly

Reviewed-by: Casey Bodley <cbodley@redhat.com>

Merge pull request #20072 from smithfarm/wip-22766-luminous

luminous: RGW doesn't check time skew in auth v4 http header request

Reviewed-by: Robin H. Johnson <robbat2@gentoo.org>
Reviewed-by: Casey Bodley <cbodley@redhat.com>

os/bluestore: fix the wrong usage for map_any.

If Onode::flushing_count > 0, it should return true rahter than false.

Signed-off-by: Jianpeng Ma <jianpeng.ma@intel.com>
(cherry picked from commit 039ee8328e92d6834cd0cd4b77af75dc55c65221)

os/bluestore: remove lock for removed_collections.

Current, queue removed-collections and handle removed-collections
in the same thread. So no need a lock to protect.

Signed-off-by: Jianpeng Ma <jianpeng.ma@intel.com>
(cherry picked from commit 7646b7407b7cb6fe6f1b4bf3eb9e20d7cdfc5bce)

os/bluestore: remove _reap_collection in umount.

Queue removed collections and reap removed collection in the same
thread(_kv_finalize_thread). If this thread terminated and mean all
removed collection handle. Unless expection, so add a assert.

Signed-off-by: Jianpeng Ma <jianpeng.ma@intel.com>
(cherry picked from commit ab4f797bfd3739e9722615abe8be57a7bd364c5e)

os/bluestore: Re-record unclean removed-collection.

In _reap_collection, if removed-collection still has object which in
flushing. Collection don't clean cache and  it can't
record anymore which cause can't handle in the next time.

Signed-off-by: Jianpeng Ma <jianpeng.ma@intel.com>
(cherry picked from commit 196c71f4657ba19924e371e54b9d88c35448a2c5)

cls/rbd: group_image_list incorrectly flagged as RW

Fixes: http://tracker.ceph.com/issues/23388
Signed-off-by: Jason Dillaman <dillaman@redhat.com>
(cherry picked from commit 169d84c677270cbe72ff5cddef7b021332304d6f)

Merge pull request #20906 from pdvian/wip-23317-luminous

luminous: doc: rgw: mention the civetweb support for binding to multiple ports

Reviewed-By: Abhishek Lekshmanan <abhishek@suse.com>

Merge pull request #20242 from cbodley/wip-doc-rgw-frontends-luminous

luminous: doc/rgw: add page for http frontend configuration

Reviewed-By: Nathan Cutler <ncutler@suse.com>
Reviewed-By: Abhishek Lekshmanan <abhishek@suse.com>

client: use either dentry_invalidate_cb or remount_cb to invalidate kernel dcache

Signed-off-by: Zhi Zhang <zhangz.david@outlook.com>
(cherry picked from commit 6b823dc250223c2c25c341d6ffcfa64510f47bb2)

os/filestore: fix do_copy_range replay bug

The 'from' value is an fd, not an offset.  We want 'r' to be the number of
bytes copied, or len.

It looks like this has been broken since 9499cdcd87ddac4a91abffe401baddb91f8a11d5

Fixes: http://tracker.ceph.com/issues/23298
Signed-off-by: Sage Weil <sage@redhat.com>
(cherry picked from commit df7fa174ad604822c034226593ff1eb700506e4a)

Merge pull request #20923 from ceph/backport-20910

luminous: ceph-volume: fix action plugins path in tox

Reviewed-by: Alfredo Deza <adeza@redhat.com>

rgw: reject encrypted object COPY before supported

Current implementation does not follow S3 spec and even
may result in data corruption silently when copying
multipart objects accorss pools. So reject COPY operations
on encrypted objects before it is fully functional.

Fixes: http://tracker.ceph.com/issues/23232
Signed-off-by: Jeegn Chen <jeegnchen@gmail.com>
(cherry picked from commit a1513efe21af694e04db466a4d1d63cfd857876e)

rgw: fix crash with rgw_run_sync_thread false

Fixes: http://tracker.ceph.com/issues/20448
Signed-off-by: Orit Wasserman <owasserm@redhat.com>
(cherry picked from commit f9395d63f8d7900a22f485d6d143021d0b92d8a0)

ceph-volume: fix action plugins path in tox

because of this commit [1] in ceph-ansible tests in ceph-volume needs to
be modified accordingly.

[1] https://github.com/ceph/ceph-ansible/commit/60d4b75f519c03fca91384f231b071793f582376

Signed-off-by: Guillaume Abrioux <gabrioux@redhat.com>
(cherry picked from commit 9fdc56717e72e79dca49fc22da37fd185369f216)

pybind/mgr/balancer: add sanity check against empty adjusted_map

E.g., because all OSDs are marked as **out**.

Signed-off-by: xie xingguo <xie.xingguo@zte.com.cn>
(cherry picked from commit 10a752671af71c7b6db168d75b0b416de85992d3)

doc: rgw: mention the civetweb support for binding to multiple ports

We support binding multiple ports in civetweb after the upgrade to 1.8,
clarify this in the documentation as well

Fixes: http://tracker.ceph.com/issues/20942
Signed-off-by: Abhishek Lekshmanan <abhishek@suse.com>
(cherry picked from commit 01740c6a6c1ae523ac73c24235be0e4cef71e21c)

test: cap waits in delegation tests at 60s

This adds assertions to all of the looping wait functions that we will
not do more than 60000 loop iterations (~60s).

Signed-off-by: Jeff Layton <jlayton@redhat.com>
(cherry picked from commit f3840ca301bff801eecdfe20a98726ab6d555ba6)

tests: add a test for doing a getattr while caps are recalled but before returned

Add a testcase to reproduce the deadlock I was seeing with delegations
in ganesha.

Tracker: http://tracker.ceph.com/issues/23028
Signed-off-by: Jeff Layton <jlayton@redhat.com>
(cherry picked from commit 91e5bfa2406dd64e6ed3e168d4cf47dee60ffff0)

client: allow getattr, lookup, fstatx, and readdir to use implemented caps

caps_issued_mask is a function that we call to see if we have all of
the caps provided in a mask. Usually this ignores caps that have been
revoked by the MDS but that have not yet been released by the client.
These caps are still valid though -- nothing else can use them until
they've been released.

These codepaths call caps_issued_mask only in order to determine
whether the client holds enough caps to trust cached attributes. They
don't take new references to the caps, so it should be safe to use the
"implemented" capset in these functions.

This helps break a deadlock that can occur when the client holds a
delegation that has been recalled, but needs to satisfy a getattr
request before releasing it.

Tracker: http://tracker.ceph.com/issues/23028
Signed-off-by: Jeff Layton <jlayton@redhat.com>
(cherry picked from commit dacc7b6099f5c2e758e2811a7a6722bd0946eb39)

Conflicts:
src/client/Inode.cc: Resolved in Inode::caps_issued_mask

rgw: s3website error handler uses original object name

the s3website error handler needs to use the same object name for
redirect handling that retarget() does, but s->object.name may be
modified based on get_effective_key()

Fixes: http://tracker.ceph.com/issues/23201
Signed-off-by: Casey Bodley <cbodley@redhat.com>
(cherry picked from commit 8f7cbbce26c7fbccb7e9ae46040f3472d8762a88)

test/librados_test_stub: add mock version of `create`

Signed-off-by: Jason Dillaman <dillaman@redhat.com>
(cherry picked from commit ed147192fdc1cc589cdfb107f5d1264568492e88)

librbd: create+truncate for whole-object layered discards

A whole-object layered discard is implemented as a truncate rather
than a delete: a dummy object is needed to prevent the CoW machinery
from kicking in.  However, a truncate on a non-existent object is
a no-op.  If the object doesn't exist in HEAD, a discard request is
effectively ignored -- parent blocks are still seen afterwards.

A non-exclusive create on an existing object is also a no-op, so the
fix is to do a compound create+truncate instead.

Fixes: http://tracker.ceph.com/issues/23285
Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
(cherry picked from commit 0514480b16b0a1e76a9ba31cc3ba58a32b7bee18)

test/librbd: expect_object_may_exist(..., false) in DiscardRemoveTruncate

This is !is_no_op_for_nonexistent_object() test -- NONEXISTENT case is
the interesting one.

Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
(cherry picked from commit 9d6ee9a649861f1981981e41c82af8cd2914721b)

test/librbd: clone with parent features in DiscardRemoveTruncate

... otherwise object map expectations aren't actually set.

Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
(cherry picked from commit 513baf92ef010f7170719f4157f6d1a28fe5e051)