crimson/osd: add can_discard_op method for detecting "discardable" client requests

Signed-off-by: Xuehan Xu <xxhdx1985126@gmail.com>

crimson/osd: drop repop if it can be discarded

Signed-off-by: Xuehan Xu <xxhdx1985126@gmail.com>

crimson/osd: rename is_valid_rep_op_reply to can_discard_replica_op

Signed-off-by: Xuehan Xu <xxhdx1985126@gmail.com>

Merge pull request #37015 from tchaikov/wip-seastar-packet

common/buffer_seastar: reserve the packet slots in ctor

Reviewed-by: Radoslaw Zarzynski <rzarzyns@redhat.com>

Merge pull request #36946 from zdover23/wip-docs-36796-fix

doc/cephadm: Document mgr/cephadm/manage_etc_ceph_ceph_conf

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

Merge pull request #36995 from smithfarm/wip-47305

rpm: on SUSE, podman is required for cephadm to work

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

Merge pull request #36992 from uweigand/fix-encode-float

include/encoding: Fix encode/decode of float types on big-endian systems

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

Merge pull request #36829 from changchengx/osdmap

crush,osd: refine crushmap comment and implementation

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

Merge pull request #36947 from pdvian/wip-mgr-get-rule-root

mgr/mgr_module : Fix keyerror for op if rule steps first element is opcode

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

Merge pull request #36972 from smithfarm/wip-47112

rpm,deb: drop /etc/sudoers.d/cephadm

Reviewed-by: Sebastian Wagner <sebastian.wagner@suse.com>
Reviewed-by: David Disseldorp <ddiss@suse.de>
Reviewed-by: Kefu Chai <kchai@redhat.com>

Merge pull request #36937 from tchaikov/wip-mgr-cleanup

mgr: use range-based loop and cleanups

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

Merge pull request #36999 from neha-ojha/wip-47309

qa/*/mon/mon-last-epoch-clean.sh: mark osd out instead of down

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

Merge pull request #37008 from tchaikov/wip-crimson-pg-tell

crimson/osd: add "ceph tell <pgid> <command>" support

Reviewed-by: Ronen Friedman <rfriedma@redhat.com>

common/buffer_seastar: reserve the packet slots in ctor

for better performance, and simpler this way

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

Merge pull request #37010 from tchaikov/wip-crimson-osd-net

crimson/net: move implementation into .cc file

Reviewed-by: Ronen Friedman <rfriedma@redhat.com>
Reviewed-by: Radoslaw Zarzynski <rzarzyns@redhat.com>
Reviewed-by: Yingxin Cheng <yingxin.cheng@intel.com>

Merge pull request #37001 from tchaikov/wip-crimson-basic

qa/suites: add crimson-rados/basic tests

Reviewed-by: Radoslaw Zarzynski <rzarzyns@redhat.com>

Merge PR #36600 into master

* refs/pull/36600/head:
mds: make caps dump more readable

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

crimson/osd: add "ceph tell <pgid> <command>" support

* add an abstract class of `PGCommand` for `ceph tell <pgid> <command>`
* add two sample implementations for the pg tell commands.
  - "query"
  - "mark_unfound_lost"

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

crush/CrushWrapper: refine type_map/name_map comment

type_map is the item(bucket & device) type id and its type name.
name_map is the item(bucket & device) id and its name.

Signed-off-by: Changcheng Liu <changcheng.liu@aliyun.com>

crush/CrushWrapper: use "new_bucket_id" as var name of the added bucket

Signed-off-by: Changcheng Liu <changcheng.liu@aliyun.com>

crush/CrushWrapper: align variable name with comment

In the prolog of "for loop" statement, it writes "add child in the
location with 0 weight"

Signed-off-by: Changcheng Liu <changcheng.liu@aliyun.com>

osd/OSDMap: no need to give same locations per time to create bucket

Suggested-by: Kefu Chai <kefu@redhat.com>
Signed-off-by: Changcheng Liu <changcheng.liu@aliyun.com>

crush/CrushWrapper: refine comment to improve readability

Signed-off-by: Changcheng Liu <changcheng.liu@aliyun.com>

osd/OSDMap: assert osd_weight default initialized value is CEPH_OSD_OUT

refer: http://www.cplusplus.com/reference/vector/vector/resize/
Explicitly specify initialization value to vector::resize operation.

Suggested-by: Kefu Chai <kefu@redhat.com>
Signed-off-by: Changcheng Liu <changcheng.liu@aliyun.com>

osd/OSDMap: use class member instead of parameter to reset container size

Signed-off-by: Changcheng Liu <changcheng.liu@aliyun.com>

osd/OSDMap: avoid re-initialize osd state and weight done in set_max_osd

In set_max_osd function, container's osd_weight and osd_vector have been
assigned with the right state/weight value.

Signed-off-by: Changcheng Liu <changcheng.liu@aliyun.com>

crush: correct typo

Signed-off-by: Changcheng Liu <changcheng.liu@aliyun.com>

crush: no need to double check root bucket id

root bucket id should be definitely less than 0. assert(r < 0) here.
There's no possible to find device items in find_nonshadow_roots.

Signed-off-by: Changcheng Liu <changcheng.liu@aliyun.com>

include/encoding: Fix encode/decode of float types on big-endian systems

Currently, floating-point types use "raw" encoding, which means they're
simply copied as byte stream.

This means that if the decoding happens on a machine that differs in
byte order from the source machine, the returned value will be
incorrect. As one effect of this problem, a big-endian OSD node cannot
join a cluster where the MON node is little-endian (or vice versa),
because the OSDMap (incremental) structure contains floating-point
values, and as a result of this conversion problem, the OSD node will
crash with an assertion failure as soon as it receives any OSDMap update
from the MON.

This should be fixed by always encoding floating-point values in
little-endian byte order just as is done for integers. (Note that this
still assumes source and target machines used the same floating-point
format except for byte order. But given that nearly all platforms these
days use IEEE binary32/binary64 for float/double, that seems a
reasonable assumption.)

Fixes: https://tracker.ceph.com/issues/47302
Signed-off-by: Ulrich Weigand <ulrich.weigand@de.ibm.com>

Merge pull request #36924 from ddiss/sigabbrev_np

cmake: detect and use sigdescr_np() if available

Reviewed-by: Kefu Chai <kchai@redhat.com>
Reviewed-by: Nathan Cutler <ncutler@suse.com>

Merge pull request #36929 from tchaikov/wip-python-common-no-unicode

python-common: s/unicode/str/

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

crimson/net: move implementation into .cc file

for faster compilation.

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

Merge pull request #37007 from xxhdx1985126/wip-47313-47311

crimson/osd: accurately judge whether to do drop_recovery_read on recovering objs

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

Merge pull request #37004 from tchaikov/wip-discard-repop

crimson/osd: discard op if pg changes or sender is down

Reviewed-by: Xuehan Xu <xxhdx1985126@gmail.com>

crimson/osd: add PG::mark_unfound_lost()

it is a stub implementation for supporing asock command

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

crimson/osd: add PG::dump_primary()

for implementing asock command "ceph tell 1.0 query"

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

crimson/osd: drop unused header

osd_admin.h is referenced in the .cc file, so drop it in the header

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

crimson/osd: add OSD::get_pg()

for implementing asock commands.

and add `PGMap::get_pg()` as a helper

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

osd: refactor PrimaryLogPG::do_command("query")

we should let recovery_state.dump_peering_state() close the formatter
section, because it's recovery_state.dump_peering_state(), who opens it.

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

osd/PGStateUtils: do not forward declare unused class

PG is not referenced in this header

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

crimson/osd: print pgid in OSD::load_pg()

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

crimson/os: fix indent

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

crimson/osd: don't do push related operations if there's no shard to push

Signed-off-by: Xuehan Xu <xxhdx1985126@gmail.com>

crimson/osd: accurately judge whether to do drop_recovery_read on recovering objs

Fixes: https://tracker.ceph.com/issues/47313
Signed-off-by: Xuehan Xu <xxhdx1985126@gmail.com>

crimson/osd: discard op if pg changes or sender is down

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

qa/suites: add crimson-rados/basic tests

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

Merge pull request #37000 from xxhdx1985126/wip-47226

crimson/osd: construct log_keys_debug when reading pglog entries

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

crimson/osd: construct log_keys_debug when reading pglog entries

Fixes: https://tracker.ceph.com/issues/47226
Signed-off-by: Xuehan Xu <xxhdx1985126@gmail.com>

Merge pull request #36990 from tchaikov/wip-doc-prompt

doc/dev: use prompt directive when appropriate

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

doc/dev: drop unused label

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

doc/dev/developer_guide/tests-unit-tests: use literal "make check"

the ``make check`` is better than a hyper link here.

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

doc/dev: fix broken "make check" link

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

doc/dev: link to label not file

it's less error prone, and can help us avoid broken links if we move a
file around

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

qa/*/mon/mon-last-epoch-clean.sh: mark osd out instead of down

The test should mark the OSD out to check if only "in" OSDs are considered by
the osdmap trimming logic.

Fixes: https://tracker.ceph.com/issues/47309
Signed-off-by: Neha Ojha <nojha@redhat.com>

Merge PR #36983 into master

* refs/pull/36983/head:
client: make the osdmap wait under the mount_state reader check

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

Merge PR #36580 into master

* refs/pull/36580/head:
Client: defer puting the inode
client: defer cancelling the timer event
client: remove _sync_write_commit which makes no sense any more

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

Merge PR #36850 into master

* refs/pull/36850/head:
mds: add performance counter for cap messages

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

Merge PR #36464 into master

* refs/pull/36464/head:
client: get rstat from auth mds to reduce forwarding requests

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

doc/dev: use prompt directive when appropriate

for 2 reasons:

* sphinx renders codeblock using python syntax highlighting by default,
  so it's not surprising that it highlight keywords like "export" in
  command line samples. so to render command line code blocks, we'd
  better specify the syntax explicitly for better rendering result.
* with the help of "prompt" directive, user is able to copy and paste
  the command without the prompt. for instance, with the default
  "::" directive, user will copy "$ ceph df", which is not very
  convenient, but with "prompt" directive, user only copies
  "ceph df".

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

rpm: on SUSE, podman is required for cephadm to work

On SUSE, cephadm is tested with podman only.

Fixes: https://tracker.ceph.com/issues/47305
Signed-off-by: Nathan Cutler <ncutler@suse.com>

Merge pull request #36824 from SchoolGuy/feature/replace-png-with-svg

mgr/dashboard: Exchange PNGs for SVGs

Reviewed-by: Ernesto Puerta <epuertat@redhat.com>
Reviewed-by: Sebastian Krah <skrah@suse.com>
Reviewed-by: Tiago Melo <tmelo@suse.com>

Merge pull request #34831 from rhcs-dashboard/41060-API-wrong-error-code

mgr/dashboard: REST API returns 500 when no Content-Type is specified

Reviewed-by: Avan Thakkar <athakkar@redhat.com>
Reviewed-by: Ernesto Puerta <epuertat@redhat.com>
Reviewed-by: Laura Paduano <lpaduano@suse.com>
Reviewed-by: Volker Theile <vtheile@suse.com>

mgr/mgr_module: fix keyerror for op

If rule steps first element is opcode.

Fixes: https://tracker.ceph.com/issues/47272
Signed-off-by: Prashant Dhange <pdhange@redhat.com>

Merge pull request #36059 from liu-chunmei/seastore-extentmap-tree

crimson/seastore: add extent map tree

Reviewed-by: Samuel Just <sjust@redhat.com>

Merge pull request #36865 from ishanrai05/global-styles-fix

mgr/dashboard: fix style being applied globally

Reviewed-by: Avan Thakkar <athakkar@redhat.com>
Reviewed-by: Tiago Melo <tmelo@suse.com>
Reviewed-by: Volker Theile <vtheile@suse.com>

mds: make caps dump more readable

Signed-off-by: Yanhu Cao <gmayyyha@gmail.com>

client: make the osdmap wait under the mount_state reader check

It will make no sense to check this if the unmouting is already
going on.

https://tracker.ceph.com/issues/47293
Signed-off-by: Xiubo Li <xiubli@redhat.com>

Merge pull request #36489 from zdover23/wip-doc-dev-merge-request-2020-aug

doc/dev: add proper merge commit format

Merge pull request #36977 from tchaikov/wip-47290

mon/OSDMonitor: only take in osd into consideration when trimming osd…

Reviewed-by: Dan van der Ster <daniel.vanderster@cern.ch>
Reviewed-by: Josh Durgin <jdurgin@redhat.com>
Reviewed-by: Neha Ojha <nojha@redhat.com>

doc/dev: add proper Merge PR format

This commit adds an example of proper Merge PR format
to the Developer Guide. Though this might not be of
great use to most people, Zac Dover will find it most
helpful.

Fixes: https://tracker.ceph.com/issues/46843
Signed-off-by: Zac Dover <zac.dover@gmail.com>

doc/cephadm: managing ceph.conf

This commit repairs PR#36796, which was about
updating the configuration settings in ceph.conf
without touching the file directly.

That commit wasn't building because of a couple of
back quotes that were touching the word that followed
them. So much like the princess from The Princess and
The Pea is Sphinx. So careful must we always be.

Fixes: https://tracker.ceph.com/issues/47265
Signed-off-by: Zac Dover <zac.dover@gmail.com>

Merge pull request #36954 from HeinleinSupport/wip-doc-remove-ceph-deploy

doc: remove references to ceph-deploy

Reviewed-by: Zac Dover <zac.dover@gmail.com>

Merge PR #36902 into master

* refs/pull/36902/head:
mds: fix incorrect check for if dirfrag is being fragmented

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

Merge PR #36933 into master

* refs/pull/36933/head:
cephfs_mirror: Fix Clang compile errors

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

Merge pull request #35575 from cbodley/wip-45626

test/rgw: test_datalog_autotrim filters out new entries

Reviewed-by: Yuval Lifshitz <ylifshit@redhat.com>

Merge pull request #36758 from cbodley/wip-rgw-sal-user-sync-stats

rgw: rgw_user_sync_all_stats() uses is_truncated()

Reviewed-by: Daniel Gryniewicz <dang@redhat.com>

Merge pull request #36913 from cbodley/wip-47217

radosgw-admin: period pull command is not always a raw_storage_op

Reviewed-by: Daniel Gryniewicz <dang@redhat.com>

Merge pull request #28489 from croit/fix-40255

rgw: fix bucket limit check fill_status warnings

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

Merge pull request #30584 from zhangsw/fix-rgw-expiration-header

rgw: filter prefix should be used when generating expiration header

Reviewed-By: Abhishek Lekshmanan <abhishek@suse.com>
Reviewed-by: Matt Benjamin <mbenjamin@redhat.com>

Merge pull request #31058 from cbodley/wip-rgw-skip-bilog

cls/rgw: avoid read_bucket_header() when bi logging is disabled

Reviewed-by: J. Eric Ivancich <ivancich@redhat.com>

Merge pull request #31354 from xxcs/xxcs05

rgw: The rgw_bucket overload <operator, need to consider bucket_id

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

Merge pull request #31663 from cfanz/wip-rgw-fix-partial-copy-bug

rgw: do initialization in init_processing() for range copy

Reviewed-by: Daniel Gryniewicz <dang@redhat.com>

Merge pull request #33942 from pritha-srivastava/rgw-sts-new-keycloak-examples

rgw: Updating STS document with references to new keycloak

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

Merge PR #36546 into master

* refs/pull/36546/head:
vstart_runner: log commands in a more usable form

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

Merge pull request #35122 from BryceCao/wip-swift-buckets-jammed-metadata-sync

rgw: URL encode bucket name in metadata sync

Reviwed-by: Shilpa Jagannath <smanjara@redhat.com>

Merge pull request #35123 from BryceCao/wip-bug-fix-swift-bucket-create

rgw: urlencode bucket name when forwarding request

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

Merge pull request #35551 from joke-lee/v4-auth

rgw: replace '+' with "%20" in canonical query string for s3 v4 auth.

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

Merge pull request #35984 from matthewoliver/rgw_swift_api_401_for_v1_and_keystone

rgw: Swift API anonymous access should 401

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

Merge pull request #36502 from cbodley/wip-rgw-objv-tracker-inc

rgw: RGWObjVersionTracker tracks version over increments

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

Merge PR #36971 into master

* refs/pull/36971/head:
doc: fix typo and correct the affiliation

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

mon/OSDMonitor: only take in osd into consideration when trimming osdmaps

we should not take down osd into consideration when trimming osdmap. in
e62269c892, we decrease the upper bound of range of osdmaps to be trimmed
if the given osd is out. but we should have to decrease it only if the
osd in question is still *in*.

so, in this change, the min_lec is decreased only if the osd in question
is *in*.

Fixes: https://tracker.ceph.com/issues/47290
Signed-off-by: Kefu Chai <kchai@redhat.com>

Merge pull request #36218 from mkogan1/wip-cache-put-post

rgw: add PUT and POST req support to data cache

Reviewed-by: Or Friedmann <ofriedma@redhat.com>
Reviewed-by: Matt Benjamin <mbenjamin@redhat.com>

rpm,deb: drop /etc/sudoers.d/cephadm

Current behavior (without this patch) is:

1. cephadm package installs cephadm at /usr/sbin/cephadm
2. cephadm package installs /etc/sudoers.d/cephadm
3. !!! BUT this file refers to a non-existent executable (/usr/bin/cephadm) !!!
4. the PR that introduced this sudoers file (and this discrepancy) was merged in 2019
5. nobody noticed the discrepancy until now

My conclusion: the file /etc/sudoers.d/cephadm is not needed for cephadm to
work.

Fixes: https://tracker.ceph.com/issues/47112
Signed-off-by: Nathan Cutler <ncutler@suse.com>

Merge pull request #36542 from mkogan1/wip-rgw-fix-sync-stats

rgw: fix user stats iterative increment

Reviewed-by: Daniel Gryniewicz <dang@redhat.com>

Merge pull request #36932 from theMarix/patch-1

rpm: Fix typo in summary of prometheus-alerts package

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

Merge pull request #36951 from cbodley/wip-47240

cls/rgw: unlink_instance writes updated olh log on early return

Reviewed-by: Daniel Gryniewicz <dang@redhat.com>
Reviewed-by: J. Eric Ivancich <ivancich@redhat.com>

Merge pull request #36908 from rosinL/wip-valid-chunk

msg/async/rdma: use wr_id address to check valid chunk

Reviewed-by: Changcheng Liu <changcheng.liu@aliyun.com>

Merge pull request #36894 from pcuzner/device-ls-with-lsm

orchestrator: intergrate LSM information with device ls

Merge pull request #36911 from sebastian-philipp/cephadm-lsmcli-jinja

mgr/cephadm: customize blink_device_light cmd via j2

Reviewed-by: Patrick Seidensal <pseidensal@suse.com>
Reviewed-by: Volker Theile <vtheile@suse.com>

Merge pull request #36943 from sebastian-philipp/cephadm-add-comments-to-secondary-containers

mgr/cephadm: Add comments to secondary contaieners

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