Merge pull request #12574 from liewegas/wip-bluestore-split

os/bluestore: preserve source collection cache during split

Reviewed-by: xie xingguo <xie.xingguo@zte.com.cn>

Merge pull request #12472 from dillaman/wip-18244

librbd: remove image header lock assertions

Reviewed-by: Mykola Golub <mgolub@mirantis.com>

Merge pull request #12532 from ifed01/wip-bluestore-large-onode-test

os/bluestore: add UT for an estimation of Onode in-memory size

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

Merge pull request #12505 from ceph/wip-jlayton-submount

client: set metadata["root"] from mount method when it's called with …

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

Merge pull request #12570 from wjwithagen/patch-9

common/str_list.h: fix clang warning about std::move

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

Merge pull request #12541 from liewegas/wip-bluestore-omap-flush

os/bluestore: include modified objects in flush list even if onode unchanged

Reviewed-by: Igor Fedotov <ifedotov@mirantis.com>
Reviewed-by: xie xingguo <xie.xingguo@zte.com.cn>

os/bluestore: preserve source collection cache during split

OSD split transactions look something like

 mkcoll new
 split old
 ...
 omap_rmkey_range old
 omap_setkeys old
 omap_setkeys new

The last part splits the log into two pieces.  The
problem is that the rmkey_range needs to wait on old
omap transactions to flush, and those are linked to the
old onode, and split clears the cache.  The result is
that we don't wait, rmkeyrange leaves some recent pg log
keys behind, and on OSD restart we get an error because
the object doesn't belong to the (old) collection.

Fix this by preserving objects in the old collection and
only clear out objects that are moving to the newly
split collections.  This will include the pgmeta object
that we care about.

(Note that we are one step closer to preserving the
cache contents across the split, but not quite there
yet: at this point we don't have all of the destination
collections.  A change in the ObjectStore interface is
probably needed to make that not be extremely awkward.)

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

Merge pull request #12571 from athanatos/wip-15943

osd: Fix map gaps again (bug 15943)

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

Merge pull request #12572 from liewegas/wip-python-rados-test

qa/workunits: include extension for nose tests

Merge pull request #12568 from ifed01/wip-bluestore-onode-diet2

os/bluestore: reduce Onode in-memory footprint

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

Merge pull request #12555 from socketpair/usr_bin

systemd: Fix startup of ceph-mgr on Debian 8

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

Merge pull request #12355 from chhabaramesh/bit_alloc_mempool

os/bluestore: simplified allocator interfaces to single apis

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

qa/workunits: include extension for nose tests

When you have a relative path you have to include the extension.
Weird.

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

Merge pull request #12412 from ceph/wip-jlayton-suid

client/mds: Clear setuid bits when writing or truncating

Clang/str_list.h: fix clang warning about std::move

/home/jenkins/workspace/ceph-master/src/include/str_list.h:99:10: warning: moving a local object in a return statement prevents copy elision [-Wpessimizing-move]
  return std::move(str_vec);
         ^
/home/jenkins/workspace/ceph-master/src/include/str_list.h:99:10: note: remove std::move call here
  return std::move(str_vec);
         ^~~~~~~~~~       ~
1 warning generated.

Signed-off-by: Willem Jan Withagen <wjw@digiware.nl>

Merge pull request #12565 from trociny/wip-18247-fixup

common: fix clang compilation error

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

os/bluestore: remove duplicate sbid member for Blob, use one from shared_blob member
Signed-off-by: Igor Fedotov <ifedotov@mirantis.com>

os/bluestore: declare a member for Blob's cached bufferlist if explicitly enabled by CACHE_BLOG_BL macro only

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

os/bluestore: recalculate shared blob key instead of keeping it in memory
Signed-off-by: Igor Fedotov <ifedotov@mirantis.com>

os/bluestore: include modified objects in flush list even if onode unchanged

We use the onode flush list so that we can ->flush() as
a barrier before doing any read/modify/write.  For
example, omap_rmkeyrange will flush before reading to
see what keys to erase in order to ensure that any
previous inserts are applied to the db and we see them
and remove them.

However, some omap operations don't update the onode
itself, which means write_onode() doesn't get called and
we aren't put on this list.

Add a note_modified_object() helper that can be called
instead of write_onode() for those cases.  That way we
get on the list and flush() works as expected.

We could have resolved this by just putting ourselves on
the dirty onode list, but in practice every OSD op is
writing omap keys to the pgmeta object and there is no
need to touch the onode key in this case, so doing so
would be a big regression.

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

Merge pull request #12548 from Liuchang0812/add-document-rgw

doc: explain rgw_fcgi_socket_backlog in rgw/config-ref.rst

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

Merge pull request #12561 from kylinstorage/wip-fix-osdmaptool-zero

osdmaptool: fix divide by zero error

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

Merge pull request #12497 from dillaman/wip-18243

journal: prevent repetitive error messages after being blacklisted

Reviewed-by: Mykola Golub <mgolub@mirantis.com>

remove unused wait argument and related function versions

Signed-off-by: Ramesh Chander <Ramesh.Chander@sandisk.com>

common: fix clang compilation

Signed-off-by: Mykola Golub <mgolub@mirantis.com>

osdmaptool: fix divide by zero error

Signed-off-by: Yunchuan Wen <yunchuan.wen@kylin-cloud.com>

Merge pull request #12558 from Adirl/fix_bad_len_msg

msg/async/rdma: fix bad message that went to the user

Reviewed-by: Haomai Wang <haomai@xsky.com>

zone block offset bug fix and related unit test case

Signed-off-by: Ramesh Chander <Ramesh.Chander@sandisk.com>
zone block offset related test case

Signed-off-by: Ramesh Chander <Ramesh.Chander@sandisk.com>

allocat_extent api rename and tests for new api

Signed-off-by: Ramesh Chander <Ramesh.Chander@sandisk.com>

remove cont alloc alloc_blocks calls from allocator code and test

Signed-off-by: Ramesh Chander <Ramesh.Chander@sandisk.com>

consolidate allocator calls to single alloc_extent api

Signed-off-by: Ramesh Chander <Ramesh.Chander@sandisk.com>

allocator min_alloc arguments in call stack and tests

Signed-off-by: Ramesh Chander <Ramesh.Chander@sandisk.com>

allocator alloc_extents min_alloc

Signed-off-by: Ramesh Chander <Ramesh.Chander@sandisk.com>

os/bluestore:Bluefs balance should use alloc_extent api with bluefs_alloc_size

Signed-off-by: Ramesh Chander <Ramesh.Chander@sandisk.com>

msg/async/rdma: fix bad message that went to the user

1. the printed value of "bad length" was incorrect, because 'r' was changed before the log line.
2. log appeared to user as error even though everything was calm; hence, reducing its severity

issue: 914840

Change-Id: I41520d27fc53fbbe07b61dd915e16448b87769d5
Signed-off-by: Avner BenHanoch <avnerb@mellanox.com>

Fix startup of Ceph cluster manager daemon on Debian 8

Signed-off-by: Mark Korenberg <socketpair@gmail.com>

Merge pull request #12531 from dachary/wip-18291-bluestore

ceph-disk: do not create bluestore wal/db partitions by default

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

ceph-disk: do not create bluestore wal/db partitions by default

Fixes: http://tracker.ceph.com/issues/18291
Signed-off-by: Loic Dachary <loic@dachary.org>

doc: explain rgw_fcgi_socket_backlog in rgw/config-ref.rst

Signed-off-by: liuchang0812 <liuchang0812@gmail.com>

Merge pull request #12458 from Liuchang0812/add-more-log-for-network-error

osd: enhance logging for osd network error

Reviewed-by: Kefu Chai <kchai@redhat.com>
Reviewed-by: xie xingguo <xie.xingguo@zte.com.cn>

ods: enhance logging for osd network error

Signed-off-by: liuchang0812 <liuchang0812@gmail.com>

Merge pull request #12343 from liewegas/wip-bluestore-release

os/bluestore: simplify allocator release flow

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

Merge pull request #12351 from xiexingguo/xxg-wip-bluestore-1207

os/bluestore: clear omap flag if parent has none

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

Merge pull request #12353 from xiexingguo/xxg-wip-bluestore-1207-01

os/bluestore: don't implicitly create the source object for clone

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

Merge pull request #12466 from kylinstorage/wip-remove-unused-class-declaration

remove unused declaration

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

Merge branch 'master' into wip-remove-unused-class-declaration

Merge pull request #12538 from liewegas/wip-workunits

qa/tasks/workunit: retry on ceph.git if checkout fails

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

Merge pull request #12534 from liewegas/wip-truncate-fix

osd/PrimaryLogPG: don't truncate if we don't have to for WRITEFULL

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

qa/tasks/workunit: retry on ceph.git if checkout fails

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

Merge pull request #12486 from dillaman/wip-18247

librbd: keep rbd_default_features setting as bitmask

Reviewed-by: Mykola Golub <mgolub@mirantis.com>

Merge pull request #12514 from vasukulkarni/wip-cd-shaman

qa/tasks/ceph_deploy: use dev option instead of dev-commit

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

ReplicatedPG: don't truncate if we don't have to for WRITEFULL

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

Merge pull request #10160 from jmunhoz/bugfix-16463-presigned-url

rgw: aws4: add presigned url bugfix in runtime

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

Merge pull request #12530 from liewegas/wip-bluestore-ceph-disk

ceph-disk: adjust bluestore default device sizes

Reviewed-by: Loic Dachary <ldachary@redhat.com>

test: add mempool dump to Onode size tracking test case

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

os/bluestore: add an estimation for Onode in-memory size

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

Merge pull request #12516 from hjwsm1989/fix-rpg

osd/PG: fix the unreadable log in choose_acting

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

Merge pull request #12521 from kylinstorage/wip-for-coverity

osd/osdmap: fix divide by zero error

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

client: set metadata["root"] from mount method when it's called with a pathname

Currently, we only set the root properly config file or the
--client_metadata command line option. If a userland client program
tries to call ceph_mount with a pathname, it's not being properly
set.

Since we already hold the mutex, we can just update it directly.

Signed-off-by: Jeff Layton <jlayton@redhat.com>

Merge pull request #12507 from ifed01/wip-bluestore-fix-cache

os/bluestore: fix target_buffer value overflow in Cache::trim()

Reviewed-by: xie xingguo <xie.xingguo@zte.com.cn>

Merge pull request #12525 from trociny/wip-18261

rbd: fix json formatting for image and journal status output

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

ceph-disk: default DB size of 20GB

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

ceph-disk: default WAL size of 576 MB

Enough for 2 256MB write buffers, and then some.

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

librbd: keep rbd_default_features setting as bitmask

Support both human readable, comma delimited list of feature
names and also integer bitmask value. Attempting to read the
setting will always result in the feature bitmask integer
value.

This is required to avoid breaking backwards compatibility with
librbd clients that are dependent on the older behavior (e.g.
OpenStack).

Fixes: http://tracker.ceph.com/issues/18247
Signed-off-by: Jason Dillaman <dillaman@redhat.com>

Merge pull request #12407 from majianpeng/rbd-stripe-unit

rbd : make option --stripe-unit w/ B/K/M work

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

Merge pull request #12484 from runsisi/wip-fix-rollback-warn

librbd: ignore error when object map is already locked by current client

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

rbd: fix json formatting for image and journal status output

Fixes: http://tracker.ceph.com/issues/18261
Signed-off-by: Mykola Golub <mgolub@mirantis.com>

rbd: make option --stripe-unit w/ B/K/M work.

As 'man rbd' said:  --stripe-unit size-in-B/K/M. But w/ B/K/M, the
command failed and said invalid options. So fix it.

Signed-off-by: Jianpeng Ma <jianpeng.ma@intel.com>

osdmap: fix divide by zero error

*** CID 1397255:  Incorrect expression  (DIVIDE_BY_ZERO)
/home/brad/working/src/ceph/src/osd/OSDMap.cc: 2995 in
OSDMap::summarize_mapping_stats(OSDMap*, const std::set<long,
std::less<long>, std::allocator<long>> *,
std::__cxx11::basic_string<char, std::char_traits<char>,
std::allocator<char>> *, ceph::Formatter *) const()
2989         f->open_object_section("utilization");
2990       if (newmap) {
2991         if (f) {
2992           f->dump_unsigned("moved_pgs", moved_pg);
2993           f->dump_unsigned("total_pgs", total_pg);
2994         } else {
>>>     CID 1397255:  Incorrect expression  (DIVIDE_BY_ZERO)
>>>     In expression "(float)moved_pg * 100. / (float)total_pg",
division by expression "total_pg" which may be zero has undefined
behavior.
2995           ss << "moved " << moved_pg << " / " << total_pg
2996       << " (" << ((float)moved_pg * 100.0 / (float)total_pg) <<
"%)\n";
2997         }
2998       }
2999       if (f) {
3000         f->dump_float("avg_pgs", avg_pg);

Signed-off-by: Yunchuan Wen <yunchuan.wen@kylin-cloud.com>

  osd/PG: fix the unreadable log in choose_acting
          To avoid the log like: failed to fill position ^B
Signed-off-by: huangjun <hjwsm1989@gmail.com>

librbd: ignore error when object map is already locked by current client

otherwise when using rbd cli to rollback image with object-map feature
enabled, the following error message will be printed out on the screen,
which is confusing to users:
  librbd::object_map::LockRequest: failed to lock object map: (17) File exists

Fixes: http://tracker.ceph.com/issues/16179
Signed-off-by: runsisi <runsisi@zte.com.cn>

rgw: aws4: add rgw_s3_auth_aws4_force_boto2_compat conf option

Runtime bugfix to handle presigned urls computed with canonical requests using
the port number once.

Boto2 computes canonical requests using the port number twice although it
should be used once only. This behaviour is a bug supported by AWS S3. Boto2 is
used in RGW S3 as reference implementation.

The client-side tools not supporting this boto2 bug will fail although they
should work too.

In order to support both presigned url implementations this patch adds a config
option to compute a second signature. With this option disabled, the code will
compute two signatures when the first signature is not valid. The aws4 auth
succeed if some of the two signatures is valid.

The config option rgw_s3_auth_aws4_force_boto2_compat, is enabled by default so
one signature, working with boto2, is computed only.

Fixes: http://tracker.ceph.com/issues/16463
Signed-off-by: Javier M. Mellid <jmunhoz@igalia.com>

Merge pull request #12463 from wjwithagen/wip-wjw-freebsd-osd-avoidports

FreeBSD/OSD.cc: add client_messenger to the avoid_ports set.

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

use dev option instead of dev-commit

Signed-off-by: Vasu Kulkarni <vasu@redhat.com>

Merge pull request #12500 from badone/wip-doc-PGPool

doc/dev/osd_internals: add pgpool.rst

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

Merge pull request #12511 from liewegas/wip-workunits

qa/workunits/rbd: fix

Reviewed-by: Jason Dillaman <dillaman@redhat.com>
Reviewed-by: Mykola Golub <mgolub@mirantis.com>

qa/workunits: use relative path instead of wget from git

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

Merge pull request #12508 from liewegas/wip-qa-admin-socket

qa/tasks/admin_socket: subst in repo name

doc/dev/osd_internals: add pgpool.rst

Signed-off-by: Brad Hubbard <bhubbard@redhat.com>

os/bluestore: clone_range of len 0 creates but does not size object

This mimics, roughly,

 open(..., O_WRONLY|O_CREAT, ...)
 seek(...)
 if (len > 0)
   write(...)

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

os/bluestore: remove commit_{start,finish} from Allocator

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

os/bluestore/BlueFS: do not call commit_{start,finish}

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

os/bluestore/BlueFS: defer releasing extents explicitly

Don't rely on the Allocator to do this for us (that is
about to go away!).

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

os/bluestore: do not call commit_{start,finish}

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

os/bluestore: defer bluefs reclaimed extent release

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

os/bluestore: release alloc directly into usable pool

Wait until after the kv transaction commits, and then
release extents directly into the usable pool.

This will let us remove the commit_{start,end} business.

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

qa/tasks/workunit.py: add CEPH_BASE env var

Root of git checkout

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

qa/tasks/workunit: leave workunits inside git checkout

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

PG: fix cached_removed_snaps bug in PGPool::update after map gap

5798fb3bf6d726d14a9c5cb99dc5902eba5b878a actually made 15943 worse
by always creating an out-of-date cached_removed_snaps value after
a map gap rather than only in the case where the the first map after
the gap did not remove any snapshots.

Introduced: 5798fb3bf6d726d14a9c5cb99dc5902eba5b878a
Fixes: http://tracker.ceph.com/issues/15943
Signed-off-by: Samuel Just <sjust@redhat.com>

qa/config/rados.yaml: enable osd_debug_verify_cached_snaps

Also, make map gaps more likely.

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

PG::handle_advance_map: add debugging option to verify cached_removed_snaps

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

qa/tasks/admin_socket: subst in repo name

It is either ceph.git or ceph-ci.git.

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

os/bluestore: fix target_buffer value overflow in Cache::trim() when metadata_ratio is set to 1

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

Merge pull request #12396 from xiaoxichen/fix_diff

librbd: use proper snapshot when computing diff parent overlap

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

Merge pull request #12482 from badone/wip-librados-objectcursor-leak

librados: Memory leaks in object_list_begin and object_list_end

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

Merge pull request #12286 from dillaman/wip-18123

rados: optionally support reading omap key from file

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

Merge pull request #12462 from XinzeChi/wip-blk-link

common/blkdev: use realpath instead of readlink to resolve the recurs…

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

Merge pull request #12488 from liewegas/wip-bluestore-enc

os/bluestore: fix ondisk encoding for blobs

Reviewed-by: Igor Fedotov <ifedotov@mirantis.com>
Reviewed-by: Varada Kari <varada.kari@sandisk.com>

os/bluestore/bluestore_types: bluestore_onode_t: varint encoding

Changing the uint32_t encoding to varint encoding saves around 120 bytes
per onode.

Signed-off-by: Varada Kari <varada.kari@sandisk.com>

Merge pull request #12502 from kylinstorage/wip-ctypos2

fix typos

Reviewed-by: xie xingguo <xie.xingguo@zte.com.cn>