librados: Add rados_pool_get_base_tier call

Fixes: #8188
Signed-off-by: Adam Crume <adamcrume@gmail.com>

Merge pull request #2618 from ceph/wip-coverity-20141001

Fixes for Coverity issues

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

Merge pull request #2452 from ceph/wip-pool-ls

mon: add 'osd pool ls [detail]' command

Reviewed-by: Joao Eduardo Luis <joao@redhat.com>

test_librbd.cc: fix compiler warning

Signed-off-by: Danny Al-Gaaf <danny.al-gaaf@bisect.de>

client/SyntheticClient.cc: remove dead code

Fix for:

const: At condition do_sync, the value of do_sync must be equal to 0.
dead_error_condition: The condition do_sync cannot be true.
CID 716893 (#1 of 1): Logically dead code (DEADCODE)
dead_error_begin: Execution cannot reach this statement op.OSDOp()

Signed-off-by: Danny Al-Gaaf <danny.al-gaaf@bisect.de>

doc: A couple of fixes to the CloudStack documentation

SyntheticClient.cc: remove dead code

Fix for coverity issue:

const: At condition unsafe > 0, the value of unsafe must be equal to 0.
dead_error_condition: The condition unsafe > 0 cannot be true.

CID 716892 (#1 of 1): Logically dead code (DEADCODE)
dead_error_begin: Execution cannot reach this statement

Signed-off-by: Danny Al-Gaaf <danny.al-gaaf@bisect.de>

LogEntry.h: init LogEntry::prio in constructor

CID 1240202 (#1 of 1): Uninitialized scalar field (UNINIT_CTOR)
 uninit_member: Non-static class member prio is not initialized
 in this constructor nor in any functions that it calls.

Signed-off-by: Danny Al-Gaaf <danny.al-gaaf@bisect.de>

osd: fix need_journal call

From 2955b3da4e2c472c18b45e81eba791f8ccd58061

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

Merge pull request #2607 from yuyuyu101/wip-9580

ObjectStore: Add "need_journal" interface to make aware of journal devic...

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

Merge pull request #2608 from yuyuyu101/pipe-iovec

Pipe: avoid redundancy new/delete for struct iovec

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

ObjectStore: Add "need_journal" interface to make aware of journal device

Impl feature #9580

Signed-off-by: Haomai Wang <haomaiwang@gmail.com>

Pipe: avoid redundancy new/delete for struct iovec

Signed-off-by: Haomai Wang <haomaiwang@gmail.com>

Merge pull request #2605 from ceph/wip-osd-objecter

osd: do not bind ms_objecter messenger

Reviewed-by: Greg Farnum <greg@inktank.com>

Merge pull request #2597 from cchengleo/wip-test-shared-cache

test shared cache

Reviewed-by: Loic Dachary <loic-201408@dachary.org>

test-shared-cache:
  Add test for "lower_bound" method.
  Add multithread tests for "lookup" and "lower_bound" methods.

Signed-off-by: Cheng Cheng <ccheng.leo@gmail.com>
               Loic Dachary <loic@dachary.org>

test-shared-cache:
  Initial draft for the unit test of "common/shared_cache.hpp".

Signed-off-by: Cheng Cheng <ccheng.leo@gmail.com>
               Loic Dachary <loic@dachary.org>

Merge pull request #2609 from dachary/wip-doc-context

common: document C_GatherBuilder and C_Gather

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

common: document C_GatherBuilder and C_Gather

* Introduction explaining what they are used for
* Commented example to illustrate the abstract description

Signed-off-by: Loic Dachary <loic-201408@dachary.org>

Merge pull request #2502 from yuyuyu101/kvstore-meta-cache

KeyValueStore: performance improvement and docs

Merge pull request #2523 from ceph/wip-6926

rbd: Fix rbd diff for non-existent objects

Reviewed-by: Josh Durgin <josh.durgin@inktank.com>

Merge remote-tracking branch 'gh/giant'

Conflicts:
qa/workunits/cephtool/test.sh
src/test/erasure-code/TestErasureCodeIsa.cc

osd: do not bind ms_objecter messenger

The objecter messenger is only used as a client to initiate client-side
connections to other OSDs.  It doesn't need to bind to a port.

This was added in 558d9fc956204b7a18001b148e171b893cb63d85 to push client
traffic to the cluster interface.  This doesn't actually help/work because
we are still connecting to our peers' client-facing addresses.

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

Merge pull request #2602 from athanatos/wip-9574

PG: check full ratio again post-reservation

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

Merge pull request #2604 from athanatos/wip-9113

ReplicatedPG: clean out completed trimmed objects as we go

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

Merge pull request #2549 from ceph/wip-9545

os/FileJournal: do not request sync while shutting down

Reviewed-by: Samuel Just <sam.just@inktank.com>

Merge pull request #2550 from ceph/wip-8629

osd: fix cache_evict vs make_writeable/finish_ctx snapdir bug #8629

Reviewed-by: Samuel Just <sam.just@inktank.com>

Merge pull request #2510 from somnathr/wip-obj-delete-fix

FileStore: Race condition during object delete is fixed

Reviewed-by: Samuel Just <sam.just@inktank.com>

Merge pull request #2598 from ceph/wip-9582

librados: fix other timeout segfault

Reviewed-by: Greg Farnum <greg@inktank.com>

Merge pull request #2458 from ceph/wip-9341

client: trim unused inodes before reconnecting to recovering MDS

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

Merge pull request #2555 from theanalyst/wip/mailmap

mailmap updates

Reviewed-by: Loic Dachary <loic-201408@dachary.org>

mailmap: Riccardo Ferretti affiliation

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

mailmap: Casey Marshall affiliation

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

mailmap: Feng Wang affiliation

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

mailmap: Scott A. Brandt affiliation

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

mailmap: Gerben Meijer affiliation

Signed-off-by: Abhishek Lekshmanan <abhishek.lekshmanan@gmail.com>
Reviewed-by: Gerben Meijer <gerben@daybyday.nl>

mailmap: John Wilkins name normalization

Also adding John Wilkins to .peoplemap to track org change

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

mailmap: John Spray name normalization

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

mailmap: Samuel Just name normalization

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

mailmap: Pavan Rallabhandi name normalization

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

mailmap: Swami Reddy name normalization

Signed-off-by: Abhishek Lekshmanan <abhishek.lekshmanan@gmail.com>
Reviewed-by: Swami Reddy <swamireddy@gmail.com>

mailmap: Volker Voigt affiliation

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

mailmap: Laurent Guerby affiliation

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

mailmap: Johnu George affiliation

Signed-off-by: Abhishek Lekshmanan <abhishek.lekshmanan@gmail.com>
Reviewed-by: Johnu George <johnugeo@cisco.com>

mailmap: Janne Grunau affiliation

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

mailmap: Daniel Schepler affiliation

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

mailmap: Cheng Cheng affiliation

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

mailmap: Sahid Ferdjaoui affiliation

Signed-off-by: Abhishek Lekshmanan <abhishek.lekshmanan@gmail.com>
Reviewed-by: Sahid Orentino Ferdjaoui <sahid.ferdjaoui@redhat.com>

mailmap: Jiang Heng affiliation

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

mailmap: JuanJose Galvez affiliation

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

mailmap: Roman Haritonov affiliation

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

mailmap: Yongyue Sun affiliation

Signed-off-by: Abhishek Lekshmanan <abhishek.lekshmanan@gmail.com>
Reviewed-by: Loic Dachary <loic-201408@dachary.org>

Merge pull request #2594 from dachary/wip-9620-test-mon-thrash

qa/workunits/cephtool/test.sh: fix thrash (ultimate)

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

Merge pull request #2582 from ceph/wip-crush-location

crushtool: add --location <id> command

Reviewed-by: Loic Dachary <loic-201408@dachary.org>

qa/workunits/cephtool/test.sh: fix thrash (ultimate)

Keep the osd trash test to ensure it is a valid command but make it a
noop by giving it a zero argument (meaning thrash 0 OSD maps).

Remove the loops that were added after the command in an attempt to wait
for the cluster to recover and not pollute the rest of the tests. Actual
testing of osd thrash would require a dedicated cluster because it the
side effects are random and it is unnecessarily difficult to ensure they
are finished.

http://tracker.ceph.com/issues/9620 Fixes: #9620

Signed-off-by: Loic Dachary <loic-201408@dachary.org>

Merge pull request #2590 from dachary/wip-9592-librados-large-object

librados large object early check

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

librados: cap the IoCtxImpl::{aio_}*{write,append} buffer length

If the value of the len parameter is greater than UINT_MAX/2,
IoCtxImpl::aio_write, IoCtxImpl::aio_write_full, IoCtxImpl::aio_append,
IoCtxImpl::write, IoCtxImpl::append will fail with E2BIG.

IoCtxImpl::write_full is the exception because it does not have a
length argument to check.

For more information see 33501d242661a545211df43bf645398b492398ae

http://tracker.ceph.com/issues/9592 Fixes: #9592

Signed-off-by: Loic Dachary <loic-201408@dachary.org>

Merge pull request #2400 from majianpeng/fix2

osd: Make RPGTransaction::get_bytes_written return the correct size.

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

librados: test s/E2BIG/TooBig/

Because E2BIG does not allow selection

./ceph_test_rados_api_aio --gtest_filter=LibRadosAio.E2BIG
Running main() from gtest_main.cc
Note: Google Test filter = LibRadosAio.E2BIG
[==========] Running 0 tests from 0 test cases.
[==========] 0 tests from 0 test cases ran. (0 ms total)
[  PASSED  ] 0 tests.

probably because it contains a number.

Signed-off-by: Loic Dachary <loic-201408@dachary.org>

librados: cap the rados_aio_*{write,append} buffer length

If the value of the len parameter is greater than UINT_MAX/2,
rados_aio_write, rados_aio_write_full and rados_aio_append will fail
with E2BIG.

For more information see 33501d242661a545211df43bf645398b492398ae

http://tracker.ceph.com/issues/9592 Fixes: #9592

Signed-off-by: Loic Dachary <loic-201408@dachary.org>

osd: Make RPGTransaction::get_bytes_written return the correct size.

It record size larger than clien wrote. It should like
ECTransaction::get_bytes_written only return the size which clien
wrote. It should contain omap data.

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

crushtool: add --show-location <id> command

Include some tests.

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

Merge pull request #2584 from dachary/wip-9592-librados-large-object

librados: cap the rados*{write,append} buffer length

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

librados: cap the rados*{write,append} buffer length

When the caller submits a payload that will end up being rejected with

  rados.Error: Ioctx.write(rbd): failed to write hw: errno EMSGSIZE

it is stored in a bufferlist whose length is an unsigned int. If the
value of the len parameter is greater than UINT_MAX/2, rados_write,
rados_write_full and rados_append will fail with E2BIG.

Multiple calls to rados_write or rados_append can fill objects larger
than UINT_MAX/2.

http://tracker.ceph.com/issues/9592 Fixes: #9592

Signed-off-by: Loic Dachary <loic-201408@dachary.org>

Merge pull request #2580 from cernceph/wip-scientific

ceph-disk: add Scientific Linux as a Redhat clone

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

ceph-disk: add Scientific Linux as a Redhat clone

Scientific Linux is a RHEL clone and needs to use partx.

Signed-off-by: Dan van der Ster <daniel.vanderster@cern.ch>
(cherry picked from commit 5ca7ea5b531f28b3c87161d53aee907d6957cfc2)

Merge pull request #2568 from johnugeorge/wip-9492-crush

Crush: Ensuring at most num-rep osds are selected for any rule

Reviewed-by: Loic Dachary <loic-201408@dachary.org>

Crush: Ensuring at most num-rep osds are selected

Crush temporary buffers are allocated as per replica size configured
by the user.When there are more final osds (to be selected as per
rule) than the replicas, buffer overlaps and it causes crash.Now, it
ensures that at most num-rep osds are selected even if more number of
osds are allowed by the rule.

Fixes: #9492
Signed-off-by: Johnu George <johnugeo@cisco.com>

ceph-disk: add Scientific Linux as a Redhat clone

Scientific Linux is a RHEL clone and needs to use partx.

Signed-off-by: Dan van der Ster <daniel.vanderster@cern.ch>

Merge pull request #2572 from ceph/wip-9562

osdc/Filer: drop probe/purge locks before calling objecter

Reviewed-by: Greg Farnum <greg@inktank.com>
Reviewed-by: Yan, Zheng <ukernel@gmail.com>

Merge pull request #2575 from ceph/wip-zafman-cleanup

osd: Remove unused PG functions queue_notify(), queue_info(), queue_log(...

Reviewed-by: Loic Dachary <loic@dachary.org>

Merge pull request #2524 from ceph/wip-5768

rbd-fuse: Fix memory leak in enumerate_images

Reviewed-by: Josh Durgin <josh.durgin@inktank.com>

Merge pull request #2531 from dachary/wip-9536-isa-alignment

erasure-code: isa plugin alignment fixes

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

osd: improve debug output for do_{notifies,queries,infos}

Hunting #9389

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

Merge pull request #2540 from ceph/wip-giant-messenger-fixes

giant messenger fixes

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

osdc/Objecter: only post_rx_buffer if no op timeout

If we post an rx buffer and there is a timeout, the revocation can happen
while the reader has consumed the buffers but before it has decoded and
constructed the message.  In particular, we calculate a crc32c over the
data portion of the message after we've taken the buffers and dropped the
lock.

Instead of fixing this race (for example, by reverifying rx_buffers under
the lock while calculating the crc.. bleh), just skip the rx buffer
optimization entirely when a timeout is present.

Note that this doesn't cover the op_cancel() paths, but none of those users
provide static buffers to read into.

Fixes: #9582
Backport: firefly, dumpling
Signed-off-by: Sage Weil <sage@redhat.com>

Merge pull request #2574 from ceph/wip-msgr-shutdown

msg: allow calling dtor immediately after ctor

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

erasure-code: test isa encode/decode with various object sizes

Create an encode_decode() helper method to be called from the
encode_decode test function with various object size arguments. The
helper method is a copy/paste of the previous test that was using a
single object of a fixed size. The test is slightly adapted to
accommodate for different object sizes but the logic is not modified.

The object sizes being tested are chosen to be under the size of the
required size alignment or on multiple pages, size aligned or not.

Signed-off-by: Loic Dachary <loic-201408@dachary.org>

erasure-code: add test for isa chunk_size method

Signed-off-by: Loic Dachary <loic-201408@dachary.org>

msg: allow calling dtor immediately after ctor

Asserting on reaper_stop only made sense if the
messenger had ever been started: as it stood,
one couldn't create and destroy a messenger
without also starting and stopping it.

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

erasure-code: isa encode tests adapted to per chunk alignment

The encode tests use the alignment constraints. It has been changed to
be aligned on a per chunk basis instead of computing a more expensive
object alignement constraint. The test function is modified to take the
change into account but the logic is otherwise unmodified.

Signed-off-by: Loic Dachary <loic-201408@dachary.org>

erasure-code: isa test compare chunks with memcmp instead of strncmp

Because they may contain null characters.

Signed-off-by: Loic Dachary <loic-201408@dachary.org>

erasure-code: run isa tests via libtool and valgrind

Because running valgrind with no libtool does not test the binary but
the enclosing shell script.

Signed-off-by: Loic Dachary <loic-201408@dachary.org>

erasure-code: do not use typed tests for isa

Because there only is one type.

Signed-off-by: Loic Dachary <loic-201408@dachary.org>

erasure-code: isa uses per chunk alignment constraints

Copy code from the jerasure plugin to enforce alignment constraints per
chunk instead of using the total object size. It is simpler and reduces
the size of the chunks. See
https://github.com/ceph/ceph/commit/c7daaaf5e63d0bd1d444385e62611fe276f6ce29
for more information.

Signed-off-by: Loic Dachary <loic-201408@dachary.org>

erasure-code: [ISA] modify get_alignment function to imply a platform/compiler independent alignment constraint of 32-byte aligned buffer addresses & length

doc/release-notes: v0.67.11

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

Merge pull request #2571 from dachary/wip-9579-isa-documentation

documentation: erasure-code plugin isa does not require k/m

Reviewed-by: Andreas Peters <andreas.joachim.peters@cern.ch>

osdc/Filer: drop probe/purge locks before calling objecter

Fixes: #9562
Signed-off-by: John Spray <john.spray@redhat.com>

documentation: erasure-code plugin isa does not require k/m

http://tracker.ceph.com/issues/9579 Refs: #9579

Signed-off-by: Loic Dachary <loic-201408@dachary.org>

mailmap: Yan Zheng affiliation

Also adding Yan Zheng to .peoplemap to track org. change

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

mailmap: Thorsten Glaser affiliation

Signed-off-by: Abhishek Lekshmanan <abhishek.lekshmanan@gmail.com>
Reviewed-by: Thorsten Glaser <tg@mirbsd.de>

osd: Remove unused PG functions queue_notify(), queue_info(), queue_log()

Signed-off-by: David Zafman <dzafman@redhat.com>

ReplicatedPG: clean out completed trimmed objects as we go

Also, explicitely maintain a max number of concurrently trimming
objects.

Fixes: 9113
Backport: dumpling, firefly, giant
Signed-off-by: Samuel Just <sam.just@inktank.com>

Merge remote-tracking branch 'origin/giant'

Merge pull request #2567 from dachary/wip-6697-strncmp-vs-memcmp

tests: use memcmp to compare binary buffers

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

tests: use memcmp to compare binary buffers

instead of strncmp because it will stop at the first \0

http://tracker.ceph.com/issues/6697 Fixes: #6697

Signed-off-by: Loic Dachary <loic-201408@dachary.org>

Merge pull request #2506 from dachary/wip-9304-unintended-implicit-ruleset

erasure-code: pool create must not always create a ruleset

Reviewed-by: João Eduardo Luís <joao@redhat.com>

Merge remote-tracking branch 'origin/giant'

PG: check full ratio again post-reservation

Otherwise, we might queue 30 pgs for backfill at 0.80 fullness
and then never check again filling the osd after pg 11.

Fixes: #9574
Backport: dumpling, firefly, giant
Signed-off-by: Samuel Just <sam.just@inktank.com>