Jason Dillaman
61f86e7ec1
cls_rbd: mirror image status summary should read full directory
...
Previously only retrieved the status for the first 64 images in
the rbd_mirroring directory.
Fixes: http://tracker.ceph.com/issues/16178
Signed-off-by: Jason Dillaman <dillaman@redhat.com>
2016-06-07 09:37:16 -04:00
Igor Fedotov
e7446ef2fd
os/BlueStore: Fixes improper blob length assignment when creating a new blob for large writes
...
Signed-off-by: Igor Fedotov <ifedotov@mirantis.com>
2016-06-07 16:37:04 +03:00
Igor Fedotov
6f8cc6111a
os/BlueStore: Adds checkes for lextent range overlaps to _fsck
...
Signed-off-by: Igor Fedotov <ifedotov@mirantis.com>
2016-06-07 16:36:44 +03:00
Sage Weil
cc9a07cabe
Merge pull request #9533 from jjhuo/no_pagecache
...
os/bluestore: make rocksdb not to use page cache
Reviewed-by: Sage Weil <sage@redhat.com>
2016-06-07 09:14:13 -04:00
Sage Weil
a577659d09
Merge pull request #8233 from majianpeng/bluestore-perfcount
...
os/bluestore: perf counters
Reviewed-by: Sage Weil <sage@redhat.com>
2016-06-07 09:10:25 -04:00
huanwen ren
47edef3739
tool/rbd: cleanup in utils.cc
...
Signed-off-by: huanwen ren <ren.huanwen@zte.com.cn>
2016-06-07 20:14:53 +08:00
Jason Dillaman
636129653c
librbd: flush journal commit positions before starting op
...
Ensure all IO has been properly flushed and committed to the
journal before starting an op that could affect the IO path.
Signed-off-by: Jason Dillaman <dillaman@redhat.com>
2016-06-07 07:30:48 -04:00
Jason Dillaman
d3d4e06839
librbd: leaked AioCompletion if IO issued after image shut down
...
Signed-off-by: Jason Dillaman <dillaman@redhat.com>
2016-06-07 07:30:48 -04:00
Jason Dillaman
1032f19b3c
librbd: resize and snap create can hang on journal replay failure
...
Signed-off-by: Jason Dillaman <dillaman@redhat.com>
2016-06-07 07:30:48 -04:00
Jason Dillaman
a7f5bc181c
librbd: missing journal state for flushing/restarting replay
...
Fixes: http://tracker.ceph.com/issues/16077
Signed-off-by: Jason Dillaman <dillaman@redhat.com>
2016-06-07 07:30:48 -04:00
Jason Dillaman
eb8dda5296
librbd: track IO in the order issued
...
Fixes: http://tracker.ceph.com/issues/15034
Signed-off-by: Jason Dillaman <dillaman@redhat.com>
2016-06-07 07:30:48 -04:00
Jason Dillaman
133b1ffb21
librbd: AioCompletion should retrieve CephContext from ImageCtx
...
Signed-off-by: Jason Dillaman <dillaman@redhat.com>
2016-06-07 07:30:48 -04:00
Jason Dillaman
82587a31f1
librbd: Journal no longer requires AioCompletion for IO events
...
Signed-off-by: Jason Dillaman <dillaman@redhat.com>
2016-06-07 07:30:48 -04:00
Jason Dillaman
79c934742f
librbd: bubble journal error up via existing AioObjectRequests
...
The journal cannot directly fail the AioCompletion because it
might have one-or-more outstanding requests associated with
it.
Fixes: http://tracker.ceph.com/issues/15791
Signed-off-by: Jason Dillaman <dillaman@redhat.com>
2016-06-07 07:30:48 -04:00
Jianpeng Ma
5e5bdadcbf
os/bluestore/BlueStore: remove the duplicated code.
...
Signed-off-by: Jianpeng Ma <jianpeng.ma@intel.com>
2016-06-07 15:17:49 +08:00
xie xingguo
45efd19722
client: fix sanity check of preadv/pwritev
...
The preadv/pwritev call is only supposed to return a postive
number indicating how many bytes we succeed in reading from
or writing into disk, which is obviously not what we want
under the following cases.
Signed-off-by: xie xingguo <xie.xingguo@zte.com.cn>
2016-06-07 14:26:45 +08:00
Kefu Chai
70bbf1aded
configure.ac,debian,rpm: do not package .a files in rpm/deb packages
...
this change removes *.a from -dev or -devel packages.
* configure.ac:
- use LT_INIT instead of AM_PROG_LIBTOOL. the later is deprecated.
- disable static lib:
.a static libraries not not useful, it's tricky to generate both
static libraries and shared libraries in cmake.
see http://www.spinics.net/lists/ceph-devel/msg30637.html
* ceph.spec.in: fedora packaging policy forbids this
see
https://fedoraproject.org/wiki/Packaging:Guidelines#Packaging_Static_Libraries
* debian/*-dev.install: debian does not requires this
see
https://www.debian.org/doc/debian-policy/ch-sharedlibs.html#s-sharedlibs-static
Signed-off-by: Kefu Chai <kchai@redhat.com>
2016-06-07 14:08:59 +08:00
Kefu Chai
4ba2bb750b
debian: do not install *.la files
...
rationales:
* .la files are worthless
see http://www.spinics.net/lists/ceph-devel/msg30637.html
* debian does not like it:
see https://wiki.debian.org/ReleaseGoals/LAFileRemoval
* neither does fedora:
https://fedoraproject.org/wiki/Packaging:Guidelines#Packaging_Static_Libraries
* also they are created by libtools, and cmake does not use libtools.
Signed-off-by: Kefu Chai <kchai@redhat.com>
2016-06-07 14:08:59 +08:00
Kefu Chai
d596146e6d
makefile,debian,rpm: do not package librgw_file* anymore
...
they are unit tests for rgw.
Signed-off-by: Kefu Chai <kchai@redhat.com>
2016-06-07 14:08:59 +08:00
xie xingguo
5c4a65a521
bluestore/BitAllocator: literal cleanups
...
Signed-off-by: xie xingguo <xie.xingguo@zte.com.cn>
2016-06-07 12:28:14 +08:00
Shehbaz Jaffer
c1561a3ab1
os/bluestore: benchmark Host Aware SMR Drives for linear writes
...
* add script to write files linearly using dd command
* add script to write files linearly using SMR zone aware commands
Script needs libzbc support. Check is provided so that script does
not run if libzbc commands are not installed.
Fixes: http://tracker.ceph.com/issues/16174
Signed-off-by: Shehbaz Jaffer <shehbazjaffer007@gmail.com>
2016-06-07 00:05:29 -04:00
Kefu Chai
8d7b8e1f93
Merge pull request #9296 from tchaikov/wip-test-chain_xattr
...
test: Fix the order of ASSERT_EQ() in chain_xattr tests
Reviewed-by: Brad Hubbard <bhubbard@redhat.com>
2016-06-07 12:05:26 +08:00
Kefu Chai
39725ea785
Merge pull request #9361 from stiopaa1/osd_osd_override
...
osd/OSD.h: add override to virtual functions
Reviewed-by: Kefu Chai <kchai@redhat.com>
2016-06-07 12:03:19 +08:00
Vikhyat Umrao
a00720d9aa
rbd: add error message "snapshot is already protected"
...
Fixes: http://tracker.ceph.com/issues/15807
Signed-off-by: Vikhyat Umrao <vumrao@redhat.com>
2016-06-07 09:06:25 +05:30
Kefu Chai
3caafc4476
buffer: add a test for iterator::copy(huge_chunk, dest)
...
Fixes: http://tracker.ceph.com/issues/16010
Signed-off-by: Kefu Chai <kchai@redhat.com>
2016-06-07 11:23:28 +08:00
Kefu Chai
053bfa650b
buffer: fix advance(unsigned) to avoid int underflow
...
by chaning
- the signature of advance(unsigned) to advance(size_t)
- the signature of seek(int) to seek(ssize_t)
Fixes: http://tracker.ceph.com/issues/16010
Signed-off-by: Kefu Chai <kchai@redhat.com>
2016-06-07 11:23:28 +08:00
Jason Dillaman
276fed6b70
librbd: potential duplicate snap removal can result in crash
...
Signed-off-by: Jason Dillaman <dillaman@redhat.com>
2016-06-06 22:34:30 -04:00
Jianjian Huo
5bb63ebdb6
os/bluestore/BlueFS: convert all read/write IO to direct IO.
...
RocksDB is the sole user of BlueFS, after BlueFS uses direct IO,
RocksDB will have to use its own block cache to cache raw data
blocks.
Signed-off-by: Jianjian Huo <jianjian.huo@ssi.samsung.com>
2016-06-06 18:11:41 -07:00
Jianjian Huo
8e5891306c
os/bluestore: convert read_buffered() to read_random() with buffered flag
...
New function will handle unaligned random read for both buffered or
direct read.
Signed-off-by: Jianjian Huo <jianjian.huo@ssi.samsung.com>
2016-06-06 18:11:30 -07:00
Jianpeng Ma
7ae1db2ba0
os/bluestore/BitAllocator: fix compiler warning.
...
Signed-off-by: Jianpeng Ma <jianpeng.ma@intel.com>
2016-06-07 07:42:28 +08:00
Jianpeng Ma
fca97f8619
os/bluestore/BitMapAllocator: fix free space beyond size of BitAllocator
...
When Using bitmap allocator, the osd crashed. The stack info as
follows:
ceph-osd: os/bluestore/BitAllocator.cc:910: bool
BitAllocator::is_allocated(int64_t, int64_t): Assertion `start_block >=
0 && (start_block + num_blocks <= size())' failed.
*** Caught signal (Aborted) **
in thread 7f795bbc48c0 thread_name:ceph-osd ceph version
10.2.0-1333-g3f4cf16
(3f4cf16
)
1: (()+0x98e25e) [0x55e112a9625e]
2: (()+0x109f0) [0x7f795a6719f0]
3: (gsignal()+0x38) [0x7f7958542a28]
4: (abort()+0x16a) [0x7f795854462a]
5: (()+0x2d227) [0x7f795853b227]
6: (()+0x2d2d2) [0x7f795853b2d2]
7: (()+0x7eaa59) [0x55e1128f2a59]
8: (BitAllocator::free_blocks(long, long)+0x22) [0x55e1128f2e82]
9: (BitMapAllocator::insert_free(unsigned long, unsigned long)+0x22c)
[0x55e1128ef12c]
10: (BitMapAllocator::init_add_free(unsigned long, unsigned long)+0x22c)
[0x55e1128ef41c]
11: (BlueFS::_init_alloc()+0x2b9) [0x55e1128c27b9]
12: (BlueFS::mkfs(uuid_d)+0x45a) [0x55e1128d475a]
13: (BlueStore::_open_db(bool)+0xd76) [0x55e1127bbcb6]
14: (BlueStore::mkfs()+0x8b1) [0x55e1127e4f11]
15: (OSD::mkfs(CephContext, ObjectStore, std::__cxx11::basic_string,
std::allocator > const&, uuid_d, int)+0x117) [0x55e112484de7]
16: (main()+0x101f) [0x55e112425f2f]
17: (__libc_start_main()+0xf0) [0x7f795852e580]
18: (_start()+0x29) [0x55e1124684e9]
This because in BitAllocator::init will decrease size of blkdev which
make size align w/ zone-size.
The later add free extent will beyond the size.
Signed-off-by: Jianpeng Ma <jianpeng.ma@intel.com>
2016-06-07 07:37:13 +08:00
Jianpeng Ma
add4832ca8
os/bluestore/BlueStore: remove the unused perfcounter:l_bluestore_state_wal_done_lat
...
Signed-off-by: Jianpeng Ma <jianpeng.ma@intel.com>
2016-06-07 06:29:08 +08:00
Casey Bodley
b262e46599
Merge pull request #9527 from linuxbox2/rgw-ldap
...
rgw ldap: fix ldap bindpw parsing
Reviewed-by: Casey Bodley <cbodley@redhat.com>
2016-06-06 17:55:04 -04:00
Matt Benjamin
72b2f3e8dc
rgw ldap: fix ldap bindpw parsing
...
Also add additional LDAP debugging output at 0, 10, and 15 to make
troubleshooting easier.
Fixes DN search issue using QE configuration of MS AD.
Signed-off-by: Matt Benjamin <mbenjamin@redhat.com>
2016-06-06 17:50:32 -04:00
Greg Farnum
5d9cc299e6
doc: include link to CephFS' experimental features
...
Signed-off-by: Greg Farnum <gfarnum@redhat.com>
2016-06-06 14:46:44 -07:00
Jianjian Huo
17129122ac
RocksDBStore: add option to configure number of block cache shards
...
rocksdb_cache_size will be splitted into number of shards evenly.
More shards will reduce lock contentions.
Signed-off-by: Jianjian Huo <jianjian.huo@ssi.samsung.com>
2016-06-06 14:44:41 -07:00
Michal Jarzabek
689819bab4
common/ceph_json.h: add const to member function
...
Signed-off-by: Michal Jarzabek <stiopa@gmail.com>
2016-06-06 21:55:46 +01:00
Samuel Just
f14c8ae61f
Merge pull request #9525 from liewegas/wip-wq
...
unittest_workqueue: fix wq test for 0 threads
Reviewed-by: Samuel Just <sjust@redhat.com>
2016-06-06 13:53:06 -07:00
Josh Durgin
27edd204aa
Merge pull request #9517 from idryomov/wip-rbd-concurrent
...
krbd: don't segfault if images are unmapped concurrently
Reviewed-by: Josh Durgin <jdurgin@redhat.com>
2016-06-06 12:57:37 -07:00
Samuel Just
2d79aa9cfc
Merge pull request #9432 from ceph/wip-16113
...
Clean up hobject_t::max() values in decode, avoid comparing directly to get_max()
Reviewed-by: Sage Weil <sage@redhat.com>
2016-06-06 12:23:46 -07:00
Sage Weil
9c7d248e0d
unittest_workqueue: fix wq test for 0 threads
...
In 81517aea06
we added
support for a wq with 0 worker threads. Fix unit test
accordingly.
Signed-off-by: Sage Weil <sage@redhat.com>
2016-06-06 14:20:30 -04:00
Sage Weil
f836280c46
Merge pull request #9522 from ceph/wip-rocksdb
...
Makefile-rocksdb: updates
2016-06-06 14:18:22 -04:00
Samuel Just
cef5151a62
hobject: enforce max canonical value
...
Signed-off-by: Samuel Just <sjust@redhat.com>
2016-06-06 09:34:09 -07:00
Samuel Just
7c5f2acaa5
src/: remove all direct comparisons to get_max()
...
get_max() now returns a special singleton type from which hobject_t's
can be assigned and constructed, but which cannot be directly compared.
This patch also cleans up all such uses to use is_max() instead.
This should prevent some issues like 16113 by preventing us from
checking for max-ness by comparing against a sentinel value. The more
complete fix will be to make all fields of hobject_t private and enforce
a canonical max() representation that way. That patch will be hard to
backport, however, so we'll settle for this for now.
Fixes: http://tracker.ceph.com/issues/16113
Signed-off-by: Samuel Just <sjust@redhat.com>
2016-06-06 09:34:04 -07:00
Jianpeng Ma
189c8f190b
os/bluestore/BlueStore: add perfcoutner:l_bluestore_write_penalty_read_ops
...
For write, because the align requirement it maybe need read data.
Using this perfcounter to record the read ops.
Signed-off-by: Jianpeng Ma <jianpeng.ma@intel.com>
2016-06-07 00:31:02 +08:00
Sage Weil
f3d68eb61c
Merge pull request #9518 from cbodley/wip-mds-override
...
mds: fix MDRequestImpl::print override
Reviewed-by: John Spray <jspray@redhat.com>
2016-06-06 12:11:34 -04:00
Jianpeng Ma
f8b7eb980d
os/bluestore/BlueStore: add perfcounter: l_bluestore_wal_write_ops/bytes.
...
Using this record wal ops and wal bytes.
Signed-off-by: Jianpeng Ma <jianpeng.ma@intel.com>
2016-06-06 23:45:53 +08:00
Jianpeng Ma
b3f353e2cf
os/bluestore/BlueStore: Add perfcounter::l_bluestore_write_pad_bytes.
...
For write, sometimes it need padd zero. Use this perfcounter to
record the padding bytes.
Signed-off-by: Jianpeng Ma <jianpeng.ma@intel.com>
2016-06-06 23:44:32 +08:00
Yan Jun
dc0ac7b304
rgw/basic_type: use the standard usage of string
...
string::find() returns a value of size_t type, we'd better
use the std::string::npos when it find nothing.
Signed-off-by: Yan Jun <yan.jun8@zte.com.cn>
2016-06-06 23:38:25 +08:00
Sage Weil
6f022acc85
Makefile-rocksdb: updates
...
Signed-off-by: Sage Weil <sage@redhat.com>
2016-06-06 11:16:52 -04:00