]>
git.apps.os.sepia.ceph.com Git - ceph.git/log
Yan, Zheng [Fri, 31 Jul 2015 09:18:58 +0000 (17:18 +0800)]
test/admin_socket: check error message according to OS
Signed-off-by: Yan, Zheng <zyan@redhat.com>
Yan, Zheng [Fri, 31 Jul 2015 02:01:01 +0000 (10:01 +0800)]
compat: move definitions in porting.h into include/compat.h
Signed-off-by: Yan, Zheng <zyan@redhat.com>
Yan, Zheng [Thu, 30 Jul 2015 10:15:18 +0000 (18:15 +0800)]
TestLFNIndex.cc: don't use "cp --preserve=xattr"
--preserve option is not supported by OSX's cp. use 'cp -a' instead
Signed-off-by: Yan, Zheng <zyan@redhat.com>
Yan, Zheng [Thu, 30 Jul 2015 08:03:56 +0000 (16:03 +0800)]
tests: replace std::tr1::shared_ptr with ceph::shared_ptr
Signed-off-by: Yan, Zheng <zyan@redhat.com>
Yan, Zheng [Thu, 30 Jul 2015 08:01:46 +0000 (16:01 +0800)]
tests: disable unittest_blkdev on OSX
the test uses sysfs, which does not exist on OSX
Signed-off-by: Yan, Zheng <zyan@redhat.com>
Yan, Zheng [Thu, 30 Jul 2015 06:31:20 +0000 (14:31 +0800)]
test/librados: replace sem_init() with sem_open()
sem_init() always fails on OSX (it's no longer supported).
Signed-off-by: Yan, Zheng <zyan@redhat.com>
Yan, Zheng [Wed, 29 Jul 2015 09:44:52 +0000 (17:44 +0800)]
buffer: make buffer::exception classes undefined in dynamic objects
On OSX, if the an exception class is declared and defined in header file,
but it ends up being compiled as private symbols in different binaries.
The exception handling code will take the two compiled exception classes
as different types! In our case, the one in libcls_xxx.so and the one is
ceph-osd are considered as different classes, thus the try-catch statement
fails to work.
The fix is force buffer::exception classes undefined in libcls_xxx.so. The
ibcls_xxx.so are compiled with '-undefined dynamic_lookup' option. when
it is loaded into ceph-osd, buffer::exception classes in ceph-osd will be
used.
Signed-off-by: Yan, Zheng <zyan@redhat.com>
Yan, Zheng [Tue, 28 Jul 2015 09:55:30 +0000 (17:55 +0800)]
test/librados: replace errno -125 with -ECANCELED
Signed-off-by: Yan, Zheng <zyan@redhat.com>
Yan, Zheng [Mon, 27 Jul 2015 11:08:48 +0000 (19:08 +0800)]
client: convert XATTR_{CREATE,REPLACE} to CEPH_XATTR_{CREATE,REPLACE}
Signed-off-by: Yan, Zheng <zyan@redhat.com>
Yan, Zheng [Mon, 27 Jul 2015 08:01:31 +0000 (16:01 +0800)]
xattr: convert ENOATTR to ENODATA on DARWIN
On Darwin, getxattr/removexattr return ENOATTR when extended
attribute does not exist. But lots of ceph codes expect ENODATA
is returned in that case. On Darwin, ENOATTR and ENODATA have
different values, so we convert ENOATTR to ENODATA.
Signed-off-by: Yan, Zheng <zyan@redhat.com>
Yan, Zheng [Fri, 24 Jul 2015 10:13:22 +0000 (18:13 +0800)]
test_libcephfs: disable flock test on OSX
there is no sem_timedwait on OSX. For inter-threads communication,
we can replace sem_t with pthread_cond_t. But for inter-processes
communication, it's extremely hard to figure out how to do a timed
wait. So disable the test case on OSX
Signed-off-by: Yan, Zheng <zyan@redhat.com>
Yan, Zheng [Fri, 24 Jul 2015 08:06:01 +0000 (16:06 +0800)]
libradosstriper/striping.cc: include "include/types.h"
include definitions of __le{32,64)
Signed-off-by: Yan, Zheng <zyan@redhat.com>
Yan, Zheng [Fri, 24 Jul 2015 08:03:48 +0000 (16:03 +0800)]
test_c_headers: don't use -Wold-style-declaration option for clang
clang does not support this option
Signed-off-by: Yan, Zheng <zyan@redhat.com>
Yan, Zheng [Fri, 24 Jul 2015 07:30:03 +0000 (15:30 +0800)]
libcephfs: define loff_t as off_t on OSX
Signed-off-by: Yan, Zheng <zyan@redhat.com>
Yan, Zheng [Fri, 24 Jul 2015 03:32:08 +0000 (11:32 +0800)]
tools: link ceph-client-debug to LIBCLIENT
Signed-off-by: Yan, Zheng <zyan@redhat.com>
Yan, Zheng [Thu, 23 Jul 2015 07:07:45 +0000 (15:07 +0800)]
init-ceph: check if /lib/lsb/init-functions exists
On OSX/FreeBSD, /lib/lsb/init-functions does not exist
Signed-off-by: Yan, Zheng <zyan@redhat.com>
Yan, Zheng [Thu, 23 Jul 2015 07:05:33 +0000 (15:05 +0800)]
init-ceph: don't use procfs to check if daemon is running
use ps(1) instead, which is portable
Signed-off-by: Yan, Zheng <zyan@redhat.com>
Yan, Zheng [Thu, 23 Jul 2015 02:58:05 +0000 (10:58 +0800)]
init-ceph: replace 'echo -n' with printf
'echo -n' is no portable. On OSX, output of 'echo -n' can be '-n'
Signed-off-by: Yan, Zheng <zyan@redhat.com>
Yan, Zheng [Thu, 23 Jul 2015 02:24:33 +0000 (10:24 +0800)]
vstart.sh: use portable way to get ip address
Signed-off-by: Yan, Zheng <zyan@redhat.com>
Yan, Zheng [Thu, 23 Jul 2015 02:20:26 +0000 (10:20 +0800)]
vstart.sh: append ceph library path to {LD,DYLD}_LIBRARY_PATH
Signed-off-by: Yan, Zheng <zyan@redhat.com>
Yan, Zheng [Thu, 23 Jul 2015 01:52:08 +0000 (09:52 +0800)]
On Darwin: subfix of dynamic library is dylib
Signed-off-by: Yan, Zheng <zyan@redhat.com>
Kefu Chai [Fri, 10 Jul 2015 03:53:55 +0000 (11:53 +0800)]
porting.h: add TODO comment
Signed-off-by: Kefu Chai <kchai@redhat.com>
Kefu Chai [Fri, 10 Jul 2015 03:52:51 +0000 (11:52 +0800)]
Makefile: add porting.h to dist tarball
Signed-off-by: Kefu Chai <kchai@redhat.com>
Yan, Zheng [Mon, 6 Jul 2015 01:20:33 +0000 (09:20 +0800)]
Don't use '--exclude-libs' linker option on DARWIN
Only GNU linker supports this option
Signed-off-by: Yan, Zheng <zyan@redhat.com>
Yan, Zheng [Sun, 5 Jul 2015 08:02:33 +0000 (16:02 +0800)]
client: don't try trimming kernel dcache on DARWIN/FreeBSD
still don't know if the methods used on linux also works on
DARWIN/FreeBSD
Signed-off-by: Yan, Zheng <zyan@redhat.com>
Yan, Zheng [Sun, 5 Jul 2015 07:54:27 +0000 (15:54 +0800)]
client: don't include unsupported mount options on DARWIN/FreeBSD
Signed-off-by: Yan, Zheng <zyan@redhat.com>
Yan, Zheng [Sun, 5 Jul 2015 07:49:49 +0000 (15:49 +0800)]
Link ceph-fuse to fuse on DARWIN
Signed-off-by: Yan, Zheng <zyan@redhat.com>
Yan, Zheng [Sun, 5 Jul 2015 07:31:27 +0000 (15:31 +0800)]
msg: fix encoding/decoding sockaddr_storage on DARWIN/FreeBSD
sockaddr_storage on DARWIN/FreeBSD is different from sockaddr_storage
on Linux. sockaddr_storage on DARWIN/FreeBSD includes a ss_len field.
Besides, Its ss_len and ss_family fields are 'unsigned char'.
Signed-off-by: Yan, Zheng <zyan@redhat.com>
Dennis Schafroth [Tue, 30 Jun 2015 18:39:50 +0000 (20:39 +0200)]
librbd: Need to include errno.h on DARWIN.
The file is not really required
Signed-off-by: Dennis Schafroth <dennis@schafroth.com>
Dennis Schafroth [Tue, 30 Jun 2015 18:39:03 +0000 (20:39 +0200)]
porting.h: add porting.h for porting/compatibility on DARWIN
Collection of porting/compatibility defines to compile on DARWIN
Signed-off-by: Dennis Schafroth <dennis@schafroth.com>
Dennis Schafroth [Tue, 30 Jun 2015 18:35:32 +0000 (20:35 +0200)]
tools/ceph_objectstore_tool: Missing O_LARGEFILE on DARWIN
Signed-off-by: Dennis Schafroth <dennis@schafroth.com>
Dennis Schafroth [Tue, 30 Jun 2015 18:15:23 +0000 (20:15 +0200)]
test: Fix error with clang on DARWIN (LLVM 3.6.0svn)
Signed-off-by: Dennis Schafroth <dennis@schafroth.com>
Dennis Schafroth [Tue, 30 Jun 2015 18:13:10 +0000 (20:13 +0200)]
rbd-replay: Different location of endian.h on DARWIN
Signed-off-by: Dennis Schafroth <dennis@schafroth.com>
Dennis Schafroth [Tue, 30 Jun 2015 18:12:37 +0000 (20:12 +0200)]
rbd-fuse: Add position to set/get xattr on DARWIN
Signed-off-by: Dennis Schafroth <dennis@schafroth.com>
Dennis Schafroth [Tue, 30 Jun 2015 18:10:19 +0000 (20:10 +0200)]
os/FileStore: fail with ENOTSUP if using sparse files on DARWIN
For now, fail with Not supported (ENOTSUP) if configured using sparse files on DARWIN.
Signed-off-by: Dennis Schafroth <dennis@schafroth.com>
Dennis Schafroth [Tue, 30 Jun 2015 18:07:32 +0000 (20:07 +0200)]
librbd: Include porting header
Signed-off-by: Dennis Schafroth <dennis@schafroth.com>
Dennis Schafroth [Tue, 30 Jun 2015 17:55:10 +0000 (19:55 +0200)]
common/xattr: Split out xattr on Linux and DARWIN.
DARWIN requires position parameter
Signed-off-by: Dennis Schafroth <dennis@schafroth.com>
Dennis Schafroth [Tue, 30 Jun 2015 17:52:17 +0000 (19:52 +0200)]
common/util: include sys/param.h and mount.h on DARWIN
Signed-off-by: Dennis Schafroth <dennis@schafroth.com>
Dennis Schafroth [Tue, 30 Jun 2015 17:51:32 +0000 (19:51 +0200)]
common/SubProcess: fix build on DARWIN
include <signal.h> to avoid errors/warnings on missing functions
Signed-off-by: Dennis Schafroth <dennis@schafroth.com>
Dennis Schafroth [Tue, 30 Jun 2015 17:50:27 +0000 (19:50 +0200)]
ceph_fuse: Implement set/get xattr with position parameter on OSX
Signed-off-by: Dennis Schafroth <dennis@schafroth.com>
Dennis Schafroth [Tue, 30 Jun 2015 17:49:04 +0000 (19:49 +0200)]
msg/async: Include porting.h for MSG_*
Signed-off-by: Dennis Schafroth <dennis@schafroth.com>
Dennis Schafroth [Tue, 30 Jun 2015 17:47:24 +0000 (19:47 +0200)]
msg/simple: Move MSG_ and SO_NOSIGPIPE into porting.h
Only enable SO_PRIORITY on Linux
Signed-off-by: Dennis Schafroth <dennis@schafroth.com>
Kefu Chai [Sat, 29 Aug 2015 14:32:20 +0000 (22:32 +0800)]
Merge pull request #5710 from tchaikov/wip-check-boost_random-mt
configure.ac: check for libboost_random-mt also
Reviewed-by: Loic Dachary <ldachary@redhat.com>
Sage Weil [Sat, 29 Aug 2015 12:24:01 +0000 (08:24 -0400)]
Merge pull request #5711 from dachary/wip-12781-dmcrypt
ceph-disk: fix dmcrypt typo
Reviewed-by: Kefu Chai <kchai@redhat.com>
Loic Dachary [Thu, 27 Aug 2015 10:51:14 +0000 (12:51 +0200)]
ceph-disk: fix dmcrypt typo
Fix the typo introduced by
29431944c77adbc3464a8faeb7e052b24f821780
http://tracker.ceph.com/issues/12781 Fixes: #12781
Signed-off-by: Loic Dachary <ldachary@redhat.com>
Loic Dachary [Sat, 29 Aug 2015 09:03:02 +0000 (11:03 +0200)]
Merge pull request #4731 from stiopaa1/cephdetectinitfix
ceph-detect-init/debian/__init__: improved syntax
Reviewed-by: Loic Dachary <ldachary@redhat.com>
Loic Dachary [Sat, 29 Aug 2015 08:58:12 +0000 (10:58 +0200)]
Merge pull request #5503 from zhouyuan/isal_2.14_update
erasure-code: Update ISA-L to 2.14
Reviewed-by: Loic Dachary <ldachary@redhat.com>
Kefu Chai [Sat, 29 Aug 2015 08:49:45 +0000 (16:49 +0800)]
configure.ac: check for libboost_random-mt also
Signed-off-by: Kefu Chai <kchai@redhat.com>
Loic Dachary [Sat, 29 Aug 2015 08:47:26 +0000 (10:47 +0200)]
Merge pull request #5467 from dachary/wip-ceph-release-notes
tools: ceph-release-notes unicode handling
Reviewed-by: Abhishek Lekshmanan <abhishek.lekshmanan@ril.com>
Samuel Just [Fri, 28 Aug 2015 21:49:15 +0000 (14:49 -0700)]
Merge pull request #5333 from ceph/wip-12437
osd: Keep a reference count on Connection while calling send_mes…
Reviewed-by: Samuel Just <sjust@redhat.com>
Reviewed-by: Kefu Chai <kchai@redhat.com>
Sage Weil [Fri, 28 Aug 2015 19:20:04 +0000 (15:20 -0400)]
Merge remote-tracking branch 'gh/infernalis'
Sage Weil [Fri, 28 Aug 2015 17:19:23 +0000 (13:19 -0400)]
Merge pull request #5690 from dzafman/wip-pgls-zafman
cls: Fix successful return found by compiler warning
Reviewed-by: Sage Weil <sage@redhat.com>
Sage Weil [Fri, 28 Aug 2015 16:20:18 +0000 (12:20 -0400)]
Merge pull request #5704 from tchaikov/wip-adventure-mode
common: 'enable experimental data corrupting features' now understand '*'
Reviewed-by: Sage Weil <sage@redhat.com>
Samuel Just [Fri, 28 Aug 2015 16:18:16 +0000 (09:18 -0700)]
Merge pull request #5667 from liewegas/wip-9221
ceph_test_rados_api_tier: make PromoteOn2ndRead tolerate thrashing
Reviewed-by: Samuel Just <sjust@redhat.com>
Samuel Just [Fri, 28 Aug 2015 16:17:54 +0000 (09:17 -0700)]
Merge pull request #5666 from liewegas/wip-12785
osd: fix off-by-one in hobject_t::is_temp() (bug 12785)
Reviewed-by: Samuel Just <sjust@redhat.com>
Kefu Chai [Fri, 28 Aug 2015 16:08:10 +0000 (00:08 +0800)]
common: 'enable experimental data corrupting features' now understands '*'
'*' allows any feature.
Signed-off-by: Kefu Chai <kchai@redhat.com>
Sage Weil [Fri, 28 Aug 2015 16:14:41 +0000 (12:14 -0400)]
Merge pull request #5609 from rohanmars/wip-timegm
added boost timegm impl for cross platform support
In the future it would be nice to avoid duplication of the code in each compiled module, but this works for now. The only real user right now is rgw.
Reviewed-by: Sage Weil <sage@redhat.com>
Sage Weil [Fri, 28 Aug 2015 16:13:57 +0000 (12:13 -0400)]
Merge pull request #5627 from wonzhq/agent-work
osd: do evict before flush in agent_work
Reviewed-by: Sage Weil <sage@redhat.com>
Sage Weil [Fri, 28 Aug 2015 16:13:38 +0000 (12:13 -0400)]
Merge pull request #5640 from liewegas/wip-mutex-lock
mutex: avoid trylock unless instrumentation is enabled
Reviewed-by: Samuel Just <sjust@redhat.com>
Reviewed-by: Piotr Dałek <piotr.dalek@ts.fujitsu.com>
Sage Weil [Fri, 28 Aug 2015 16:12:45 +0000 (12:12 -0400)]
Merge pull request #5656 from XinzeChi/wip-buffer-emtpyptr
common: fix insert empty ptr when bufferlist rebuild
Reviewed-by: Kefu Chai <kchai@redhat.com>
Reviewed-by: Sage Weil <sage@redhat.com>
Kefu Chai [Fri, 28 Aug 2015 15:45:25 +0000 (23:45 +0800)]
Merge pull request #5453 from wonzhq/recover-primary
osd: recovery fixes
Reviewed-by: Sage Weil <sage@redhat.com>
Kefu Chai [Fri, 28 Aug 2015 15:40:41 +0000 (23:40 +0800)]
Merge pull request #5415 from wonzhq/do-recovery
osd: do_recovery cleanup
Reviewed-by: Sage Weil <sage@redhat.com>
Reviewed-by: Kefu Chai <kchai@redhat.com>
Sage Weil [Fri, 28 Aug 2015 15:14:31 +0000 (11:14 -0400)]
Merge pull request #5703 from ceph/wip-boost-random-cmake
CMake: update for boost_random
John Spray [Fri, 28 Aug 2015 14:33:12 +0000 (15:33 +0100)]
CMake: update for boost_random
This became a dependency in
dbcaa544
Signed-off-by: John Spray <john.spray@redhat.com>
Loic Dachary [Fri, 28 Aug 2015 10:23:19 +0000 (12:23 +0200)]
Merge pull request #5701 from guce/h3c_mail_organization_map
h3c mail organization map
Reviewed-by: Loic Dachary <ldachary@redhat.com>
guce [Fri, 28 Aug 2015 09:56:19 +0000 (17:56 +0800)]
h3c mail organization map
update .mailmap and .organizationmap for h3c organization
Signed-off-by: Ce Gu <guce@h3c.com>
Kefu Chai [Fri, 28 Aug 2015 02:08:56 +0000 (10:08 +0800)]
Merge pull request #5114 from branch-predictor/bp-reduced-memzero
ObjectStore/FileJournal: reduce impact of memset()
Reviewed-by: Sage Weil <sage@redhat.com>
David Zafman [Thu, 27 Aug 2015 23:21:50 +0000 (16:21 -0700)]
cls: Fix successful return found by compiler warning
Signed-off-by: David Zafman <dzafman@redhat.com>
David Zafman [Thu, 27 Aug 2015 19:58:47 +0000 (12:58 -0700)]
Merge pull request #5688 from dzafman/wip-decode-zafman
osd: Decode use_gmt_hitset with a unique version
Reviewed-by: Sage Weil <sage@redhat.com>
Samuel Just [Thu, 27 Aug 2015 18:59:49 +0000 (11:59 -0700)]
Merge pull request #5679 from wonzhq/store-test
test: disable newstore test until it's merged
Reviewed-by: Samuel Just <sjust@redhat.com>
David Zafman [Thu, 27 Aug 2015 18:24:25 +0000 (11:24 -0700)]
osd: Decode use_gmt_hitset with a unique version
Signed-off-by: David Zafman <dzafman@redhat.com>
John Spray [Thu, 27 Aug 2015 17:14:37 +0000 (18:14 +0100)]
Merge pull request #5545 from ceph/wip-objclass-filters
Allow object classes to define PGLS filters
Reviewed-by: David Zafman <dzafman@redhat.com>
Sage Weil [Thu, 27 Aug 2015 17:05:40 +0000 (13:05 -0400)]
Merge remote-tracking branch 'gh/wip-osd-compat'
Reviewed-by: Samuel Just <sjust@redhat.com>
Sage Weil [Thu, 27 Aug 2015 16:41:54 +0000 (12:41 -0400)]
Merge pull request #5683 from tchaikov/wip-hitset-gmt-yet-another-feature-bit
osd: do not let OSD_HITSET_GMT reuse the feature bit
Reviewed-by: Sage Weil <sage@redhat.com>
Kefu Chai [Thu, 27 Aug 2015 16:32:38 +0000 (00:32 +0800)]
osd: do not let OSD_HITSET_GMT reuse the feature bit
* to ease the backport to hammer
Signed-off-by: Kefu Chai <kchai@redhat.com>
Joao Eduardo Luis [Thu, 27 Aug 2015 15:21:16 +0000 (16:21 +0100)]
Merge pull request #5524 from tchaikov/wip-monstore-cache
mon: add a cache layer over MonitorDBStore
Reviewed-by: Joao Eduardo Luis <joao@suse.de>
Reviewed-by: Sage Weil <sage@redhat.com>
Sage Weil [Thu, 27 Aug 2015 14:10:55 +0000 (10:10 -0400)]
Merge pull request #5669 from yuyuyu101/wip-12745
ms/simple: Messenger/MessengerTest.SyntheticInjectTest2/1 hang
Reviewed-by: Sage Weil <sage@redhat.com>
Sage Weil [Thu, 27 Aug 2015 14:06:07 +0000 (10:06 -0400)]
doc/release-notes: v0.94.3
Fix up the release timeline link for v9.0.3 too.
Signed-off-by: Sage Weil <sage@redhat.com>
John Spray [Thu, 27 Aug 2015 13:12:59 +0000 (14:12 +0100)]
osd: separate filter init from construction
...so that implementations can readily handle
decode errors in client args and return an
error code.
Signed-off-by: John Spray <john.spray@redhat.com>
John Spray [Tue, 11 Aug 2015 12:30:19 +0000 (13:30 +0100)]
test: add a test for filter in cls hello
Signed-off-by: John Spray <john.spray@redhat.com>
John Spray [Tue, 11 Aug 2015 12:17:22 +0000 (13:17 +0100)]
cls: add a filter to the hello class for testing
Signed-off-by: John Spray <john.spray@redhat.com>
John Spray [Tue, 11 Aug 2015 14:56:59 +0000 (15:56 +0100)]
objclass: enable unregistering filter factory
Signed-off-by: John Spray <john.spray@redhat.com>
Haomai Wang [Wed, 26 Aug 2015 02:49:32 +0000 (10:49 +0800)]
AsyncConnection: Fix uninitialized variable compile warning
Signed-off-by: Haomai Wang <haomaiwang@gmail.com>
Haomai Wang [Wed, 26 Aug 2015 02:36:10 +0000 (10:36 +0800)]
TestMsgr: Fix forever hang under lossless policy and one is WAIT another down
Signed-off-by: Haomai Wang <haomaiwang@gmail.com>
Orit Wasserman [Thu, 27 Aug 2015 08:40:23 +0000 (10:40 +0200)]
Merge pull request #5678 from ceph/wip-12208
rgw: delete finisher only after finalizing watches
Orit Wasserman [Thu, 27 Aug 2015 08:39:47 +0000 (10:39 +0200)]
Merge pull request #5677 from ceph/wip-11455
rgw: init some manifest fields when handling explicit objs
Orit Wasserman [Thu, 27 Aug 2015 08:38:36 +0000 (10:38 +0200)]
Merge pull request #5651 from theanalyst/rgw/fixiso8601
rgw: be more flexible with iso8601 timestamps
Zhiqiang Wang [Thu, 27 Aug 2015 04:27:52 +0000 (12:27 +0800)]
test: don't unmount when no store is created
When store is not created in SetUp for some reasons, unmounting it would
lead to segment fault.
Signed-off-by: Zhiqiang Wang <zhiqiang.wang@intel.com>
Zhiqiang Wang [Thu, 27 Aug 2015 03:17:17 +0000 (11:17 +0800)]
test: handle the case when ObjectStore::create returns NULL
When the objectstore type doesn't exist, NULL is returned in SetUp.
Handle the NULL return code to avoid a segment fault.
Signed-off-by: Zhiqiang Wang <zhiqiang.wang@intel.com>
Zhiqiang Wang [Thu, 27 Aug 2015 03:05:58 +0000 (11:05 +0800)]
test: disable newstore test until it's merged
Newstore hasn't been merged. It leads to a segment fault in one of the
teuthology testing job since ObjectStore::create() returns NULL.
Signed-off-by: Zhiqiang Wang <zhiqiang.wang@intel.com>
Sage Weil [Thu, 27 Aug 2015 00:41:05 +0000 (20:41 -0400)]
Merge pull request #4456 from ceph/wip-user
run ceph daemons as user and group ceph
Sage Weil [Thu, 27 Aug 2015 00:38:53 +0000 (20:38 -0400)]
Merge pull request #5674 from ceph/wip-selinux-post-script
ceph.spec.in: Restart services only if they are running
Reviewed-by: Milan Broz <mbroz@redhat.com>
Sage Weil [Tue, 25 Aug 2015 16:32:22 +0000 (12:32 -0400)]
debian: /var/run/ceph should be owned by ceph:ceph
Fix upstart and sysvinit scripts to create /var/run/ceph properly.
Chown existing dir on upgrade.
Signed-off-by: Sage Weil <sage@redhat.com>
Sage Weil [Wed, 19 Aug 2015 23:20:38 +0000 (19:20 -0400)]
upstart: setuser ceph
Signed-off-by: Sage Weil <sage@redhat.com>
Sage Weil [Wed, 19 Aug 2015 21:20:13 +0000 (17:20 -0400)]
debian/ceph-common.dirs: install /var/lib/ceph
It is the ceph user's home dir and we need to set the ownership.
Signed-off-by: Sage Weil <sage@redhat.com>
Sage Weil [Tue, 18 Aug 2015 18:50:42 +0000 (14:50 -0400)]
debian/ceph-common.postinst: fix adduser, addgroup
Drop the extra groups thing--it's broken anyway.
Signed-off-by: Sage Weil <sage@redhat.com>
Sage Weil [Tue, 18 Aug 2015 17:46:34 +0000 (13:46 -0400)]
debian/ceph-common.postinst: fix /var/log/ceph permissions
Signed-off-by: Sage Weil <sage@redhat.com>
Sage Weil [Tue, 18 Aug 2015 13:12:51 +0000 (09:12 -0400)]
debian: rename ceph-common.postinst
There was a trailing space!
Signed-off-by: Sage Weil <sage@redhat.com>
Sage Weil [Wed, 12 Aug 2015 20:54:55 +0000 (16:54 -0400)]
PendingReleaseNotes: more notes about the 'ceph' user
Signed-off-by: Sage Weil <sage@redhat.com>
Sage Weil [Wed, 12 Aug 2015 20:50:31 +0000 (16:50 -0400)]
global_init: ignore --set{user,group} if not root
Assume these options can only drop privileges; normal users can't setuid
root (and even if they could we wouldn't want to).
This makes ceph-disk behave when it sees the ceph user as 'root' and
invokes things with --setuser root but runs as a non-root user.
Signed-off-by: Sage Weil <sage@redhat.com>