Commit Graph

80428 Commits

Author SHA1 Message Date
Sage Weil
7f0d82d93b
Merge pull request #18826 from liewegas/wip-22090
qa/suite/rados: fix balancer vs firefly tunables failures

Reviewed-by: Kefu Chai <kchai@redhat.com>
2017-11-12 19:50:14 -06:00
Sage Weil
ef54a059e6
Merge pull request #18805 from liewegas/wip-22039
os/bluestore: fix SharedBlob unregistration

Reviewed-by: Igor Fedotov <ifedotov@mirantis.com>
2017-11-12 16:50:30 -06:00
Sage Weil
0fd48dddd6
Merge pull request #18878 from liewegas/wip-perf-health
qa/suites/rados/perf: whitelist health warnings
2017-11-12 16:50:08 -06:00
Sage Weil
7e9c888c83
Merge pull request #18884 from liewegas/wip-internal-safe-to-start-threads
common/config: make internal_safe_to_start_threads internal

Reviewed-by: Kefu Chai <kchai@redhat.com>
2017-11-12 16:49:13 -06:00
Mykola Golub
095ee4ba7d
Merge pull request #18825 from dillaman/wip-release-notes-rbd-caps
doc/release-notes: clarify that you need to keep your existing OSD caps

Reviewed-by: Mykola Golub <to.my.trociny@gmail.com>
2017-11-12 17:22:48 +02:00
Sage Weil
9e2fc91d2d common/config: make internal_safe_to_start_threads internal
There is no reason for this to be exposed like a normal config option
(even a legacy one) since it is an internal safety flag.

We do keep the clear_() helper there for the unit test.

Signed-off-by: Sage Weil <sage@redhat.com>
2017-11-10 17:25:25 -06:00
Casey Bodley
16e3439423
Merge pull request #18404 from cbodley/wip-vstart-mrgw
rgw: mrgw.sh uses instance name 'client.rgw'

Reviewed-by: Orit Wasserman <owasserm@redhat.com>
2017-11-10 14:20:26 -05:00
Yuri Weinstein
54b657ec0d
Merge pull request #18821 from prallabh/wip-22084
rgw: Fix swift object expiry not deleting objects

Reviewed-by: Casey Bodley <cbodley@redhat.com>
2017-11-10 08:04:31 -08:00
Yuri Weinstein
c6d5612063
Merge pull request #18846 from yaozongyou/fix-lifecycle-bug
rgw: check going_down() when lifecycle processing

Reviewed-by: Casey Bodley <cbodley@redhat.com>
2017-11-10 08:04:11 -08:00
Sage Weil
c1c81a04f5 qa/suites/rados/perf: whitelist health warnings
We may see these as cluster stabilizes.

Signed-off-by: Sage Weil <sage@redhat.com>
2017-11-10 09:00:39 -06:00
Haomai Wang
39da34bfba
Merge pull request #18877 from Jeegn-Chen/wip-issue-22103
rados: make ceph_perf_msgr_client work for multiple jobs

Reviewed-by: Haomai Wang <haomai@xsky.com>
2017-11-10 08:43:13 -06:00
Sage Weil
6455954d29 qa/suites/rados: stop testing firefly tunables
We can't mix the balancer compat-set testing with firefly tunables because
it requires that all buckets be straw2.

Signed-off-by: Sage Weil <sage@redhat.com>
2017-11-10 07:10:11 -06:00
Sage Weil
476b2335fd mgr/balancer: fail a bit more gracefully if we can't create a compat weight-set
Signed-off-by: Sage Weil <sage@redhat.com>
2017-11-10 07:10:11 -06:00
Jeegn Chen
5c1a860d61 rados: make ceph_perf_msgr_client work for multiple jobs
Move the invocation of msgr->start() before a client initiates a connection.
So the nonce is probably initialized and the server side is able to differenciate
the connections from different Messengers from the same client.

Fixes: http://tracker.ceph.com/issues/22103
Signed-off-by: Jeegn Chen <jeegnchen@gmail.com>
2017-11-10 20:59:01 +08:00
Abhishek L
f9d8317cbc
Merge pull request #18876 from smithfarm/wip-rn-typos
doc: release notes: fix grammar/style nits

Reviewed-By: Abhishek Lekshmanan <abhishek@suse.com>
2017-11-10 13:34:20 +01:00
Nathan Cutler
d12ce99fd4 doc: release notes: fix grammar/style nits
Signed-off-by: Nathan Cutler <ncutler@suse.com>
2017-11-10 11:49:14 +01:00
Haomai Wang
8ccaa64bf1
Merge pull request #18862 from Jeegn-Chen/wip-issue-22100
rados: Copy payload in ceph_perf_msgr_client

Reviewed-by: Haomai Wang <haomai@xsky.com>
2017-11-09 21:47:28 -06:00
Jeegn Chen
f1625e8905 rados: Copy payload in ceph_perf_msgr_client
Copy payload in ceph_perf_msgr_client so that payload can keep unchanged

Fixes: http://tracker.ceph.com/issues/22100
Signed-off-by: Jeegn Chen <jeegnchen@gmail.com>
2017-11-10 10:14:50 +08:00
Sage Weil
9ef5f20bf4
Merge pull request #18707 from liewegas/wip-filestore-rocksdb-compression
os/filestore: disable rocksdb compression

Reviewed-by: Douglas Fuller <dfuller@redhat.com>
2017-11-09 20:08:04 -06:00
Andrew Schoen
cbf06de900
Merge pull request #18823 from ceph/wip-bz1498200
ceph-volume: support GPT and other deployed OSDs

Reviewed-by: Andrew Schoen <aschoen@redhat.com>
2017-11-09 12:01:12 -06:00
Sage Weil
78a465b90b os/bluestore: fix SharedBlob unregistration
We use the SharedBlobSet remove() in three cases:

- from SharedBlob::put(), we try to remove ourselves from the set, but
  have to deal with a racing lookup, so the removal is conditional on
  nref still being 0.
- from split_cache(), we move the SharedBlob to another collection
- from make_blob_unshared(), we remove the entry when we clear the sbid.

The problem is that the condtiional remove() (for the first case) was being
used for all three cases, and in the second two cases nref is always != 0,
so it doesn't actually happen.  This can lead to a crash during cache
shutdown.

Fix by making two variants: remove() that is unconditional, and
try_remove() that is conditional.

Set the sb->coll pointer after because remove() asserts the parent matches
where we are unregistering.

Fixes: http://tracker.ceph.com/issues/22039
Signed-off-by: Sage Weil <sage@redhat.com>
2017-11-09 11:57:13 -06:00
Yuri Weinstein
89c26b9838
Merge pull request #18817 from ZVampirEM77/wip-em-zone-cleanup
rgw cleanup: some unnecessary function called and repeated assignment

Reviewed-by: Casey Bodley <cbodley@redhat.com>
2017-11-09 09:36:26 -08:00
Sage Weil
3855d2cbfd
Merge pull request #18683 from tchaikov/wip-erasure-cleanup
erasure-code: refactor the interfaces to hide internals from public

Reviewed-by: Sage Weil <sage@redhat.com>
2017-11-09 11:10:27 -06:00
Alfredo Deza
e2f281c115 doc/ceph-volume include the top level systemd.rst in the toctree
Signed-off-by: Alfredo Deza <adeza@redhat.com>
2017-11-09 12:02:16 -05:00
Alfredo Deza
ff60592897 ceph-volume simple.activate add --file in help docstring as an example
Signed-off-by: Alfredo Deza <adeza@redhat.com>
2017-11-09 12:02:16 -05:00
Alfredo Deza
29913951b7 doc/ceph-volume activate has a --file option for json files
Signed-off-by: Alfredo Deza <adeza@redhat.com>
2017-11-09 12:02:16 -05:00
Alfredo Deza
366fb8c43a ceph-volume simple.activate enable the osd when activating
Signed-off-by: Alfredo Deza <adeza@redhat.com>
2017-11-09 12:02:16 -05:00
Alfredo Deza
cb470622fd ceph-volume systemd.systemctl always force symlink when masking ceph-disk
Signed-off-by: Alfredo Deza <adeza@redhat.com>
2017-11-09 12:02:16 -05:00
Alfredo Deza
c954c91339 ceph-volume simple.activate use the contents of the JSON file, fallback to CLI args
Signed-off-by: Alfredo Deza <adeza@redhat.com>
2017-11-09 12:02:15 -05:00
Alfredo Deza
826c2709dd ceph-volume simple.activate only check for missing id and uuid if file is not passed in
Signed-off-by: Alfredo Deza <adeza@redhat.com>
2017-11-09 12:02:15 -05:00
Alfredo Deza
bd3498f3ea ceph-volume simple.activate allow to pass a file to activate
Signed-off-by: Alfredo Deza <adeza@redhat.com>
2017-11-09 12:02:15 -05:00
Alfredo Deza
d7fc26ab4e ceph-volume tests.util add tests for OSDPath validator
Signed-off-by: Alfredo Deza <adeza@redhat.com>
2017-11-09 12:02:15 -05:00
Alfredo Deza
cda47c247b ceph-volume systemd.systemctl masking ceph-disk needs to be done by directly symlinking to dev/null
Signed-off-by: Alfredo Deza <adeza@redhat.com>
2017-11-09 12:02:15 -05:00
Alfredo Deza
bad2f62b34 ceph-volume systemd.systemctl masking needs to use systemctl mask, not disable
Signed-off-by: Alfredo Deza <adeza@redhat.com>
2017-11-09 12:02:15 -05:00
Alfredo Deza
6b2fc4c0ba ceph-volume simple.scan if mounted device cannot be detected, raise an error
Signed-off-by: Alfredo Deza <adeza@redhat.com>
2017-11-09 12:02:14 -05:00
Alfredo Deza
230faf04a1 ceph-volume simple.activate remove str formatting from logging
Signed-off-by: Alfredo Deza <adeza@redhat.com>
2017-11-09 12:02:14 -05:00
Alfredo Deza
b5acf4d720 ceph-volume util.arg_validators always convert the path to an absolute path
Signed-off-by: Alfredo Deza <adeza@redhat.com>
2017-11-09 12:02:14 -05:00
Pavan Rallabhandi
70adfaae50 rgw: Fix swift object expiry not deleting objects
In cls_timeindex_list() though `to_index` has expired for a timespan, the marker is set for a subsequent index during the time boundary check.
This marker is further returned to RGWObjectExpirer::process_single_shard(), where this out_marker is trimmed from the respective shard,
resulting in a lost removal hint and a leaked object.

Fixes: http://tracker.ceph.com/issues/22084
Signed-off-by: Pavan Rallabhandi <PRallabhandi@walmartlabs.com>
2017-11-09 21:40:00 +05:30
Casey Bodley
36d83f16e0
Merge pull request #18781 from iliul/correct-error-messages
rgw/rgw_op.cc: Fix error message in rgw_user_get_all_buckets_stats

Reviewed-by: Casey Bodley <cbodley@redhat.com>
2017-11-09 09:48:26 -05:00
Casey Bodley
48d612001a
Merge pull request #18796 from ivancich/wip-fix-fmt-warnings
rgw: address warnings due to incorrect format code

Reviewed-by: Casey Bodley <cbodley@redhat.com>
Reviewed-by: Orit Wasserman <owasserm@redhat.com>
2017-11-09 09:43:33 -05:00
Kefu Chai
e2ade46c84
Merge pull request #17659 from majianpeng/osd-wait-ratherthan-waitinterval
osd/OSD: Using Wait rather than WaitInterval to wait queue.is_empty().

Reviewed-by: Sage Weil <sage@redhat.com>
Reviewed-by: Kefu Chai <kchai@redhat.com>
2017-11-09 22:18:47 +08:00
Kefu Chai
a8f18c1539
Merge pull request #18759 from tchaikov/wip-mon-log-last
mon/LogMonitor: "log last" should return up to n entries

Reviewed-by: John Spray <john.spray@redhat.com>
2017-11-09 22:17:31 +08:00
Kefu Chai
cb58b34271
Merge pull request #18779 from tchaikov/wip-osd-remove-unused-objects_read_async
osd: remove unused ReplicatedBackend::objects_read_async()

Reviewed-by: xie xingguo <xie.xingguo@zte.com.cn>
Reviewed-by: Xinze Chi <xinze@xsky.com>
2017-11-09 22:16:51 +08:00
Kefu Chai
b4f552d8cd
Merge pull request #18809 from tchaikov/wip-21904
mgr/zabbix: ignore osd with 0 kb capacity

Reviewed-by: John Spray <john.spray@redhat.com>
Reviewed-by: Wido den Hollander <wido@42on.com>
2017-11-09 22:15:06 +08:00
iliul
d60b57ef9c rgw/rgw_op.cc: Fix error message in rgw_user_get_all_buckets_stats
Signed-off-by: iliul <liul.stone@gmail.com>
2017-11-09 13:36:43 +00:00
Yao Zongyou
b761989033 rgw: check going_down() when lifecycle processing
Fixes: http://tracker.ceph.com/issues/22099

Signed-off-by: Yao Zongyou <yaozongyou@vip.qq.com>
2017-11-09 20:40:15 +08:00
Alfredo Deza
376dad7415 ceph-volume util.arg_validators require superuser privs to check paths
Signed-off-by: Alfredo Deza <adeza@redhat.com>
2017-11-09 06:47:43 -05:00
Alfredo Deza
6bc1e9ee0b ceph-volume simple.scan skip directories to scan
Signed-off-by: Alfredo Deza <adeza@redhat.com>
2017-11-09 06:47:43 -05:00
Alfredo Deza
cde0d51756 ceph-volume tests.devices.simple add activate CLI unit tests
Signed-off-by: Alfredo Deza <adeza@redhat.com>
2017-11-09 06:47:43 -05:00
Alfredo Deza
dcc7c72196 ceph-volume tests.devices add help menu unit test
Signed-off-by: Alfredo Deza <adeza@redhat.com>
2017-11-09 06:47:43 -05:00