1
0
mirror of https://github.com/ceph/ceph synced 2024-12-23 20:03:56 +00:00
Commit Graph

33228 Commits

Author SHA1 Message Date
Samuel Just
ddf1e986dc osd: track the number of hit_set archive objects in a pg
Also, use this value in agent_choose_mode instead of the max
number.

Related: 
Signed-off-by: Samuel Just <sam.just@inktank.com>
2014-04-21 10:52:58 -07:00
Samuel Just
1fb90c94b3 ReplicatedPG::hit_set_persist: clean up degraded check
Signed-off-by: Samuel Just <sam.just@inktank.com>
2014-04-21 10:46:11 -07:00
Samuel Just
95d0278dcf ReplicatedPG::mark_all_unfound_lost: delete local copy if necessary
There might be a local copy for an EC pool in the DELETE case.  The replica
copies should be already handled by merge_log.

Fixes: 
Signed-off-by: Samuel Just <sam.just@inktank.com>
2014-04-20 18:59:21 -07:00
Chris Glass
61b6564b70 Simple mechanical cleanups
* Removed trailing and useless whitespaces
* Removed useless imports.

Signed-off-by: Christopher Glass <christopher.glass@canonical.com>
2014-04-19 14:58:33 +02:00
Sage Weil
7a61cdbfd5 buffer: adjust #include order
The pthread.h include is somehow clobbering things, although it is not
clear how.  :(

Signed-off-by: Sage Weil <sage@inktank.com>
2014-04-18 17:33:52 -07:00
Samuel Just
74f4d57347 Merge pull request from ceph/wip-8097
buffer: use Mutex instead of Spinlock for raw crcs

Reviewed-by: Samuel Just <sam.just@inktank.com>
2014-04-18 15:12:09 -07:00
Sage Weil
0234bcfca7 Merge pull request from ceph/wip-num_objects_omap
osd_types::object_stat_sum_t: fix add/sub for num_objects_omap

Reviewed-by: Sage Weil <sage@inktank.com>
2014-04-18 14:24:06 -07:00
Sage Weil
e087eae868 Merge pull request from ceph/wip-8153
Revert "ReplicatedPG::get_snapset_context: assert snap obj is not missin...

Reviewed-by: Sage Weil <sage@inktank.com>
2014-04-18 14:09:37 -07:00
Samuel Just
f9e9365fda Revert "ReplicatedPG::get_snapset_context: assert snap obj is not missing"
This breaks mark_lost_unfound_revert.

This reverts commit 0d2177a180.
2014-04-18 13:59:22 -07:00
Sage Weil
dec77c3407 Merge pull request from ceph/wip-7997
mon: fix get_version race (more)

Reviewed-by: Joao Eduardo Luis <joao.luis@inktank.com>
2014-04-18 13:54:30 -07:00
Sage Weil
4413670d78 osd: throttle snap trimmming with simple delay
This is not particularly smart, but it is *a* knob that lets you make
the snap trimmer slow down.  It's a flow and a simple delay, so it is
adjustable at runtime.  Default is 0 (no change in behavior).

Partial solution for .

Signed-off-by: Sage Weil <sage@inktank.com>
2014-04-18 13:50:11 -07:00
Greg Farnum
82edda23aa test: handle the create-pg delay when testing cache split syntax
Signed-off-by: Greg Farnum <greg@inktank.com>
Reviewed-by: Samuel Just <sam.just@inktank.com>
2014-04-18 13:41:00 -07:00
Sage Weil
d07ce84148 Merge pull request from ceph/wip-7784
mon: OSDMonitor: HEALTH_WARN on 'mon osd down out interval == 0'

Reviewed-by: Sage Weil <sage@inktank.com>
2014-04-18 11:47:59 -07:00
Joao Eduardo Luis
b2112d5087 mon: OSDMonitor: HEALTH_WARN on 'mon osd down out interval == 0'
A 'status' or 'health' request will return a HEALTH_WARN whenever the
monitor handling the request has the option set to zero.

Fixes: 7784

Signed-off-by: Joao Eduardo Luis <joao.luis@inktank.com>
2014-04-18 19:15:52 +01:00
Sage Weil
09985d25a8 mon: wait for PaxosService readable in handle_get_version
We were waiting for the election to finish, but we need to *also* wait for
paxos to recover.  Being a peon or leader is not sufficient and we may
return a map that is still old.

Fixes: 
Signed-off-by: Sage Weil <sage@inktank.com>
2014-04-18 11:12:23 -07:00
Sage Weil
7251983d8e Merge pull request from ceph/wip-8092
Wip 8092

Reviewed-by: Sage Weil <sage@inktank.com>
2014-04-17 21:21:59 -07:00
Sage Weil
375e4ee8e8 Merge pull request from ceph/wip-8108
osd: OSDMap: have osdmap json dump print valid boolean instead of string

Reviewed-by: Sage Weil <sage@inktank.com>
2014-04-17 21:19:33 -07:00
Sage Weil
8fb2388d82 osd_types: pg_t: add get_ancestor() method
Give us the ancestor for when the pool had a past value for pg_num.

Signed-off-by: Sage Weil <sage@inktank.com>
2014-04-17 21:17:40 -07:00
Gregory Farnum
2dd2b11f1f Merge pull request from ceph/wip-mds-op-prio
mds: dynamically adjust priority of committing dirfrags

Reviewed-by: Greg Farnum <greg@inktank.com>
2014-04-17 17:53:40 -07:00
Sage Weil
dea701125d Merge pull request from ceph/wip-8091
Wip 8091

Reviewed-by: Sage Weil <sage@inktank.com>
2014-04-17 14:51:18 -07:00
Samuel Just
7e697b1bc2 ReplicatedPG::recover_replicas: do not recover clones while snap obj is missing
Otherwise, we cannot safely read the snapset for the clone.

Fixes: 
Signed-off-by: Samuel Just <sam.just@inktank.com>
2014-04-17 14:46:34 -07:00
Samuel Just
3ad51c8ed3 osd_types::object_stat_sum_t: fix add/sub for num_objects_omap
Introduced in a130a4452e
Signed-off-by: Samuel Just <sam.just@inktank.com>
2014-04-17 13:31:09 -07:00
Samuel Just
79e7db7505 Merge pull request from ceph/wip-8048
osd/ReplicatedPG: check clones for degraded

Reviewed-by: Samuel Just <sam.just@inktank.com>
2014-04-17 13:18:21 -07:00
Sage Weil
ac014510ff Merge pull request from ceph/wip-8132
mon: set leader commands prior to first election

Reviewed-by: Joao Eduardo Luis <joao.luis@inktank.com>
2014-04-17 13:18:01 -07:00
Sage Weil
3d0e80acd9 osd/ReplicatedPG: check clones for degraded
We check whether the head is degraded, and we check whether a clone is
unreadable, but in the case where we have a cache op on a degraded object,
we don't check.  That leads to an assert when the repop hits the replica
and the object is in the peer's missing set.

Fix this by adding a check on the clone when write_ordered is true.  Note
that checking write_ordered is better than whether it is a cache op because
we want to preserve write ordering even for reads that are flagged by the
client.

Fixes: 
Signed-off-by: Sage Weil <sage@inktank.com>
2014-04-17 13:11:54 -07:00
Sage Weil
224a0f5749 Merge pull request from ceph/wip-8086
ReplicatedPG::agent_work: skip hitset objects before getting object cont...

Reviewed-by: Sage Weil <sage@inktank.com>
2014-04-17 12:49:58 -07:00
Yehuda Sadeh
26f4d5b061 Merge pull request from ceph/wip-8130
osdc/Objecter: fix osd target for newly-homeless op

Reviewed-by: Yehuda Sadeh <yehuda@inktank.com>
2014-04-17 10:50:40 -07:00
Sage Weil
93c0515fd9 osdc/Objecter: fix osd target for newly-homeless op
If we recalculate the mapping and find that there is no primary, we need
to set the 'osd' field to -1.  Otherwise, the caller will try to resend
to a dead session with bad results.

This was introduced in the refactor 860d72770c.

Fixes: 
Signed-off-by: Sage Weil <sage@inktank.com>
2014-04-17 10:48:26 -07:00
Sage Weil
fe71a12d78 Merge pull request from onlyjob/debian
spelling corrections

Reviewed-by: Sage Weil <sage@inktank.com>
2014-04-17 10:07:40 -07:00
Sage Weil
b0338ca361 Merge pull request from ceph/wip-7699
mds: Fix respawn (add path resolution)

Reviewed-by: Greg Farnum <greg@inktank.com>
Reviewed-by: Sage Weil <sage@inktank.com>
2014-04-17 10:05:22 -07:00
Sage Weil
3a794d5fe1 Merge pull request from ceph/wip-poolset-noblock
mon: Don't block on EAGAIN from `osd pool set`

Reviewed-by: Sage Weil <sage@inktank.com>
2014-04-17 10:03:26 -07:00
Sage Weil
881680ee93 mon: set leader commands prior to first election
If we have just started and receive a command, we currently will reply with
EINVAL because the leader commands are empty.  Note that this race is very
difficult to reach because the (old) peon needs to forward a command to
the mon while it still thinks it has quorum, and the message needs to get
sent after the leader mon has restarted and reset its connection but before
it has declared a new election.

To fix this, we should assume at startup time that our commands are
valid.  If it is an internal command that does not require quorum, that
is fine.  If it does require quorum, we will retry the command after the
election completes and we will revalidate the command then.

Fixes: 
Signed-off-by: Sage Weil <sage@inktank.com>
2014-04-17 09:33:44 -07:00
John Spray
40e8dbbb6b mon: EBUSY instead of EAGAIN when pgs creating
In 69321bf, EAGAIN changed behaviour to block indefinitely
rather than returning to user.  Change the return for
`osd pool set` operations that are blocked by creating PGs
to return EBUSY instead of EAGAIN, so that they are excepted
from this blocking behaviour.

Signed-off-by: John Spray <john.spray@inktank.com>
2014-04-17 15:28:22 +01:00
Gregory Farnum
2e375b6f4e Merge pull request from guangyy/wip-bench
Make rados/rest bench work for multiple write instances without metadata conflict.

Reviewed-by: Greg Farnum <greg@inktank.com>
2014-04-16 21:57:41 -07:00
Dmitry Smirnov
f22e2e9a02 spelling corrections 2014-04-17 12:43:30 +10:00
Samuel Just
75a5bd5d49 Merge pull request from ceph/wip-8043
mon/OSDMonitor: require force argument to split a cache pool

Reviewed-by: Samuel Just <sam.just@inktank.com>
2014-04-16 18:16:11 -07:00
Sage Weil
6d58e3c9fc Merge pull request from ceph/wip-8020
OSD: split pg stats during pg split

Reviewed-by: Sage Weil <sage@inktank.com>
2014-04-16 18:13:01 -07:00
Samuel Just
18caa1cd8f OSD: split pg stats during pg split
Fixes: 
Signed-off-by: Samuel Just <sam.just@inktank.com>
2014-04-16 18:10:02 -07:00
Samuel Just
5e4a5dc6ea osd_types::osd_stat_sum_t: fix floor for num_objects_omap
Introduced in a130a4452e
Signed-off-by: Samuel Just <sam.just@inktank.com>
2014-04-16 18:08:35 -07:00
David Zafman
a3d759ebdd Merge branch 'wip-8100'
Reviewed-by: Mark Nelson <mark.nelson@inktank.com>
2014-04-16 15:09:12 -07:00
David Zafman
a3d452acdf common/obj_bencher: Fix error return check from read that is negative on error
Fixed read return value in d99f1d9f68

Fixes: 

Signed-off-by: David Zafman <david.zafman@inktank.com>
2014-04-16 15:06:55 -07:00
Sage Weil
24da7d0c17 Merge pull request from ceph/wip-7786
civetweb: update subproject
2014-04-16 11:49:58 -07:00
David Zafman
4db1984c2b osd/ReplicatedPG: add missing whitespace in debug output
Signed-off-by: David Zafman <david.zafman@inktank.com>
2014-04-16 11:08:23 -07:00
Yan, Zheng
924064f83b mds: dynamically adjust priority of committing dirfrags
Adjust priority of committing dirfrags according to number of
expiring log segments. The more expiring log segments, the higher
priority. Because it mean MDS does not trim log segments quickly
enough.

Signed-off-by: Yan, Zheng <zheng.z.yan@intel.com>
2014-04-16 15:11:38 +08:00
Yan, Zheng
0640a08534 mds: fix cap revoke confirmation
when the _revokes list is emptied, it doesn't mean that client has
released the revoking caps. It's possible that client was flusing
dirty metadata.

Signed-off-by: Yan, Zheng <zheng.z.yan@intel.com>
2014-04-16 13:51:04 +08:00
Guang Yang
8c7a5ab861 Use string instead of char* when saving arguments for rest-bench 2014-04-16 01:28:16 +00:00
Samuel Just
0d2177a180 ReplicatedPG::get_snapset_context: assert snap obj is not missing
Signed-off-by: Samuel Just <sam.just@inktank.com>
2014-04-15 15:37:46 -07:00
Sage Weil
015df934af mon/OSDMonitor: require force argument to split a cache pool
There are several perils when splitting a cache pool:

 - split invalidstes pg stats, which disables the agent
 - a scrub must be manually triggered post-split to rebuild stats
 - the pool may fill the OSDs during that period.
 - or, the pool may end up beyond the 'full' mark and once scrub does
   complete and the agent activate we may block IO for a long time while
   we catch up with flush/evict

Make it a bit harder for users to shoot themselves in the foot.

Fixes: 
Signed-off-by: Sage Weil <sage@inktank.com>
2014-04-15 13:57:21 -07:00
Joao Eduardo Luis
c252345889 osd: OSDMap: have osdmap json dump print valid boolean instead of string
Fixes: 8108

Signed-off-by: Joao Eduardo Luis <joao.luis@inktank.com>
2014-04-15 18:32:17 +01:00
John Spray
aa6df59e99 mds: Fix respawn (add path resolution)
Previously assumed that ceph-mds executable was in
PWD - now use /proc/self/exe to find the
executable whereever it may be.  Leave in old version
as a fallback for non-linux environments.

Also add a 'respawn' command so that it's easy to test
respawn with `ceph mds tell <id> respawn`

Fixes: 
2014-04-15 12:28:09 +01:00