]> git.apps.os.sepia.ceph.com Git - ceph.git/log
ceph.git
11 years agomon: pool set <pool> crush_ruleset must not use rule_exists 1962/head
John Spray [Tue, 20 May 2014 15:50:18 +0000 (16:50 +0100)]
mon: pool set <pool> crush_ruleset must not use rule_exists

Implement CrushWrapper::ruleset_exists that iterates over the existing
rulesets to find the one matching the ruleset argument.

ceph osd pool set <pool> crush_ruleset must not use
CrushWrapper::rule_exists, which checks for a *rule* existing, whereas
the value being set is a *ruleset*. (cherry picked from commit
fb504baed98d57dca8ec141bcc3fd021f99d82b0)

A test via ceph osd pool set data crush_ruleset verifies the ruleset
argument is accepted.

http://tracker.ceph.com/issues/8599 fixes: #8599

Backport: firefly, emperor, dumpling
Signed-off-by: John Spray <john.spray@inktank.com>
Signed-off-by: Loic Dachary <loic@dachary.org>
11 years agoinit-ceph: continue after failure doing osd data mount
Sage Weil [Mon, 9 Jun 2014 03:18:49 +0000 (20:18 -0700)]
init-ceph: continue after failure doing osd data mount

If we are starting many daemons and hit an error, we normally note it and
move on.  Do the same when doing the pre-mount step.

Fixes: #8554
Signed-off-by: Sage Weil <sage@inktank.com>
(cherry picked from commit 6a7e20147cc39ed4689809ca7d674d3d408f2a17)

11 years agoFix for bug #6700
Steve Taylor [Tue, 10 Jun 2014 18:42:55 +0000 (12:42 -0600)]
Fix for bug #6700

When preparing OSD disks with colocated journals, the intialization process
fails when using dmcrypt. The kernel fails to re-read the partition table after
the storage partition is created because the journal partition is already in use
by dmcrypt. This fix unmaps the journal partition from dmcrypt and allows the
partition table to be read.

Signed-off-by: Stephen F Taylor <steveftaylor@gmail.com>
(cherry picked from commit 673394702b725ff3f26d13b54d909208daa56d89)

11 years agodoc: Added Disable requiretty commentary. 1923/head
John Wilkins [Thu, 5 Jun 2014 18:29:20 +0000 (11:29 -0700)]
doc: Added Disable requiretty commentary.

Signed-off-by: John Wilkins <john.wilkins@inktank.com>
11 years agoReplicatedPG::start_flush: fix clone deletion case
Samuel Just [Fri, 16 May 2014 23:56:33 +0000 (16:56 -0700)]
ReplicatedPG::start_flush: fix clone deletion case

dsnapc.snaps will be non-empty most of the time if there
have been snaps before prev_snapc.  What we really want to
know is whether there are any snaps between oi.snaps.back()
and prev_snapc.

Fixes: 8334
Backport: firefly
Signed-off-by: Samuel Just <sam.just@inktank.com>
(cherry picked from commit 29f615b7ac9e92f77cdef9927070727fee9d5e33)

11 years agoReplicatedPG::start_flush: send delete even if there are no snaps
Samuel Just [Mon, 12 May 2014 22:08:07 +0000 (15:08 -0700)]
ReplicatedPG::start_flush: send delete even if there are no snaps

Even if all snaps for the clone have been removed, we still have to
send the delete to ensure that when the object is recreated the
new snaps aren't included in the wrong clone.

Signed-off-by: Samuel Just <sam.just@inktank.com>
(cherry picked from commit 2ec2182745fa7c22526a7cf3dedb25bc314c9db4)

11 years agoHashIndex: in cleanup, interpret missing dir as completed merge
Samuel Just [Fri, 16 May 2014 03:53:27 +0000 (20:53 -0700)]
HashIndex: in cleanup, interpret missing dir as completed merge

If we stop between unlinking the empty subdir and removing the root
merge marker, we get ENOENT on the get_info.  That's actually fine.

Backport: firefly
Fixes: 8332
Signed-off-by: Samuel Just <sam.just@inktank.com>
(cherry picked from commit 5ff95dbdd2dbb533d344f37fea722ca4f140e670)

11 years agoadd backport of collections.Counter for python2.6
Alfredo Deza [Wed, 28 May 2014 15:48:12 +0000 (11:48 -0400)]
add backport of collections.Counter for python2.6

Using Raymond Hettinger's MIT backport

Signed-off-by: Alfredo Deza <alfredo.deza@inktank.com>
(cherry picked from commit 23b75b550507438c79b3aa75e06721e5f7b134a4)

11 years agorest-api: key missing for per "rx" and "rwx"
Ailing [Wed, 28 May 2014 19:37:48 +0000 (12:37 -0700)]
rest-api: key missing for per "rx" and "rwx"

commit 85a1cf31e6 and db266a3fb2 introduce new per "rx" and "rwx", but key missing for per "rx" and "rwx" in permmap

Signed-off-by: Ailing Zhang <ailzhang@cisco.com>
(cherry picked from commit 0b5a67410793ec28cac47e6e44cbbcf5684d77e7)

11 years agocephfs-java: build against older jni headers
Greg Farnum [Thu, 22 May 2014 04:41:23 +0000 (21:41 -0700)]
cephfs-java: build against older jni headers

Older versions of the JNI interface expected non-const parameters
to their memory move functions. It's unpleasant, but won't actually
change the memory in question, to do a cast_const in order to satisfy
those older headers. (And even if it *did* modify the memory, that
would be okay given our single user.)

Signed-off-by: Greg Farnum <greg@inktank.com>
(cherry picked from commit 4d4b77e5b6b923507ec4a0ad9d5c7018e4542a3c)

11 years agoOSDMonitor: set next commit in mon primary-affinity reply
Ilya Dryomov [Fri, 16 May 2014 15:03:13 +0000 (19:03 +0400)]
OSDMonitor: set next commit in mon primary-affinity reply

Commit 8c5c55c8b47e ("mon: set next commit in mon command replies")
fixed MMonCommand replies to include the right version, but the
primary-affinity handler was authored before that.  Fix it.

Backport: firefly
Signed-off-by: Ilya Dryomov <ilya.dryomov@inktank.com>
(cherry picked from commit a78b14ec1769ef37bef82bfda6faabb581b4cd7d)

11 years agoprioritise use of `javac` executable (gcj provides it through alternatives).
Dmitry Smirnov [Mon, 12 May 2014 04:08:44 +0000 (14:08 +1000)]
prioritise use of `javac` executable (gcj provides it through alternatives).

 On Debian this fixes FTBFS when gcj-jdk and openjdk-7-jdk are installed at
 the same time because build system will use default `javac` executable
 provided by current JDK through `update-alternatives` instead of blindly
 calling GCJ when it is present.

Signed-off-by: Dmitry Smirnov <onlyjob@member.fsf.org>
(cherry picked from commit 8b682d167e4535df582f1c77542e2b1ea0981228)

11 years agopass '-classpath' option (gcj/javah ignores CLASSPATH environment variable).
Dmitry Smirnov [Mon, 12 May 2014 04:02:53 +0000 (14:02 +1000)]
pass '-classpath' option (gcj/javah ignores CLASSPATH environment variable).

 This should not affect OpenJDK which understands '-classpath' as well.

 With gcj-jdk we still get FTBFS later:

 ~~~~
    java/native/libcephfs_jni.cc:2878:55: error: invalid conversion from 'const jbyte* {aka const signed char*}' to 'jbyte* {aka signed char*}' [-fpermissive]
                 reinterpret_cast<const jbyte*>(rawAddress));
                                                           ^
    In file included from java/native/libcephfs_jni.cc:27:0:
    /usr/lib/gcc/x86_64-linux-gnu/4.8/include/jni.h:1471:8: error:   initializing argument 4 of 'void _Jv_JNIEnv::SetByteArrayRegion(jbyteArray, jsize, jsize, jbyte*)' [-fpermissive]
       void SetByteArrayRegion (jbyteArray val0, jsize val1, jsize val2, jbyte * val3)
            ^
    make[5] *** [java/native/libcephfs_jni_la-libcephfs_jni.lo] Error 1
 ~~~~

Signed-off-by: Dmitry Smirnov <onlyjob@member.fsf.org>
(cherry picked from commit 89fe0353582bde7e2fbf32f1626d430a20002dd0)

11 years agolook for "jni.h" in gcj-jdk path, needed to find "jni.h" with gcj-jdk_4.9.0
Dmitry Smirnov [Mon, 12 May 2014 03:57:20 +0000 (13:57 +1000)]
look for "jni.h" in gcj-jdk path, needed to find "jni.h" with gcj-jdk_4.9.0

Signed-off-by: Dmitry Smirnov <onlyjob@member.fsf.org>
(cherry picked from commit 0f4120c0115e7977ae7c03458addcc2b2916db07)

11 years agoceph-disk: partprobe before settle when preparing dev
Sage Weil [Thu, 8 May 2014 15:52:51 +0000 (08:52 -0700)]
ceph-disk: partprobe before settle when preparing dev

Two users have reported this fixes a problem with using --dmcrypt.

Fixes: #6966
Tested-by: Eric Eastman <eric0e@aol.com>
Signed-off-by: Sage Weil <sage@inktank.com>
(cherry picked from commit 0f196265f049d432e399197a3af3f90d2e916275)

11 years agotest: fix some templates to match new output code
Greg Farnum [Tue, 13 May 2014 20:15:28 +0000 (13:15 -0700)]
test: fix some templates to match new output code

Signed-off-by: Greg Farnum <greg@inktank.com>
Reviewed-by: Samuel Just <sam.just@inktank.com>
(cherry picked from commit 00225d739cefa1415524a3de45fb9a5a2db53018)

11 years agoOSD: fix an osdmap_subscribe interface misuse
Greg Farnum [Thu, 15 May 2014 23:50:43 +0000 (16:50 -0700)]
OSD: fix an osdmap_subscribe interface misuse

When calling osdmap_subscribe, you have to pass an epoch newer than the
current map's. _maybe_boot() was not doing this correctly -- we would
fail a check for being *in* the monitor's existing map range, and then
pass along the map prior to the monitor's range. But if we were exactly
one behind, that value would be our current epoch, and the request would
get dropped. So instead, make sure we are not *in contact* with the monitor's
existing map range.

Signed-off-by: Greg Farnum <greg@inktank.com>
Reviewed-by: Samuel Just <sam.just@inktank.com>
(cherry picked from commit 290ac818696414758978b78517b137c226110bb4)

11 years agoosd: skip out of order op checks on tiered pools
Sage Weil [Mon, 19 May 2014 17:32:12 +0000 (10:32 -0700)]
osd: skip out of order op checks on tiered pools

When we send redirected ops, we do not assign a new tid, which means that
a given client's ops for a pool may not have strictly ordered tids.  Skip
this check if the pool is tiered to avoid false positives.

Fixes: #8380
Signed-off-by: Sage Weil <sage@inktank.com>
(cherry picked from commit cf2b172c843da0599164901956b66c306a59e570)

11 years agoReplicatedPG: block scrub on blocked object contexts
Samuel Just [Tue, 6 May 2014 18:50:14 +0000 (11:50 -0700)]
ReplicatedPG: block scrub on blocked object contexts

Fixes: #8011
Backport: firefly
Signed-off-by: Samuel Just <sam.just@inktank.com>
(cherry picked from commit 7411477153219d66625a74c5886530029c516036)

11 years agomsg: Fix inconsistent message sequence negotiation during connection reset
Guang Yang [Fri, 9 May 2014 09:21:23 +0000 (09:21 +0000)]
msg: Fix inconsistent message sequence negotiation during connection reset

Backport: firefly, emperor, dumpling

Signed-off-by: Guang Yang (yguang@yahoo-inc.com)
Reviewed-by: Greg Farnum <greg@inktank.com>
(cherry picked from commit bdee119076dd0eb65334840d141ccdf06091e3c9)

11 years agoObjecter::_op_submit: only replace the tid if it's 0
Samuel Just [Tue, 15 Apr 2014 19:55:47 +0000 (12:55 -0700)]
Objecter::_op_submit: only replace the tid if it's 0

Otherwise, redirected ops will suddenly have a different tid
and will become uncancelable.

Fixes: #7588
Signed-off-by: Samuel Just <sam.just@inktank.com>
(cherry picked from commit 76568aa0db4e16ac1af8fe6405edade1e61cbc81)

11 years agomon/OSDMonitor: force op resend when pool overlay changes
Sage Weil [Thu, 8 May 2014 17:42:42 +0000 (10:42 -0700)]
mon/OSDMonitor: force op resend when pool overlay changes

If a client is sending a sequence of ops (say, a, b, c, d) and partway
through that sequence it receives an OSDMap update that changes the
overlay, the ops will get send to different pools, and the replies will
come back completely out of order.

To fix this, force a resend of all outstanding ops any time the overlay
changes.

Fixes: #8305
Signed-off-by: Sage Weil <sage@inktank.com>
(cherry picked from commit 63d92ab0969357f78fdade749785136a509bc81b)

11 years agoosd: discard client ops sent before last_force_op_resend
Sage Weil [Thu, 8 May 2014 17:50:51 +0000 (10:50 -0700)]
osd: discard client ops sent before last_force_op_resend

If an op is sent before last_force_op_resend, and the client's feature is
present, drop the op because we know they will resend.

Signed-off-by: Sage Weil <sage@inktank.com>
(cherry picked from commit 45e79a17a932192995f8328ae9f6e8a2a6348d10)

11 years agoosdc/Objecter: resend ops in the last_force_op_resend epoch
Sage Weil [Thu, 8 May 2014 17:52:11 +0000 (10:52 -0700)]
osdc/Objecter: resend ops in the last_force_op_resend epoch

If we are a client, and process a map that sets last_force_op_resend to
the current epoch, force a resend of this op.

If the OSD expects us to do this, it will discard our previous op.  If the
OSD is old, it will process the old one, this will appear as a dup, and we
are no worse off than before.

Signed-off-by: Sage Weil <sage@inktank.com>
(cherry picked from commit dd700bdf7115223cb3e517b851f462d75dd76a2b)

11 years agoosd/osd_types: add last_force_op_resend to pg_pool_t
Sage Weil [Thu, 8 May 2014 17:40:10 +0000 (10:40 -0700)]
osd/osd_types: add last_force_op_resend to pg_pool_t

Signed-off-by: Sage Weil <sage@inktank.com>
(cherry picked from commit 3152faf79f498a723ae0fe44301ccb21b15a96ab)

11 years agoosd: handle race between osdmap and prepare_to_stop
Sage Weil [Fri, 9 May 2014 16:20:34 +0000 (09:20 -0700)]
osd: handle race between osdmap and prepare_to_stop

If we get a MOSDMarkMeDown message and set service.state == STOPPING, we
kick the prepare_to_stop() thread.  Normally, it will wake up and then
set osd.state == STOPPING, and when we process the map message next we
will not warn.  However, if dispatch() takes the lock instead and processes
the map, it will fail the preparing_to_stop check and issue a spurious
warning.

Fix by checking for either preparing_to_stop or stopping.

Fixes: #8319
Backport: firefly, emperor, dumpling
Signed-off-by: Sage Weil <sage@inktank.com>
(cherry picked from commit 6b858be0676f937a99dbd51321497f30c3a0097f)

11 years agoosd/ReplicatedPG: do not queue NULL dup_op
Sage Weil [Sat, 10 May 2014 17:29:11 +0000 (10:29 -0700)]
osd/ReplicatedPG: do not queue NULL dup_op

We call start_flush() with a NULL op in a couple different places.  Do not
put a NULL pointer on the dup_ops list or we will crash later.

Fixes: #8328
Signed-off-by: Sage Weil <sage@inktank.com>
(cherry picked from commit 0d67f9b0695765824bdc4a65fbed88edf8ea232e)

11 years ago0.80.1 v0.80.1
Jenkins [Mon, 12 May 2014 15:12:54 +0000 (15:12 +0000)]
0.80.1

11 years ago0.80.1
Jenkins [Mon, 12 May 2014 15:11:33 +0000 (15:11 +0000)]
0.80.1

11 years ago0.80.1
Jenkins [Mon, 12 May 2014 15:10:56 +0000 (15:10 +0000)]
0.80.1

11 years ago0.80.1
Jenkins [Mon, 12 May 2014 15:09:01 +0000 (15:09 +0000)]
0.80.1

11 years agoRevert "ReplicatedPG: block scrub on blocked object contexts"
Samuel Just [Fri, 2 May 2014 23:21:26 +0000 (16:21 -0700)]
Revert "ReplicatedPG: block scrub on blocked object contexts"

This reverts commit e66f2e36c06ca00c1147f922d3513f56b122a5c0.
Reviewed-by: Sage Weil <sage@inktank.com>
0f3235d46c8fd6c537bd4aa8a3faec6c00f311a8 is the firefly commit
corresponding to e66f2e36c06ca00c1147f922d3513f56b122a5c0.

(cherry picked from commit 84728058dbb91b8ed062240b3373b18078f0c9ca)

11 years agorgw: fix stripe_size calculation
Yehuda Sadeh [Tue, 6 May 2014 23:55:27 +0000 (16:55 -0700)]
rgw: fix stripe_size calculation

Fixes: #8299
Backport: firefly
The stripe size calculation was broken, specifically affected cases
where we had manifest that described multiple parts.

Signed-off-by: Yehuda Sadeh <yehuda@inktank.com>
(cherry picked from commit 9968b938b5d47fdf3a67db134bd2ea6bf3a28086)

11 years agorgw: cut short object read if a chunk returns error
Yehuda Sadeh [Tue, 6 May 2014 18:06:29 +0000 (11:06 -0700)]
rgw: cut short object read if a chunk returns error

Fixes: #8289
Backport: firefly, dumpling
When reading an object, if we hit an error when trying to read one of
the rados objects then we should just stop. Otherwise we're just going
to continue reading the rest of the object, and since it can't be sent
back to the client (as we have a hole in the middle), we end up
accumulating everything in memory.

Signed-off-by: Yehuda Sadeh <yehuda@inktank.com>
(cherry picked from commit 03b0d1cfb7bd30a77fedcf75eb06476b21b14e95)

11 years agorgw: send user manifest header field
Yehuda Sadeh [Mon, 21 Apr 2014 22:34:04 +0000 (15:34 -0700)]
rgw: send user manifest header field

Fixes: #8170
Backport: firefly
If user manifest header exists (swift) send it as part of the object
header data.

Signed-off-by: Yehuda Sadeh <yehuda@inktank.com>
(cherry picked from commit 5cc5686039a882ad345681133c9c5a4a2c2fd86b)

11 years agoclient: add asok command to kick sessions that were remote reset
Yan, Zheng [Fri, 11 Apr 2014 07:03:37 +0000 (15:03 +0800)]
client: add asok command to kick sessions that were remote reset

Fixes: #8021
Signed-off-by: Yan, Zheng <zheng.z.yan@intel.com>
(cherry picked from commit 09a1bc5a4601d356b9cc69be8541e6515d763861)

11 years agoosd: throttle snap trimmming with simple delay
Sage Weil [Fri, 18 Apr 2014 20:50:11 +0000 (13:50 -0700)]
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 #6278.

Signed-off-by: Sage Weil <sage@inktank.com>
(cherry picked from commit 4413670d784efc2392359f0f22bca7c9056188f4)

11 years agomon/MonClient: remove stray _finish_hunting() calls
Sage Weil [Fri, 2 May 2014 21:48:35 +0000 (14:48 -0700)]
mon/MonClient: remove stray _finish_hunting() calls

Callig _finish_hunting() clears out the bool hunting flag, which means we
don't retry by connection to another mon periodically.  Instead, we send
keepalives every 10s.  But, since we aren't yet in state HAVE_SESSION, we
don't check that the keepalives are getting responses.  This means that an
ill-timed connection reset (say, after we get a MonMap, but before we
finish authenticating) can drop the monc into a black hole that does not
retry.

Instead, we should *only* call _finish_hunting() when we complete the
authentication handshake.

Fixes: #8278
Backport: firefly, dumpling
Signed-off-by: Sage Weil <sage@inktank.com>
Reviewed-by: Joao Eduardo Luis <joao.luis@inktank.com>
(cherry picked from commit 77a6f0aefebebf057f02bfb95c088a30ed93c53f)

11 years agoosd/ReplicatedPG: fix trim of in-flight hit_sets
Sage Weil [Fri, 2 May 2014 23:41:26 +0000 (16:41 -0700)]
osd/ReplicatedPG: fix trim of in-flight hit_sets

We normally need to stat the hit_set to know how many bytes to adjust the
stats by.  If the hit_set was just written, we will get ENOENT.

Get the obc instead, which will either get the in-memory copy (because the
repop is still in flight) or load it off of disk.

Fixes: #8283
Backport: firefly
Signed-off-by: Sage Weil <sage@inktank.com>
(cherry picked from commit 72fdd557c35cb721d4b502c5a8f68c878f11a19c)

11 years agoosd/ReplicatedPG: fix whiteouts for other cache mode
Sage Weil [Tue, 6 May 2014 18:01:27 +0000 (11:01 -0700)]
osd/ReplicatedPG: fix whiteouts for other cache mode

We were special casing WRITEBACK mode for handling whiteouts; this needs to
also include the FORWARD and READONLY modes.  To avoid having to list
specific cache modes, though, just check != NONE.

Fixes: #8296
Backport: firefly
Signed-off-by: Sage Weil <sage@inktank.com>
(cherry picked from commit 3e387d62ed95898db8a7d7163c2bacc363b8f617)

11 years agoosd: Prevent divide by zero in agent_choose_mode()
Sage Weil [Thu, 1 May 2014 23:53:17 +0000 (16:53 -0700)]
osd: Prevent divide by zero in agent_choose_mode()

Fixes: #8175
Backport: firefly

Signed-off-by: David Zafman <david.zafman@inktank.com>
Signed-off-by: Sage Weil <sage@inktank.com>
(cherry picked from commit f47f867952e6b2a16a296c82bb9b585b21cde6c8)

11 years agoosd, common: If agent_work() finds no objs to work on delay 5 (default) secs
David Zafman [Tue, 22 Apr 2014 06:52:04 +0000 (23:52 -0700)]
osd, common: If agent_work() finds no objs to work on delay 5 (default) secs

Add config osd_agent_delay_time of 5 seconds
Honor delay by ignoring agent_choose_mode() calls
Add tier_delay to logger
Treat restart after delay like we were previously idle

Fixes: #8113
Backport: firefly

Signed-off-by: David Zafman <david.zafman@inktank.com>
(cherry picked from commit b7d31e5f5952c631dd4172bcb825e77a13fc60bc)

11 years agoosd/ReplicatedPG: agent_work() fix next if finished early due to start_max
David Zafman [Fri, 2 May 2014 01:54:30 +0000 (18:54 -0700)]
osd/ReplicatedPG: agent_work() fix next if finished early due to start_max

Backport: firefly

Signed-off-by: David Zafman <david.zafman@inktank.com>
(cherry picked from commit 9cf470cac8dd4d8f769e768f2de6b9eb67a3c3af)

11 years agoFix clone problem
Haomai Wang [Sat, 3 May 2014 04:53:06 +0000 (12:53 +0800)]
Fix clone problem

When clone happened, the origin header also will be updated in GenericObjectMap,
so the new header wraper(StripObjectHeader) should be updated too.

Fix #8282
Signed-off-by: Haomai Wang <haomaiwang@gmail.com>
(cherry picked from commit 3aee1e0ffe0583f74c02d9c9e86c7fb267f3515c)

11 years ago0.80 v0.80
Jenkins [Tue, 6 May 2014 14:03:28 +0000 (14:03 +0000)]
0.80

11 years agoMerge pull request #1763 from ceph/wip-blacklist
Sage Weil [Sat, 3 May 2014 22:11:58 +0000 (15:11 -0700)]
Merge pull request #1763 from ceph/wip-blacklist

Wip blacklist

Backport: firefly
Reviewed-by: Sage Weil <sage@inktank.com>
11 years agoosd: check blacklisted clients in ReplicatedPG::do_op() 1763/head
Yan, Zheng [Sat, 3 May 2014 21:17:15 +0000 (05:17 +0800)]
osd: check blacklisted clients in ReplicatedPG::do_op()

OSD checks if client is blacklisted only when receiving OSD request.
It's possible that OSD request's sender get blacklisted while OSD
request in in some waiting list.

Signed-off-by: Yan, Zheng <zheng.z.yan@intel.com>
11 years agoceph-object-corpus: v0.80-rc1-35-g4812150
Sage Weil [Sat, 3 May 2014 14:52:08 +0000 (07:52 -0700)]
ceph-object-corpus: v0.80-rc1-35-g4812150

Signed-off-by: Sage Weil <sage@inktank.com>
11 years agomon/PGMonitor: set tid on no-op PGStatsAck
Sage Weil [Fri, 2 May 2014 22:10:43 +0000 (15:10 -0700)]
mon/PGMonitor: set tid on no-op PGStatsAck

The OSD needs to know the tid.  Both generally, and specifically because
the flush_pg_stats may be blocking on it.

Fixes: #8280
Signed-off-by: Sage Weil <sage@inktank.com>
Reviewed-by: Joao Eduardo Luis <joao.luis@inktank.com>
(cherry picked from commit 5a6ae2a978dcaf96ef89de3aaa74fe951a64def6)

11 years agomon/OSDMonitor: share latest map with osd on dup boot message
Sage Weil [Fri, 2 May 2014 22:00:11 +0000 (15:00 -0700)]
mon/OSDMonitor: share latest map with osd on dup boot message

If we get a dup boot message, share the newer maps with the osd so that
they know they are living in the past.

Fixes: #8279
Signed-off-by: Sage Weil <sage@inktank.com>
Reviewed-by: Joao Eduardo Luis <joao.luis@inktank.com>
(cherry picked from commit 2e6b24868da0b203c2d70ac91071166d95d1d851)

11 years agoMerge pull request #1751 from ceph/wip-mds-shutdown
Sage Weil [Fri, 2 May 2014 00:08:36 +0000 (17:08 -0700)]
Merge pull request #1751 from ceph/wip-mds-shutdown

mds: remove mdsdir in the final step of shutdown MDS

Reviewed-by: Sage Weil <sage@inktank.com>
11 years agomds: remove mdsdir in the final step of shutdown MDS 1751/head
Yan, Zheng [Thu, 1 May 2014 21:22:35 +0000 (05:22 +0800)]
mds: remove mdsdir in the final step of shutdown MDS

Otherwise we may get bad subtree map if we restart the MDS before
the shutdown process finishes.

Signed-off-by: Yan, Zheng <zheng.z.yan@intel.com>
11 years agoReplicatedPG: block scrub on blocked object contexts
Samuel Just [Fri, 18 Apr 2014 00:26:17 +0000 (17:26 -0700)]
ReplicatedPG: block scrub on blocked object contexts

Fixes: #8011
Signed-off-by: Samuel Just <sam.just@inktank.com>
Reviewed-by: Sage Weil <sage@inktank.com>
(cherry picked from commit e66f2e36c06ca00c1147f922d3513f56b122a5c0)

11 years agorados.h,ReplicatedPG: add CEPH_OSD_FLAG_ENFORCE_SNAPC and use on flush
Samuel Just [Tue, 1 Apr 2014 23:27:20 +0000 (16:27 -0700)]
rados.h,ReplicatedPG: add CEPH_OSD_FLAG_ENFORCE_SNAPC and use on flush

We need to ensure that even with pool snaps, we use the snapc provided in order
to ensure that the clones are written back correctly.

Fixes: #7941
Signed-off-by: Samuel Just <sam.just@inktank.com>
(cherry picked from commit 499adb1db1cd225c91acce31b5f48fad5145043b)

11 years agoECBackend::continue_recovery_op: handle a source shard going down
Samuel Just [Thu, 24 Apr 2014 19:48:44 +0000 (12:48 -0700)]
ECBackend::continue_recovery_op: handle a source shard going down

get_min_avail_to_read_shards might return an error if there are
no longer enough sources to reconstruct the missing shards.
This is possible if osds went down while we were writing the
previous chunk -- we already notice in check_recovery_sources
if a source goes down during a read.

Fixes: #8161
Signed-off-by: Samuel Just <sam.just@inktank.com>
(cherry picked from commit 1885792c517670086332a8bab237c58558ee6dda)

11 years agoReplicatedPG: we can get EAGAIN on missing clone flush
Samuel Just [Fri, 25 Apr 2014 23:28:38 +0000 (16:28 -0700)]
ReplicatedPG: we can get EAGAIN on missing clone flush

Signed-off-by: Samuel Just <sam.just@inktank.com>
(cherry picked from commit 060105c313c5b4a777c55f17115eeb95ebb17117)

11 years agoReplicatedPG: do not preserve op context during flush
Samuel Just [Fri, 11 Apr 2014 01:15:30 +0000 (18:15 -0700)]
ReplicatedPG: do not preserve op context during flush

Any information stashed in the OpContext may be obsolete by the time we
actually mark the object clean.  Instead, let the start_flush caller
clean up its OpContext and in try_flush_mark_clean we'll create a new
one.  The primary reason to keep the OpContext would have been locking,
but we can set the obc as blocking without holding an OpContext, and
that would allow trimming to happen in the mean time (which is good
since trim_object does not respect rw locks since it doesn't change user
visible state).  In try_flush_mark_clean, we requeue the fop->op along
with (but ahead of) the fop->dup_ops.

Fixes: #8068
Signed-off-by: Samuel Just <sam.just@inktank.com>
(cherry picked from commit d83b8f58513e86c68aa6c25d4f909d9a3c4a103e)

11 years agorgw: fix url escaping
Yehuda Sadeh [Fri, 25 Apr 2014 21:11:27 +0000 (14:11 -0700)]
rgw: fix url escaping

Fixes: #8202
This fixes the radosgw side of issue #8202. Needed to cast value
to unsigned char, otherwise it'd get padded.

Backport: dumpling

Signed-off-by: Yehuda Sadeh <yehuda@inktank.com>
(cherry picked from commit bcf92c496aba0dfde432290fc2df5620a2767313)

11 years agodoc/release-notes: v0.67.8 notes
Sage Weil [Mon, 28 Apr 2014 19:40:04 +0000 (12:40 -0700)]
doc/release-notes: v0.67.8 notes

Signed-off-by: Sage Weil <sage@inktank.com>
11 years agoMerge pull request #1729 from ceph/wip-7966
Yan, Zheng [Sat, 26 Apr 2014 12:49:16 +0000 (20:49 +0800)]
Merge pull request #1729 from ceph/wip-7966

readlink result in resapwn

Reviewed-by: Yan, Zheng <zheng.z.yan@intel.com>
11 years agomds: terminate readlink result in resapwn 1729/head
Sage Weil [Sat, 26 Apr 2014 02:46:24 +0000 (19:46 -0700)]
mds: terminate readlink result in resapwn

readlink(2) does not null terminate the buffer; we need to do that.

Fixes: #7966
Signed-off-by: Sage Weil <sage@inktank.com>
11 years agoMerge pull request #1727 from ceph/wip-8193
wusui [Fri, 25 Apr 2014 23:20:20 +0000 (16:20 -0700)]
Merge pull request #1727 from ceph/wip-8193

ceph_test_rados_api_tier: increase HitSetTrim timeouts

11 years agoMerge pull request #1725 from FlorentCoppint/master
Sage Weil [Fri, 25 Apr 2014 22:58:47 +0000 (15:58 -0700)]
Merge pull request #1725 from FlorentCoppint/master

Skipping '_netdev' Debian fstab option

Reviewed-by: Sage Weil <sage@inktank.com>
11 years agoceph_test_rados_api_tier: increase HitSetTrim timeouts 1727/head
Sage Weil [Fri, 25 Apr 2014 22:49:06 +0000 (15:49 -0700)]
ceph_test_rados_api_tier: increase HitSetTrim timeouts

...so that they pass when they get unlucky with thrashing.

This will vastly decrease the probability of failure, but failure will
always be possible when a timeout is in place.

Fixes: #8193
Signed-off-by: Sage Weil <sage@inktank.com>
11 years agoSkipping '_netdev' Debian fstab option 1725/head
FlorentCoppint [Fri, 25 Apr 2014 07:20:02 +0000 (09:20 +0200)]
Skipping '_netdev' Debian fstab option

Signed-off-by: Florent Bautista <florent@coppint.com>
11 years agoMerge pull request #1717 from dachary/wip-auid
Loic Dachary [Thu, 24 Apr 2014 20:42:09 +0000 (22:42 +0200)]
Merge pull request #1717 from dachary/wip-auid

mon: add ceph osd pool set <pool> auid

Reviewed-by: Greg Farnum <greg@inktank.com>
11 years agoMerge pull request #1724 from ceph/wip-uselocalgithubforqemu-wusui
wusui [Thu, 24 Apr 2014 20:27:40 +0000 (13:27 -0700)]
Merge pull request #1724 from ceph/wip-uselocalgithubforqemu-wusui

Use new git mirror for qemu-iotests

11 years agoUse new git mirror for qemu-iotests 1724/head
Warren Usui [Thu, 24 Apr 2014 19:55:26 +0000 (12:55 -0700)]
Use new git mirror for qemu-iotests

Fixes: 8191
Signed-off-by: Warren Usui <warren.usui@inktank.com>
11 years agoMerge remote-tracking branch 'gh/firefly'
Sage Weil [Thu, 24 Apr 2014 01:00:59 +0000 (18:00 -0700)]
Merge remote-tracking branch 'gh/firefly'

11 years agoMerge pull request #1720 from jdurgin/wip-list-children-test
Sage Weil [Thu, 24 Apr 2014 00:23:12 +0000 (17:23 -0700)]
Merge pull request #1720 from jdurgin/wip-list-children-test

test_rbd.py: ignore children in cache pools

Reviewed-by: Sage Weil <sage@inktank.com>
11 years agoMerge pull request #1719 from ceph/wip-8168
Sage Weil [Wed, 23 Apr 2014 23:07:02 +0000 (16:07 -0700)]
Merge pull request #1719 from ceph/wip-8168

Wip 8168

Reviewed-by: Sage Weil <sage@inktank.com>
11 years agoReplicatedPG::do_osd_ops: consider head whiteout in list-snaps 1719/head
Samuel Just [Tue, 22 Apr 2014 23:03:48 +0000 (16:03 -0700)]
ReplicatedPG::do_osd_ops: consider head whiteout in list-snaps

Signed-off-by: Samuel Just <sam.just@inktank.com>
11 years agoReplicatedPG::do_op: don't return ENOENT for whiteout on snapdir read
Samuel Just [Tue, 22 Apr 2014 22:12:52 +0000 (15:12 -0700)]
ReplicatedPG::do_op: don't return ENOENT for whiteout on snapdir read

Signed-off-by: Samuel Just <sam.just@inktank.com>
11 years agoMerge pull request #1718 from ceph/wip-7882-wusui
Josh Durgin [Wed, 23 Apr 2014 20:47:02 +0000 (13:47 -0700)]
Merge pull request #1718 from ceph/wip-7882-wusui

Support latest qemu iotest code

Reviewed-by: Josh Durgin <josh.durgin@inktank.com>
11 years agotest_rbd.py: ignore children in cache pools 1720/head
Josh Durgin [Wed, 23 Apr 2014 20:27:48 +0000 (13:27 -0700)]
test_rbd.py: ignore children in cache pools

This is necessary until http://tracker.ceph.com/issues/8187 is fixed.

Signed-off-by: Josh Durgin <josh.durgin@inktank.com>
11 years agomon: add ceph osd pool set <pool> auid 1717/head
Loic Dachary [Wed, 23 Apr 2014 19:55:57 +0000 (21:55 +0200)]
mon: add ceph osd pool set <pool> auid

When a pool is created with ceph osd pool create, the auid is not
inferred from the session auid and is set to zero. Add the

  ceph osd pool set <pool> auid <int>

command to set it after it is created, and the matching get:

  ceph osd pool get <pool> auid

Signed-off-by: Loic Dachary <loic@dachary.org>
11 years agoSupport latest qemu iotest code 1718/head
Warren Usui [Wed, 23 Apr 2014 20:20:14 +0000 (13:20 -0700)]
Support latest qemu iotest code

Modified qemu-iotests workunit script to check for versions
that use the latest qemu (currently only Trusty).  Limit the
tests to those that are applicable to rbd.

Fixes: 7882
Signed-off-by: Warren Usui <warren.usui@inktank.com>
11 years agoMerge pull request #1714 from ceph/wip-fs-client
Sage Weil [Wed, 23 Apr 2014 15:51:26 +0000 (08:51 -0700)]
Merge pull request #1714 from ceph/wip-fs-client

two small fixes for client

Reviewed-by: Sage Weil <sage@inktank.com>
11 years agoclient: check cap ID when handling cap export message 1714/head
Yan, Zheng [Mon, 21 Apr 2014 08:26:33 +0000 (16:26 +0800)]
client: check cap ID when handling cap export message

handle following sequence of events:
- mds0 exports an inode to mds1. client receives the cap import
  message from mds1. caps from mds0 are removed while handling
  the cap import message.
- mds1 exports an inode to mds0. client receives the cap export
  message from mds1. handle_cap_export() adds placeholder caps
  for mds0
- client receives the first cap export message (for exporting
  inode from mds0 to mds1)

Signed-off-by: Yan, Zheng <zheng.z.yan@intel.com>
11 years agoclient: avoid releasing caps that are being used
Yan, Zheng [Tue, 22 Apr 2014 02:26:50 +0000 (10:26 +0800)]
client: avoid releasing caps that are being used

To avoid releasing caps that are being used, encode_inode_release()
should send implemented caps to MDS.

Signed-off-by: Yan, Zheng <zheng.z.yan@intel.com>
11 years agoMerge pull request #1713 from ceph/wip-7439
Samuel Just [Wed, 23 Apr 2014 00:36:20 +0000 (17:36 -0700)]
Merge pull request #1713 from ceph/wip-7439

Wip 7439

Reviewed-by: Sage Weil <sage@inktank.com>
11 years agoReplicatedPG: handle ec pools in mark_all_unfound_lost 1713/head
Samuel Just [Tue, 22 Apr 2014 21:56:08 +0000 (14:56 -0700)]
ReplicatedPG: handle ec pools in mark_all_unfound_lost

Signed-off-by: Samuel Just <sam.just@inktank.com>
11 years ago0.80-rc1 v0.80-rc1
Jenkins [Tue, 22 Apr 2014 21:21:45 +0000 (21:21 +0000)]
0.80-rc1

11 years agoReplicatedPG: enable mark_unfound_lost delete for ec pools
Samuel Just [Tue, 22 Apr 2014 19:45:28 +0000 (12:45 -0700)]
ReplicatedPG: enable mark_unfound_lost delete for ec pools

revert is tricky to implement at this time for ec pools, so
we'll instead just implement delete for ec pools.

Fixes: #7439
Signed-off-by: Samuel Just <sam.just@inktank.com>
11 years agoqa/workunits/rbd/copy.sh: skip some tests when tiering is enabled
Sage Weil [Tue, 22 Apr 2014 16:42:16 +0000 (09:42 -0700)]
qa/workunits/rbd/copy.sh: skip some tests when tiering is enabled

The rados ls bit doesn't work.

Signed-off-by: Sage Weil <sage@inktank.com>
Reviewed-by: Josh Durgin <josh.durgin@inktank.com>
11 years agoqa/workunits/rbd/copy.sh: fix test
Sage Weil [Tue, 22 Apr 2014 16:37:32 +0000 (09:37 -0700)]
qa/workunits/rbd/copy.sh: fix test

I broke this in commit 9d64ac66082bd108ec3c2a74e2e77475b5564eae.

Signed-off-by: Sage Weil <sage@inktank.com>
Signed-off-by: Josh Durgin <josh.durgin@inktank.com>
11 years agoMerge pull request #1691 from ceph/wip-8139
Sage Weil [Tue, 22 Apr 2014 19:40:02 +0000 (12:40 -0700)]
Merge pull request #1691 from ceph/wip-8139

osd_types: pg_t: allow is_split to handle checks for splits prior to the most recent

Reviewed-by: Samuel Just <sam.just@inktank.com>
11 years agoECBackend: use std::swap for boost::optional
Samuel Just [Tue, 22 Apr 2014 17:21:55 +0000 (10:21 -0700)]
ECBackend: use std::swap for boost::optional

Reviewed-by: Sage Weil <sage@inktank.com>
Signed-off-by: Samuel Just <sam.just@inktank.com>
11 years agoMerge pull request #1710 from ceph/wip-coverity
Yehuda Sadeh [Tue, 22 Apr 2014 16:02:33 +0000 (09:02 -0700)]
Merge pull request #1710 from ceph/wip-coverity

a couple coverity fixes

Reviewed-by: Yehuda Sadeh <yehuda@inktank.com>
11 years agoMerge pull request #1711 from ceph/wip-coverity-respawn
Sage Weil [Tue, 22 Apr 2014 15:37:21 +0000 (08:37 -0700)]
Merge pull request #1711 from ceph/wip-coverity-respawn

mds: make strncpy in ::respawn safer

Reviewed-by: Sage Weil <sage@inktank.com>
11 years agomds: make strncpy in ::respawn safer 1711/head
John Spray [Tue, 22 Apr 2014 15:31:27 +0000 (16:31 +0100)]
mds: make strncpy in ::respawn safer

Previous code assumed null terminated argv[0]
was not longer than PATH_MAX and the resulting
strncpy was not strictly safe.

Modify the bounds to ensure that copy will not
result in an unterminated string if argv[0]
is oversized.

Signed-off-by: John Spray <john.spray@inktank.com>
11 years agoosd/osd_types: RWState: initialize snaptrimmer_write_marker 1710/head
Sage Weil [Tue, 22 Apr 2014 15:29:58 +0000 (08:29 -0700)]
osd/osd_types: RWState: initialize snaptrimmer_write_marker

** CID 1204295:  Uninitialized scalar field  (UNINIT_CTOR)
/osd/osd_types.h: 2716 in ObjectContext::RWState::RWState()()

Signed-off-by: Sage Weil <sage@inktank.com>
11 years agoosdc/Objecter: drop unused field
Sage Weil [Tue, 22 Apr 2014 15:28:52 +0000 (08:28 -0700)]
osdc/Objecter: drop unused field

This as missed by 860d72770cdf092c027d50f4ee03bed76c975599.

** CID 1204296:  Uninitialized scalar field  (UNINIT_CTOR)
/osdc/Objecter.h: 1165 in Objecter::Op::Op(const object_t &, const
object_locator_t &, std::vector<OSDOp, std::allocator<OSDOp>> &, int, Context *,
Context *, unsigned long *)()
/osdc/Objecter.h: 1165 in Objecter::Op::Op(const object_t &, const
object_locator_t &, std::vector<OSDOp, std::allocator<OSDOp>> &, int, Context *,
Context *, unsigned long *)()

Signed-off-by: Sage Weil <sage@inktank.com>
11 years agodoc/release-notes: a bit of prose about firefly
Sage Weil [Tue, 22 Apr 2014 01:33:00 +0000 (18:33 -0700)]
doc/release-notes: a bit of prose about firefly

Signed-off-by: Sage Weil <sage@inktank.com>
11 years agoosd/osd_types: pg_interval_t: include primaries in operator<< 1691/head
Sage Weil [Sun, 20 Apr 2014 05:08:41 +0000 (22:08 -0700)]
osd/osd_types: pg_interval_t: include primaries in operator<<

Also make up vs acting explicit.

Signed-off-by: Sage Weil <sage@inktank.com>
11 years agoosd/osd_types: pg_interval_t: include up_primary in pg_interval_t
Sage Weil [Sun, 20 Apr 2014 05:06:48 +0000 (22:06 -0700)]
osd/osd_types: pg_interval_t: include up_primary in pg_interval_t

Nothing uses this, but it triggers a new interval, which makes it confusing
when it is not recording in the interval itself.  Let's add it now.

Signed-off-by: Sage Weil <sage@inktank.com>
11 years agoosd/osd_types: pg_interval_t: dump primary
Sage Weil [Sun, 20 Apr 2014 05:05:27 +0000 (22:05 -0700)]
osd/osd_types: pg_interval_t: dump primary

Signed-off-by: Sage Weil <sage@inktank.com>
11 years agoosd: change in up set primary constitutes a peering interval change
Sage Weil [Sun, 20 Apr 2014 05:04:33 +0000 (22:04 -0700)]
osd: change in up set primary constitutes a peering interval change

In several places, a change in the up_primary triggers a new peering
interval, but the palces that actually generate the new past intervals,
including check_new_interval(), did not enforce that.  This becomes
somewhat obvious when you see that those callers are ignoring the
up_primary output argument for pg_to_up_acting_osds().

Fix this by adding arguments to check_new_interval and fixing the callers
to pass them in properly.  Add a unit test case to verify this.

Note that the past interval struct itself does not record who the
up_primary was; possibly it should.

Fixes: #8139
Signed-off-by: Sage Weil <sage@inktank.com>
11 years agoosd: use parent pgid (as appropriate) in generate_past_intervals()
Sage Weil [Fri, 18 Apr 2014 22:48:33 +0000 (15:48 -0700)]
osd: use parent pgid (as appropriate) in generate_past_intervals()

Feed in the ancestor pg_t (if any) when we are looking at intervals for
previous maps that may have preceded a recent split.

Fixes: #8139
Signed-off-by: Sage Weil <sage@inktank.com>
11 years agoMerge pull request #1651 from enovance/wip-brag
Sage Weil [Tue, 22 Apr 2014 03:49:43 +0000 (20:49 -0700)]
Merge pull request #1651 from enovance/wip-brag

Few bug fixes in ceph-brag

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