1 commit ba340a97c3dafc9155023da8d515eecc675c619a (tag: refs/tags/v0.67.9, refs/remotes/gh/dumpling)
2 Author: Jenkins <jenkins@inktank.com>
3 Date: Wed May 21 16:57:02 2014 +0000
7 commit b638d19d126646d2a8f6da11067c5f392a62525e
8 Author: Guang Yang <yguang@yahoo-inc.com>
9 Date: Fri May 9 09:21:23 2014 +0000
11 msg: Fix inconsistent message sequence negotiation during connection reset
13 Backport: firefly, emperor, dumpling
15 Signed-off-by: Guang Yang (yguang@yahoo-inc.com)
16 Reviewed-by: Greg Farnum <greg@inktank.com>
17 (cherry picked from commit bdee119076dd0eb65334840d141ccdf06091e3c9)
19 commit 55b1112819ed2e6fd509c09c2c1a2b434d22aac6
20 Author: Sage Weil <sage@inktank.com>
21 Date: Tue May 20 10:46:34 2014 -0700
23 OSD::handle_pg_query: on dne pg, send lb=hobject_t() if deleting
25 We will set lb=hobject_t() if we resurrect the pg. In that case,
26 we need to have sent that to the primary before hand. If we
27 finish the removal before the pg is recreated, we'll just end
28 up backfilling it, which is ok since the pg doesn't exist anyway.
31 Signed-off-by: Samuel Just <sam.just@inktank.com>
32 (cherry picked from commit 04de781765dd5ac0e28dd1a43cfe85020c0854f8)
38 commit 3b8ab41e1ec86f2ab5c6b4bee3fb4030077e2c21
39 Author: Sage Weil <sage@inktank.com>
40 Date: Fri May 2 14:48:35 2014 -0700
42 mon/MonClient: remove stray _finish_hunting() calls
44 Callig _finish_hunting() clears out the bool hunting flag, which means we
45 don't retry by connection to another mon periodically. Instead, we send
46 keepalives every 10s. But, since we aren't yet in state HAVE_SESSION, we
47 don't check that the keepalives are getting responses. This means that an
48 ill-timed connection reset (say, after we get a MonMap, but before we
49 finish authenticating) can drop the monc into a black hole that does not
52 Instead, we should *only* call _finish_hunting() when we complete the
53 authentication handshake.
56 Backport: firefly, dumpling
57 Signed-off-by: Sage Weil <sage@inktank.com>
58 Reviewed-by: Joao Eduardo Luis <joao.luis@inktank.com>
59 (cherry picked from commit 77a6f0aefebebf057f02bfb95c088a30ed93c53f)
61 commit a915350491092abcd2584d0bbb5f5d8610ed89fc
62 Merge: bd5d6f1 d4e67ff
63 Author: Sage Weil <sage@inktank.com>
64 Date: Tue May 20 10:19:00 2014 -0700
66 Merge pull request #1826 from ceph/wip-8162-dumpling
70 Reviewed-by: Sage Weil <sage@inktank.com>
72 commit bd5d6f116416d1b410d57ce00cb3e2abf6de102b
73 Author: Greg Farnum <greg@inktank.com>
74 Date: Thu May 15 16:50:43 2014 -0700
76 OSD: fix an osdmap_subscribe interface misuse
78 When calling osdmap_subscribe, you have to pass an epoch newer than the
79 current map's. _maybe_boot() was not doing this correctly -- we would
80 fail a check for being *in* the monitor's existing map range, and then
81 pass along the map prior to the monitor's range. But if we were exactly
82 one behind, that value would be our current epoch, and the request would
83 get dropped. So instead, make sure we are not *in contact* with the monitor's
86 Signed-off-by: Greg Farnum <greg@inktank.com>
87 Reviewed-by: Samuel Just <sam.just@inktank.com>
88 (cherry picked from commit 290ac818696414758978b78517b137c226110bb4)
90 commit c5ae8ffc281ac538dd0966a552a6cf729728e925
91 Merge: fced056 8d991fa
92 Author: Sage Weil <sage@inktank.com>
93 Date: Mon May 19 13:45:50 2014 -0700
95 Merge pull request #1827 from ceph/wip-6565-dumpling
99 Reviewed-by: Sage Weil <sage@inktank.com>
101 commit 8d991fa2c3fdbfd9178d93e4ca33d47504a37195
102 Author: Samuel Just <sam.just@inktank.com>
103 Date: Wed Oct 16 10:07:37 2013 -0700
105 OSD: check for splitting when processing recover/backfill reservations
108 Signed-off-by: Samuel Just <sam.just@inktank.com>
109 Reviewed-by: Sage Weil <sage@inktank.com>
110 (cherry picked from commit 15ec5332ba4154930a0447e2bcf1acec02691e97)
112 commit d4e67ff3037a3cc7ae2ecc9e1d8d086c45ae515a
113 Author: Samuel Just <sam.just@inktank.com>
114 Date: Thu May 8 13:25:32 2014 -0700
116 ReplicatedPG::recover_backfill: do not update last_backfill prematurely
118 Previously, we would update last_backfill on the backfill peer to
120 backfills_in_flight.empty() ? backfill_pos :
121 backfills_in_flight.begin()->first
123 which is actually the next backfill to complete. We want to update
124 last_backfill to the largest completed backfill instead.
126 We use the pending_backfill_updates mapping to identify the most
127 recently completed backfill. Due to the previous patch, deletes
128 will also be included in that mapping.
130 Related sha1s from master:
131 4139e75d63b0503dbb7fea8036044eda5e8b7cf1
132 7a06a71e0f2023f66d003dfb0168f4fe51eaa058
134 We don't really want to backport those due to the changes in:
135 9ec35d5ccf6a86c380865c7fc96017a1f502560a
137 This patch does essentially the same thing, but using backfill_pos.
140 Signed-off-by: Samuel Just <sam.just@inktank.com>
142 commit d620b13c51fc69b509824177af32de55c89c2259
143 Author: Samuel Just <sam.just@inktank.com>
144 Date: Mon Oct 28 16:03:25 2013 -0700
146 ReplicatedPG: add empty stat when we remove an object in recover_backfill
148 Subsequent updates to that object need to have their stats added
149 to the backfill info stats atomically with the last_backfill
152 Signed-off-by: Samuel Just <sam.just@inktank.com>
153 (cherry picked from commit ecddd12b01be120fba87f5ac60539f98f2c69a28)
155 commit fced0562f6b80035cd522cc81bdc844ccc92d9bc
156 Author: Yehuda Sadeh <yehuda@inktank.com>
157 Date: Wed Nov 27 13:34:00 2013 -0800
159 rgw: don't error out on empty owner when setting acls
162 Backport: dumpling, emperor
163 s3cmd specifies empty owner field when trying to set acls on object
164 / bucket. We errored out as it didn't match the current owner name, but
165 with this change we ignore it.
167 Signed-off-by: Yehuda Sadeh <yehuda@inktank.com>
168 (cherry picked from commit 14cf4caff58cc2c535101d48c53afd54d8632104)
170 commit 0e685c68aa6a500aa7fa433cd9b8246f70c5383e
171 Author: Yehuda Sadeh <yehuda@inktank.com>
172 Date: Mon Apr 21 15:34:04 2014 -0700
174 rgw: send user manifest header field
178 If user manifest header exists (swift) send it as part of the object
181 Signed-off-by: Yehuda Sadeh <yehuda@inktank.com>
182 (cherry picked from commit 5cc5686039a882ad345681133c9c5a4a2c2fd86b)
184 commit 20383e359cdfbf146838742fa050ec4b6073271d
185 Author: Yan, Zheng <zheng.z.yan@intel.com>
186 Date: Fri Apr 11 15:03:37 2014 +0800
188 client: add asok command to kick sessions that were remote reset
191 Signed-off-by: Yan, Zheng <zheng.z.yan@intel.com>
192 (cherry picked from commit 09a1bc5a4601d356b9cc69be8541e6515d763861)
194 commit 4e5e41deeaf91c885773d90e6f94da60f6d4efd3
195 Author: Sage Weil <sage@inktank.com>
196 Date: Fri Apr 18 13:50:11 2014 -0700
198 osd: throttle snap trimmming with simple delay
200 This is not particularly smart, but it is *a* knob that lets you make
201 the snap trimmer slow down. It's a flow and a simple delay, so it is
202 adjustable at runtime. Default is 0 (no change in behavior).
204 Partial solution for #6278.
206 Signed-off-by: Sage Weil <sage@inktank.com>
207 (cherry picked from commit 4413670d784efc2392359f0f22bca7c9056188f4)
209 commit 14650b282ecd344025714a4e743b255ae01b3ce0
210 Author: Sage Weil <sage@inktank.com>
211 Date: Tue Apr 1 16:01:28 2014 -0700
213 PG: only complete replicas should count toward min_size
215 Backport: emperor,dumpling,cuttlefish
217 Signed-off-by: Samuel Just <sam.just@inktank.com>
218 Signed-off-by: Sage Weil <sage@inktank.com>
219 (cherry picked from commit 0d5d3d1a30685e7c47173b974caa12076c43a9c4)
221 commit 3f1d7f5e0a67ad646de465335fb7ee00eb07e220
222 Author: Sage Weil <sage@inktank.com>
223 Date: Fri May 2 15:10:43 2014 -0700
225 mon/PGMonitor: set tid on no-op PGStatsAck
227 The OSD needs to know the tid. Both generally, and specifically because
228 the flush_pg_stats may be blocking on it.
231 Signed-off-by: Sage Weil <sage@inktank.com>
232 Reviewed-by: Joao Eduardo Luis <joao.luis@inktank.com>
233 (cherry picked from commit 5a6ae2a978dcaf96ef89de3aaa74fe951a64def6)