OpenStack Object Storage (swift)

Delete operation in swift-bench failed, when proxy affinity is enable

Bug #1198926 reported by Ksenia Svechnikova on 2013-07-08

This bug affects 1 person

Affects		Status	Importance	Assigned to	Milestone
	OpenStack Object Storage (swift)	Fix Released	Undecided	Ksenia Svechnikova	OpenStack Object Storage (swift) 1.10.0 "havana"

Bug Description

ProblemType: Bug
DistroRelease: Ubuntu 12.04 LTS
Package: swift 1.9.0

On SAIO after enable proxy affinity:

[app:proxy-server]
use = egg:swift#proxy
allow_account_management = true
account_autocreate = true
sorting_method = affinity
read_affinity = r1z1=100
write_affinity = r1

With 2 regions in the ring:

Devices: id region zone ip address port replication ip replication port name weight partitions balance meta
             0 1 1 127.0.0.1 6010 127.0.0.1 6010 sdb1 1.00 131072 0.00
             1 1 2 127.0.0.1 6020 127.0.0.1 6020 sdb2 1.00 131072 0.00
             2 2 3 127.0.0.1 6030 127.0.0.1 6030 sdb3 1.00 131072 0.00
             3 2 4 127.0.0.1 6040 127.0.0.1 6040 sdb4 1.00 131072 0.00

Swift-bench passed with failures, because the deletion work after replication has moved things to their proper homes.

swift@regions:/home/swift/bin$ swift-bench -A http://127.0.0.1:8080/auth/v1.0 -U test:tester -K testing -V 1.0
swift-bench 2013-07-08 13:38:58,262 INFO Auth version: 1.0
swift-bench 2013-07-08 13:38:58,969 INFO Auth version: 1.0
swift-bench 2013-07-08 13:39:01,591 INFO 1 PUTS [0 failures], 0.4/s
swift-bench 2013-07-08 13:39:16,727 INFO 60 PUTS [0 failures], 3.4/s
swift-bench 2013-07-08 13:39:31,960 INFO 112 PUTS [0 failures], 3.4/s
swift-bench 2013-07-08 13:39:47,072 INFO 170 PUTS [0 failures], 3.5/s
swift-bench 2013-07-08 13:40:02,137 INFO 224 PUTS [0 failures], 3.5/s
swift-bench 2013-07-08 13:40:17,230 INFO 274 PUTS [0 failures], 3.5/s
swift-bench 2013-07-08 13:40:32,339 INFO 323 PUTS [0 failures], 3.5/s
swift-bench 2013-07-08 13:40:47,490 INFO 381 PUTS [0 failures], 3.5/s
swift-bench 2013-07-08 13:41:02,624 INFO 439 PUTS [0 failures], 3.6/s
swift-bench 2013-07-08 13:41:17,880 INFO 486 PUTS [0 failures], 3.5/s
swift-bench 2013-07-08 13:41:33,129 INFO 533 PUTS [0 failures], 3.5/s
swift-bench 2013-07-08 13:41:48,215 INFO 579 PUTS [0 failures], 3.4/s
swift-bench 2013-07-08 13:42:03,287 INFO 628 PUTS [0 failures], 3.4/s
swift-bench 2013-07-08 13:42:18,630 INFO 681 PUTS [0 failures], 3.4/s
swift-bench 2013-07-08 13:42:33,833 INFO 740 PUTS [0 failures], 3.4/s
swift-bench 2013-07-08 13:42:49,667 INFO 789 PUTS [0 failures], 3.4/s
swift-bench 2013-07-08 13:43:04,873 INFO 831 PUTS [0 failures], 3.4/s
swift-bench 2013-07-08 13:43:20,091 INFO 879 PUTS [0 failures], 3.4/s
swift-bench 2013-07-08 13:43:35,210 INFO 920 PUTS [0 failures], 3.3/s
swift-bench 2013-07-08 13:43:50,405 INFO 967 PUTS [0 failures], 3.3/s
swift-bench 2013-07-08 13:43:58,731 INFO 1000 PUTS **FINAL** [0 failures], 3.3/s
swift-bench 2013-07-08 13:43:58,731 INFO Auth version: 1.0
swift-bench 2013-07-08 13:44:00,735 INFO 340 GETS [0 failures], 170.0/s
swift-bench 2013-07-08 13:44:15,737 INFO 3199 GETS [0 failures], 188.1/s
swift-bench 2013-07-08 13:44:30,752 INFO 5228 GETS [0 failures], 163.3/s
swift-bench 2013-07-08 13:44:45,753 INFO 7745 GETS [0 failures], 164.7/s
swift-bench 2013-07-08 13:45:00,753 INFO 9805 GETS [0 failures], 158.1/s
swift-bench 2013-07-08 13:45:03,620 INFO 10000 GETS **FINAL** [0 failures], 154.1/s
swift-bench 2013-07-08 13:45:03,620 INFO Auth version: 1.0
swift-bench 2013-07-08 13:45:05,672 INFO 110 DEL [12 failures], 53.8/s
swift-bench 2013-07-08 13:45:20,689 INFO 921 DEL [21 failures], 54.0/s
swift-bench 2013-07-08 13:45:21,976 INFO 1000 DEL **FINAL** [21 failures], 54.5/s
swift-bench 2013-07-08 13:45:21,976 INFO Auth version: 1.0

proxy.error:

Jun 26 13:58:36 regions proxy-server Handoff requested (2) (txn: tx30cf1da0b47842e19b162-0051cabb4c) (client_ip: 127.0.0.1)
Jun 26 13:58:36 regions proxy-server Handoff requested (2) (txn: txcdd5c483faf84df58a2be-0051cabb4c) (client_ip: 127.0.0.1)
Jun 26 13:58:36 regions proxy-server Handoff requested (2) (txn: tx0bb91eeba5d0447fb6f74-0051cabb4c) (client_ip: 127.0.0.1)
Jun 26 13:58:36 regions proxy-server Handoff requested (2) (txn: txf860bc3943a141e8a19dc-0051cabb4c) (client_ip: 127.0.0.1)
Jul 8 13:36:22 regions proxy-server SIGTERM received
Jul 8 13:36:22 regions proxy-server Exited
Jul 8 13:37:15 regions proxy-server Started child 29295
Jul 8 13:38:46 regions proxy-server Handoff requested (1) (txn: txdf236cf5c0cb4b1fb6f5f-0051da88a6) (client_ip: 127.0.0.1)
Jul 8 13:38:46 regions proxy-server Handoff requested (2) (txn: txdf236cf5c0cb4b1fb6f5f-0051da88a6) (client_ip: 127.0.0.1)
Jul 8 13:39:03 regions proxy-server ERROR with Container server 127.0.0.1:6021/sdb2 re: Trying to HEAD /v1/AUTH_test/fb139fa365f5417fa84751456c267efd_2: ConnectionTimeout (0.5s) (txn: tx63151415b39a4c4c9bdcc-0051da88b6)
Jul 8 13:39:03 regions proxy-server ERROR with Container server 127.0.0.1:6041/sdb4 re: Trying to HEAD /v1/AUTH_test/fb139fa365f5417fa84751456c267efd_19: ConnectionTimeout (0.5s) (txn: txd20c8cf804a3494aabe25-0051da88b6)
Jul 8 13:45:03 regions proxy-server Object DELETE returning 503 for (204, 404) (txn: txe82e5239b3654dc692a8b-0051da8a1f) (client_ip: 127.0.0.1)
Jul 8 13:45:04 regions proxy-server Object DELETE returning 503 for (404, 204) (txn: txf8eaa22100f642efa6a73-0051da8a20) (client_ip: 127.0.0.1)
Jul 8 13:45:04 regions proxy-server Object DELETE returning 503 for (204, 404) (txn: tx7c4fbfe09709402d8273a-0051da8a20) (client_ip: 127.0.0.1)
Jul 8 13:45:04 regions proxy-server Object DELETE returning 503 for (204, 404) (txn: txabcadddcbcb248e3988f5-0051da8a20) (client_ip: 127.0.0.1)
Jul 8 13:45:04 regions proxy-server Object DELETE returning 503 for (204, 404) (txn: txb545dcc4e5f54f388c8af-0051da8a20) (client_ip: 127.0.0.1)
Jul 8 13:45:04 regions proxy-server Object DELETE returning 503 for (204, 404) (txn: txeb3872e6a1e84ebf977ad-0051da8a20) (client_ip: 127.0.0.1)
Jul 8 13:45:08 regions proxy-server Object DELETE returning 503 for (204, 404) (txn: txc81de90b4cba431590c95-0051da8a23) (client_ip: 127.0.0.1)
Jul 8 13:45:08 regions proxy-server Object DELETE returning 503 for (204, 404) (txn: tx73dce0eb38484b18a9643-0051da8a24) (client_ip: 127.0.0.1)
Jul 8 13:45:08 regions proxy-server Object DELETE returning 503 for (204, 404) (txn: tx197af5afff2543909e3fc-0051da8a24) (client_ip: 127.0.0.1)
Jul 8 13:45:08 regions proxy-server Object DELETE returning 503 for (404, 204) (txn: tx5d5839e0e5364b9d8f656-0051da8a24) (client_ip: 127.0.0.1)

Revision history for this message

clayg (clay-gerrard) wrote on 2013-07-12:

It seems like maybe this is a two replica ring, can you confirm?

Seems like there's a link missing from the builder output.

-Clay

Revision history for this message

clayg (clay-gerrard) wrote on 2013-07-12:

*line missing

Revision history for this message

Ksenia Svechnikova (kdemina) wrote on 2013-07-15:

Yes, this is a two replica ring:

/etc/swift/object.builder, build version 4
262144 partitions, 2.000000 replicas, 2 regions, 4 zones, 4 devices, 0.00 balance
The minimum number of hours before a partition can be reassigned is 1
Devices: id region zone ip address port replication ip replication port name weight partitions balance meta
             0 1 1 127.0.0.1 6010 127.0.0.1 6010 sdb1 1.00 131072 0.00
             1 1 2 127.0.0.1 6020 127.0.0.1 6020 sdb2 1.00 131072 0.00
             2 2 3 127.0.0.1 6030 127.0.0.1 6030 sdb3 1.00 131072 0.00
             3 2 4 127.0.0.1 6040 127.0.0.1 6040 sdb4 1.00 131072 0.00

Revision history for this message

clayg (clay-gerrard) wrote on 2013-07-15:

Hi Ksenia,

Thank you for confirming. In a three replica geo-distributed cluster you'd get a 404 or 200 - I think both are valid and correct depending on the system state and given eventual consistency.

with one primary - 200 404 404 => 404
with two primary - 200 200 404 => 200

This is a generalize-able pre-existing two replica *bug*:

So we wrote one copy at a primary location, and one at a handoff.

We sent the delete to the two primary nodes; one said 200, one said 404, the proxy said 503 because we refuse to define a quorum with two nodes.

Personally I think we could do better, but we'll have to start accepting and triaging 2 replica bugs like this. As a start I'd want to move quorum from > 50% to >= 50%. I think it makes more sense in both the 2 and 4 replica cases and makes no difference on 3.

This is not the first time I've said this tho, so I'm not sure the fate of this bug. IMHO, we should at a minimum change the description.

Revision history for this message

Kun Huang (academicgareth) wrote on 2013-07-15:

@clayg

Do you have some advice of replicas number for building a ring in two IDC?

4/6 or 5/7, which is better? I think if 6, the ">50%" meaning finding at least 4 nodes in same results, which is same as 7 replicas. 7 makes "50%" more conceivable, but 5 saves more storage.

Revision history for this message

Ksenia Svechnikova (kdemina) wrote on 2013-07-17:

Download full text (5.8 KiB)

Hi, Clay,

Thank you for answer.
To test performance with a number of regions, I reconfigured SAIO. Now there are 6 devices, 3 regions, 3 replicas.
There are delete errors and with 3 replica, when affinity is enable:

swift@regions:/home/swift/scenario$ swift-ring-builder /etc/swift/object.builder
/etc/swift/object.builder, build version 6
262144 partitions, 3.000000 replicas, 3 regions, 6 zones, 6 devices, 0.00 balance
The minimum number of hours before a partition can be reassigned is 1
Devices: id region zone ip address port replication ip replication port name weight partitions balance meta
             0 1 1 127.0.0.1 6010 127.0.0.1 6010 sdb1 1.00 131072 0.00
             1 1 2 127.0.0.1 6020 127.0.0.1 6020 sdb2 1.00 131072 0.00
             2 2 3 127.0.0.1 6030 127.0.0.1 6030 sdb3 1.00 131072 0.00
             3 2 4 127.0.0.1 6040 127.0.0.1 6040 sdb4 1.00 131072 0.00
             4 3 5 127.0.0.1 6050 127.0.0.1 6050 sdb5 1.00 131072 0.00
             5 3 6 127.0.0.1 6060 127.0.0.1 6060 sdb6 1.00 131072 0.00

swift-bench 2013-07-17 17:34:57,226 INFO Auth version: 1.0
swift-bench 2013-07-17 17:34:59,249 INFO 44 DEL [0 failures], 21.9/s
swift-bench 2013-07-17 17:35:17,039 INFO 123 DEL [0 failures], 6.2/s
swift-bench 2013-07-17 17:35:32,220 INFO 254 DEL [0 failures], 7.3/s
swift-bench 2013-07-17 17:35:51,434 INFO 287 DEL [1 failures], 5.3/s
swift-bench 2013-07-17 17:36:11,622 INFO 296 DEL [10 failures], 4.0/s
swift-bench 2013-07-17 17:36:32,089 INFO 306 DEL [19 failures], 3.2/s
swift-bench 2013-07-17 17:36:56,990 INFO 316 DEL [29 failures], 2.6/s
swift-bench 2013-07-17 17:37:12,107 INFO 596 DEL [32 failures], 4.4/s
swift-bench 2013-07-17 17:37:24,045 INFO 1000 DEL **FINAL** [32 failures], 6.8/s
swift-bench 2013-07-17 17:37:24,045 INFO Auth version: 1.0

Hi, Clay,

Thank you for answer. 
To test performance with a number of regions, I reconfigured SAIO. Now there are 6 devices, 3 regions, 3 replicas.
There are delete errors and with 3 replica, when affinity is enable:

swift@regions:/home/swift/scenario$ swift-ring-builder /etc/swift/object.builder
/etc/swift/object.builder, build version 6
262144 partitions, 3.000000 replicas, 3 regions, 6 zones, 6 devices, 0.00 balance
The minimum number of hours before a partition can be reassigned is 1
Devices:    id  region  zone      ip address  port  replication ip  replication port      name weight partitions balance meta
             0       1     1       127.0.0.1  6010       127.0.0.1              6010      sdb1   1.00     131072    0.00
             1       1     2       127.0.0.1  6020       127.0.0.1              6020      sdb2   1.00     131072    0.00
             2       2     3       127.0.0.1  6030       127.0.0.1              6030      sdb3   1.00     131072    0.00
             3       2     4       127.0.0.1  6040       127.0.0.1              6040      sdb4   1.00     131072    0.00
             4       3     5       127.0.0.1  6050       127.0.0.1              6050      sdb5   1.00     131072    0.00
             5       3     6       127.0.0.1  6060       127.0.0.1              6060      sdb6   1.00     131072    0.00

swift@regions:/home/swift/scenario$ sudo tail -n 20 /var/log/swift/proxy.error
Jul 17 17:36:49 regions proxy-server Handoff requested (3) (txn: txad01b791de4846b69fe54-0051e69de7) (client_ip: 127.0.0.1)
Jul 17 17:36:50 regions proxy-server ERROR with Object server 127.0.0.1:6030/sdb3 re: Trying to DELETE /AUTH_test/04fbc0d8b76c46e09546c0bbffcae431_1/0508c01c397a4734b44d7d9bf2c77c5d: Timeout (10s) (txn: tx1134b5bfbf9f40d6bf01b-0051e69de8)
Jul 17 17:36:50 regions proxy-server Handoff requested (1) (txn: tx1134b5bfbf9f40d6bf01b-0051e69de8) (client_ip: 127.0.0.1)
Jul 17 17:36:50 regions proxy-server ERROR with Object server 127.0.0.1:6020/sdb2 re: Trying to DELETE /AUTH_test/04fbc0d8b76c46e09546c0bbffcae431_1/0508c01c397a4734b44d7d9bf2c77c5d: Timeout (10s) (txn: tx1134b5bfbf9f40d6bf01b-0051e69de8)
Jul 17 17:36:50 regions proxy-server Handoff requested (2) (txn: tx1134b5bfbf9f40d6bf01b-0051e69de8) (client_ip: 127.0.0.1)
Jul 17 17:36:50 regions proxy-server ERROR with Object server 127.0.0.1:6060/sdb6 re: Trying to DELETE /AUTH_test/04fbc0d8b76c46e09546c0bbffcae431_1/0508c01c397a4734b44d7d9bf2c77c5d: Timeout (10s) (txn: tx1134b5bfbf9f40d6bf01b-0051e69de8)
Jul 17 17:36:50 regions proxy-server Handoff requested (3) (txn: tx1134b5bfbf9f40d6bf01b-0051e69de8) (client_ip: 127.0.0.1)
Jul 17 17:36:52 regions proxy-server ERROR with Object server 127.0.0.1:6020/sdb2 re: Trying to DELETE /AUTH_test/04fbc0d8b76c46e09546c0bbffcae431_2/a6003bf6744e4374a8899841d4177ecf: Timeout (10s) (txn: txb65498b92ace478c86c57-0051e69dea)
Jul 17 17:36:53 regions proxy-server Handoff requested (1) (txn: txb65498b92ace478c86c57-0051e69dea) (client_ip: 127.0.0.1)
Jul 17 17:36:54 regions proxy-server ERROR with Object server 127.0.0.1:6060/sdb6 re: Trying to DELETE /AUTH_test/04fbc0d8b76c46e09546c0bbffcae431_10/d333ce10be484069ae848c88f3a9bd64: Timeout (10s) (txn: tx20d12721ef81478b91f13-0051e69de0)
Jul 17 17:36:54 regions proxy-server ERROR with Object server 127.0.0.1:6030/sdb3 re: Trying to DELETE /AUTH_test/04fbc0d8b76c46e09546c0bbffcae431_18/7295dd70124a4e8b860ee1130e3782de: Timeout (10s) (txn: txe4eb8c4255784af5862a9-0051e69dea)
Jul 17 17:36:55 regions proxy-server Handoff requested (1) (txn: txe4eb8c4255784af5862a9-0051e69dea) (client_ip: 127.0.0.1)
Jul 17 17:36:55 regions proxy-server ERROR with Object server 127.0.0.1:6010/sdb1 re: Trying to DELETE /AUTH_test/04fbc0d8b76c46e09546c0bbffcae431_17/16fec6a3820f4278a2ab2fd49d83bfd5: Timeout (10s) (txn: tx239c586bdd3a43098f93f-0051e69dea)
Jul 17 17:36:55 regions proxy-server Handoff requested (1) (txn: tx239c586bdd3a43098f93f-0051e69dea) (client_ip: 127.0.0.1)
Jul 17 17:36:55 regions proxy-server ERROR with Object server 127.0.0.1:6030/sdb3 re: Trying to DELETE /AUTH_test/04fbc0d8b76c46e09546c0bbffcae431_7/5024db77616241558fb504f6232312e6: Timeout (10s) (txn: tx2d445bec100f437eb5124-0051e69dea)
Jul 17 17:36:56 regions proxy-server Handoff requested (1) (txn: tx2d445bec100f437eb5124-0051e69dea) (client_ip: 127.0.0.1)
Jul 17 17:36:56 regions proxy-server ERROR with Object server 127.0.0.1:6040/sdb4 re: Trying to DELETE /AUTH_test/04fbc0d8b76c46e09546c0bbffcae431_1/1d4eb23b7bb0440698f8f8527d6c4aab: Timeout (10s) (txn: tx55ceb254cc424e56a8958-0051e69dea)
Jul 17 17:36:56 regions proxy-server Handoff requested (1) (txn: tx55ceb254cc424e56a8958-0051e69dea) (client_ip: 127.0.0.1)
Jul 17 17:36:56 regions proxy-server ERROR with Object server 127.0.0.1:6030/sdb3 re: Trying to DELETE /AUTH_test/04fbc0d8b76c46e09546c0bbffcae431_6/1f6286c418c143baa29faf79c8e4a9ca: Timeout (10s) (txn: txd36581b448834e048beea-0051e69dea)
Jul 17 17:36:56 regions proxy-server Handoff requested (1) (txn: txd36581b448834e048beea-0051e69dea) (client_ip: 127.0.0.1)

Will some timeout parameter in swift-bench resolve it? Timeout parameter that will wait for replication to put all replicas to proper places.

Revision history for this message

clayg (clay-gerrard) wrote on 2013-07-18:

No I don't think changing any timeout setting with swift-bench would have an effect.

It's not really swift-bench that was timing out - it just got a bad response from the proxy. It's actually hard to say exactly what happened from only the .error log

I'm not sure really what you're seeing, timeout normally means a network partition or overloaded server. Maybe one of the object servers was misconfigured and didn't start? Do functests pass against this configuration?

Revision history for this message

OpenStack Infra (hudson-openstack) wrote on 2013-07-19: Fix proposed to swift (master)

Fix proposed to branch: master
Review: https://review.openstack.org/37865

Changed in swift:
assignee:	nobody → Ksenia Demina (kdemina)
status:	New → In Progress

Revision history for this message

OpenStack Infra (hudson-openstack) wrote on 2013-08-29: Fix merged to swift (master)

Reviewed: https://review.openstack.org/37865
Committed: http://github.com/openstack/swift/commit/600e9c86722301cbe881f8a7e8a9706bb41ffacc
Submitter: Jenkins
Branch: master

commit 600e9c86722301cbe881f8a7e8a9706bb41ffacc
Author: Ksenia Demina <email address hidden>
Date: Fri Jul 19 14:32:55 2013 +0400

Add delay in swift-bench

    With enable write affinity, it's necessary to wait until
    replication has moved things to their proper homes before
    running delete request. With write affinity turned on, only
    nodes in local region will get the object right after PUT request.

Fix bug #1198926

Change-Id: I3aa8933d45c47a010ae05561e12176479e7c9bcc

Ksenia Svechnikova (kdemina) on 2013-09-10

Changed in swift:
status:	In Progress → Fix Released

Revision history for this message

OpenStack Infra (hudson-openstack) wrote on 2013-09-10: Fix proposed to swift (feature/ec)

#10

Fix proposed to branch: feature/ec
Review: https://review.openstack.org/45897

Revision history for this message

OpenStack Infra (hudson-openstack) wrote on 2013-09-11: Fix merged to swift (feature/ec)

#11

Download full text (22.6 KiB)

Reviewed: https://review.openstack.org/45897
Committed: http://github.com/openstack/swift/commit/2d5210da89380ba5892f56eb4326082ef3e448d2
Submitter: Jenkins
Branch: feature/ec

commit ce12d66cf970aec6dbbf0dc84500665c62638cd5
Author: Clay Gerrard <email address hidden>
Date: Tue Jul 30 11:44:11 2013 -0700

fix swift i18n

Change-Id: I53cea28a6d7593a1b308dbcf77dddf7f40d76cb2

commit 54d5f3bde9c8685a2f5c238fc6af4162a9dae01b
Author: Alex Gaynor <email address hidden>
Date: Mon Sep 9 14:49:39 2013 -0700

Fixed a suite that was over-indented

Change-Id: I3d05b29e57b77c3751d9f5ff694085bd082e8eb1

commit 537626ac6b79679ea775a84266901545d1a6a864
Author: Alex Gaynor <email address hidden>
Date: Mon Sep 9 09:58:26 2013 -0700

Don't stat the path in ``unlink_older_than``

``listdir`` already handles the ENOENT and returns an empty list in
that case.

Change-Id: I597d7ffa9979f668a856519062839505d26129f2

commit 816c73e0151f5d75e591e8bf490974855037ea58
Author: Dirk Mueller <email address hidden>
Date: Sat Sep 7 16:29:15 2013 +0200

Add Apache 2.0 licensing headers

Change-Id: I38fae2a78b2369a897b7f298c1aead9b963bf7c9

commit 00f9d718d2d746ca8664290b79c852bb91fca1cc
Author: Dirk Mueller <email address hidden>
Date: Fri Aug 30 23:56:55 2013 +0200

Move string expansion outside localisation (H702)

    String expansion should be done outside localisation call (_()),
    otherwise there will never be a matching string found in the
    catalogue.

Also enable gating on this Hacking check (H702).

Change-Id: Ie7c89fbfb52629e75d5e68e9afda8bcf50bf4cdd

commit 3d36a76156a5080f29aa32f6fce019d7b4f1e18b
Author: Dirk Mueller <email address hidden>
Date: Wed Aug 28 21:16:08 2013 +0200

Use Python 3.x compatible except construct

    except x,y: was deprected and is removed in Python 3.x.
    Use "except x as y:" instead which works in any Python
    version >= 2.6.

Change-Id: I7008c74b807340f3457d3a0c8bd0b83f23169d14

commit 3102ad48d56ec2df00da66307191a9dd711e4784
Author: Dirk Mueller <email address hidden>
Date: Sat Sep 7 10:14:00 2013 +0200

Do not use locals() for string formatting (H501)

    Fixes a warning triggered by Hacking 0.7.x or newer. There
    is no need to use a positional string formatting here, since
    this is not going to be localized.

Change-Id: Ie38d620aecb0b48cd113af45cc9ca0d61f8f8ff1

commit 698023f477e2b66a1463aeb7082ef4547d0e6dd7
Author: Peter Portante <email address hidden>
Date: Tue Sep 3 11:10:22 2013 -0400

Provide a method for retrieving on-disk metadata

    We hide the internal dictionary for the metadata providing a method to
    retrieve it to abstract away the implementation details of how
    DiskFile object provides and maintains that metadata.

This is in anticipation of the DiskFile API refactoring.

Change-Id: I1c0dc01a4680bd435512405e2d31fba24421720a
Signed-off-by: Peter Portante <email address hidden>

commit 9d98070f7b2e06ac8cc30d12523798df4418eed0
Author: Peter Portante <email address hidden>
Date: Tue Sep 3 16:34:09 2013 -0400

Remove reference to 'file' built-in...

Reviewed:  https://review.openstack.org/45897
Committed: http://github.com/openstack/swift/commit/2d5210da89380ba5892f56eb4326082ef3e448d2
Submitter: Jenkins
Branch:    feature/ec

commit ce12d66cf970aec6dbbf0dc84500665c62638cd5
Author: Clay Gerrard <clay.gerrard@gmail.com>
Date:   Tue Jul 30 11:44:11 2013 -0700

fix swift i18n
    
    Change-Id: I53cea28a6d7593a1b308dbcf77dddf7f40d76cb2

commit 54d5f3bde9c8685a2f5c238fc6af4162a9dae01b
Author: Alex Gaynor <alex.gaynor@gmail.com>
Date:   Mon Sep 9 14:49:39 2013 -0700

Fixed a suite that was over-indented
    
    Change-Id: I3d05b29e57b77c3751d9f5ff694085bd082e8eb1

commit 537626ac6b79679ea775a84266901545d1a6a864
Author: Alex Gaynor <alex.gaynor@gmail.com>
Date:   Mon Sep 9 09:58:26 2013 -0700

Don't stat the path in ``unlink_older_than``
    
    ``listdir`` already handles the ENOENT and returns an empty list in
    that case.
    
    Change-Id: I597d7ffa9979f668a856519062839505d26129f2

commit 816c73e0151f5d75e591e8bf490974855037ea58
Author: Dirk Mueller <dirk@dmllr.de>
Date:   Sat Sep 7 16:29:15 2013 +0200

Add Apache 2.0 licensing headers
    
    Change-Id: I38fae2a78b2369a897b7f298c1aead9b963bf7c9

commit 00f9d718d2d746ca8664290b79c852bb91fca1cc
Author: Dirk Mueller <dirk@dmllr.de>
Date:   Fri Aug 30 23:56:55 2013 +0200

Move string expansion outside localisation (H702)
    
    String expansion should be done outside localisation call (_()),
    otherwise there will never be a matching string found in the
    catalogue.
    
    Also enable gating on this Hacking check (H702).
    
    Change-Id: Ie7c89fbfb52629e75d5e68e9afda8bcf50bf4cdd

commit 3d36a76156a5080f29aa32f6fce019d7b4f1e18b
Author: Dirk Mueller <dirk@dmllr.de>
Date:   Wed Aug 28 21:16:08 2013 +0200

Use Python 3.x compatible except construct
    
    except x,y: was deprected and is removed in Python 3.x.
    Use "except x as y:" instead which works in any Python
    version >= 2.6.
    
    Change-Id: I7008c74b807340f3457d3a0c8bd0b83f23169d14

commit 3102ad48d56ec2df00da66307191a9dd711e4784
Author: Dirk Mueller <dirk@dmllr.de>
Date:   Sat Sep 7 10:14:00 2013 +0200

Do not use locals() for string formatting (H501)
    
    Fixes a warning triggered by Hacking 0.7.x or newer. There
    is no need to use a positional string formatting here, since
    this is not going to be localized.
    
    Change-Id: Ie38d620aecb0b48cd113af45cc9ca0d61f8f8ff1

commit 698023f477e2b66a1463aeb7082ef4547d0e6dd7
Author: Peter Portante <peter.portante@redhat.com>
Date:   Tue Sep 3 11:10:22 2013 -0400

Provide a method for retrieving on-disk metadata
    
    We hide the internal dictionary for the metadata providing a method to
    retrieve it to abstract away the implementation details of how
    DiskFile object provides and maintains that metadata.
    
    This is in anticipation of the DiskFile API refactoring.
    
    Change-Id: I1c0dc01a4680bd435512405e2d31fba24421720a
    Signed-off-by: Peter Portante <peter.portante@redhat.com>

commit 9d98070f7b2e06ac8cc30d12523798df4418eed0
Author: Peter Portante <peter.portante@redhat.com>
Date:   Tue Sep 3 16:34:09 2013 -0400

Remove reference to 'file' built-in
    
    Change-Id: Ie79e8ede393e92824fd906df1ff1933193c00943
    Signed-off-by: Peter Portante <peter.portante@redhat.com>

commit 12f95cc5a1d1427470f6a3cd953421005dbe51a9
Author: Alex Gaynor <alex.gaynor@gmail.com>
Date:   Wed Sep 4 16:00:44 2013 -0700

Perform fewer stat calls when doing auditing of objects
    
    Each `isdir` or `exists` call performs a stat, but that info is already
    available from the exception if it doesn't exist in `listdir`.
    
    Change-Id: I831494e3dbc8fda62ab29431471566bdb8dc6d27

commit fdfa34e119f4fa751775edca0353bae77162e6e4
Author: Peter Portante <peter.portante@redhat.com>
Date:   Tue Sep 3 10:34:23 2013 -0400

Add a delete() method to DiskFile
    
    In preparation for making the DiskFile class an abstraction layer for
    the backend to all the object server REST API services, we add a
    simple delete method which encapsulates the required behavior.
    
    Change-Id: I539ad53c297e4c572e2f93c2524188fb89b2db40
    Signed-off-by: Peter Portante <peter.portante@redhat.com>

commit 9e7c56a2c98907c6eb0f4ce41302efd682106f80
Author: Peter Portante <peter.portante@redhat.com>
Date:   Tue Sep 3 10:26:39 2013 -0400

Rename DiskFile.writer to create
    
    Change-Id: I3af71c64dd4c61e665fe128a0b5241fd87ba71e0
    Signed-off-by: Peter Portante <peter.portante@redhat.com>

commit 57f97e71d7685249a6f42f2a27f2739a20ef4c2c
Author: Florian Hines <syn@ronin.io>
Date:   Thu Sep 5 18:36:01 2013 -0500

Remove dupe account_audit when reporting stats
    
    - Remove duplicate account_audit call that happened once an hour when stats
    where reported.
    - And fixed a flake8 complaint
    
    Change-Id: Ide686c475f8648664e9cffdd696629cf6ba9f75c

commit a9aec73098d5f4a7ab490bb16faab0834598243a
Author: John Dickinson <me@not.mn>
Date:   Thu Sep 5 12:27:18 2013 -0700

add reseller_admin_role to sample config
    
    Change-Id: Ia8e62eef5af9e849e86c3ff14ce7f8aaa5f21abf

commit 22451b22cb7ec1fbc9b9940ae8f0ef346a36e3ca
Author: Peter Portante <peter.portante@redhat.com>
Date:   Sun Sep 1 15:10:39 2013 -0400

Pep8 final two unit test modules and enforce (12 of 12)
    
    We also fix up any other pep8 failures that snuck in from merges along
    the way.
    
    Change-Id: I4ea984780ac2eac458c98fe181684eef4e04beaf
    Signed-off-by: Peter Portante <peter.portante@redhat.com>

commit 14037b46e694399eb722c89665680a05684c2d33
Author: Peter Portante <peter.portante@redhat.com>
Date:   Sun Sep 1 12:50:07 2013 -0400

Flake8 container test_server.py (11 of 12)
    
    Change-Id: Ie06ed2955838556d5399a49ba3599dfbca4f7512
    Signed-off-by: Peter Portante <peter.portante@redhat.com>

commit 9cdee330fb13c44f4f247402408ea0e33aa761ed
Author: Peter Portante <peter.portante@redhat.com>
Date:   Wed Sep 4 09:21:43 2013 -0400

Refactor finalize_put as an object method
    
    Change-Id: Iec3f53c564505c29739bb5531a71a0a101f64193
    Signed-off-by: Peter Portante <peter.portante@redhat.com>

commit aae122dd4c2fad215c6ce886efc8189914918c3a
Author: Peter Portante <peter.portante@redhat.com>
Date:   Tue Sep 3 22:31:30 2013 -0400

Rename param to avoid use of built-in name hash
    
    Change-Id: I505e8b15ca41c52cd4942a0a7dd34996a1849bb3
    Signed-off-by: Peter Portante <peter.portante@redhat.com>

commit 8d614eeb4aead01e3b953ccfc3d9368fd1814e4a
Author: Alex Gaynor <alex.gaynor@gmail.com>
Date:   Tue Sep 3 10:47:06 2013 -0700

Made the HTTP comparisons slightly more idiomatic
    
    This way they are more obviously range tests
    
    Change-Id: Ia782a8f5ddb5958f9da828e2caadf0a8e3c3ff11

commit 6ae4e17af1bf6f0cbc6a5266564b8097260277ca
Author: Peter Portante <peter.portante@redhat.com>
Date:   Sun Sep 1 01:54:03 2013 -0400

Pep8 account and proxy server unit tests (10 of 12)
    
    Change-Id: Ib83d164997b0d98be921c8b4857caa2429344aa4
    Signed-off-by: Peter Portante <peter.portante@redhat.com>

commit 1d9213bed4c930ba2d0f75e9e0e17944b09b96a7
Author: Peter Portante <peter.portante@redhat.com>
Date:   Sun Sep 1 01:30:36 2013 -0400

Pep8 container test_sync and test_diskfile (9 of 12)
    
    Change-Id: I469e3b75b6846fddb5e9d2b8317506129836d0a6
    Signed-off-by: Peter Portante <peter.portante@redhat.com>

commit b5a0b830e247517f9ec717b03d77b37806ac66e5
Author: Peter Portante <peter.portante@redhat.com>
Date:   Sun Sep 1 01:14:40 2013 -0400

Pep8 remaining unit test modules in common (8 of 12)
    
    Change-Id: I6fa3291eeacb7ee5c095ad9bccbd33f027bf11e3
    Signed-off-by: Peter Portante <peter.portante@redhat.com>

commit 0ebddd87d159072daa0ff36d0ed8af8e4b60fd66
Author: Peter Portante <peter.portante@redhat.com>
Date:   Sun Sep 1 00:35:53 2013 -0400

Pep8 unit tests in middleware > 20 violations (7 of 12)
    
    Change-Id: I9dfae1a473a8212ef25dcc01e338d8bdade7ef4e
    Signed-off-by: Peter Portante <peter.portante@redhat.com>

commit 56593a1323455ced7047f05fa638b9641fec72f1
Author: Peter Portante <peter.portante@redhat.com>
Date:   Sat Aug 31 23:42:43 2013 -0400

Pep8 unit test modules w/ <= 20 violations (6 of 12)
    
    Change-Id: I7317beb97e1530cb18c62da55ccf4c64206ff362
    Signed-off-by: Peter Portante <peter.portante@redhat.com>

commit be1cff4f1f14bff7d310749d1424104f05883b36
Author: Peter Portante <peter.portante@redhat.com>
Date:   Sat Aug 31 23:13:15 2013 -0400

Pep8 unit test modules w/ <= 10 violations (5 of 12)
    
    Change-Id: I8e82c14ada52d44df5a31e08982ac79cd7e5c969
    Signed-off-by: Peter Portante <peter.portante@redhat.com>

commit c067abd21ee05ae96b0f70c6058efc1bfde10de4
Author: Peter Portante <peter.portante@redhat.com>
Date:   Sat Aug 31 22:36:58 2013 -0400

Pep8 unit test modules for hacking and one liners (4 of 12)
    
    Address all the "hacking" lines that are flagged, and all the modules
    that just have one item flagged.
    
    Change-Id: I372a4bdf9c7748f73e38c4fd55e5954f1afade5b
    Signed-off-by: Peter Portante <peter.portante@redhat.com>

commit c57f526f58ba0a529c9f1697c3419ebfec1b0a90
Author: Peter Portante <peter.portante@redhat.com>
Date:   Sat Aug 31 20:56:39 2013 -0400

Pep8 the functionalnosetests tests (3 of 12)
    
    Change-Id: Ia88356d6d36d1c4914f706652aaac5c840a79f13
    Signed-off-by: Peter Portante <peter.portante@redhat.com>

commit 0014f2c828dae5f9481a2bac68fe90a7857b2f59
Author: Peter Portante <peter.portante@redhat.com>
Date:   Sat Aug 31 20:25:25 2013 -0400

Pep8 the functional tests (2 of 12)
    
    Change-Id: Id8a9ae0c82052baccf8bdacd6aaeb76a76ad2ce1
    Signed-off-by: Peter Portante <peter.portante@redhat.com>

commit 0b7432bc4d5d42f9b630e85aadbe22591d2543f9
Author: Peter Portante <peter.portante@redhat.com>
Date:   Sat Aug 31 20:07:08 2013 -0400

Pep8 the probe tests (1 of 12)
    
    Change-Id: I20571816d9838e125a60a2775219b2f3cbb36ddd
    Signed-off-by: Peter Portante <peter.portante@redhat.com>

commit 68da00aae390ceda0ca301866e65d3542e8b6a64
Author: Clay Gerrard <clay.gerrard@gmail.com>
Date:   Fri Aug 30 21:37:07 2013 -0700

Always format messages sent to FakeLogger
    
    Change-Id: I0268587cce0d47b69470b7c4f9224e180fb98f77

commit 016e22469a3ca111a1562db6d04fae20febf2dbb
Author: Clay Gerrard <clay.gerrard@gmail.com>
Date:   Fri Aug 30 21:01:12 2013 -0700

Fix string format error in DiskFile close
    
    Add a 'get_lines_for_level' method to FakeLogger in order to actually
    validate the log line format string is valid for the given args.
    
    Change-Id: Ib2559ddf5c9d89e17c6ac08d26cc46a4193a2df2

commit 4893aacc67ae4167674d1eaffc924cdc91774f44
Author: David Goetz <dpgoetz@gmail.com>
Date:   Tue Jul 23 15:00:13 2013 -0700

Allow users to extract tars to containers with ACLs set
    
    This fixes bug: https://bugs.launchpad.net/swift/+bug/1203182 (well
    it should :) I don't have keystone installed but the same issue
    existed with tempauth).
    
    Change-Id: I6f4045f484b27c0153a9244e0dbf2641cbc9e84e

commit 45eb76370c949433ed2d078ba81ae95cfe474bfd
Author: Alex Gaynor <alex.gaynor@gmail.com>
Date:   Fri Aug 30 14:30:39 2013 -0700

Made the signature of HeaderKeyDict.__init__ match dict's
    
    Change-Id: I75dc6228bc39de44f58d7ae9beb8ca3bf24b100f

commit 921e5835b827d85021ea256840494102594ecd46
Author: Alex Gaynor <alex.gaynor@gmail.com>
Date:   Fri Aug 30 13:21:20 2013 -0700

Two small performance improvements to header generation
    
    These both apply to the proxy servers:
    
    First, instead of consturction a dict just to use with `.update()` simply set a
    set of keys in the headers dict.
    
    Second, generating the headers for proxy requests once, no matter how many
    backend servers are accessed (instead of once per backend server), they all get
    the same headers.
    
    Change-Id: Ib42bf7750e175a6ad9b777fa35655380feeafb93

commit fc9cd6bf5a821682a91cbbe5a8110046fc6ba3f7
Author: Chuck Thier <cthier@gmail.com>
Date:   Fri Aug 30 20:09:26 2013 +0000

Fix changes in XML that broke old clients
    
    Certain old clients that shall not be named do not like some of the
    recent changes to the xml output for container listings.  This fixes
    those changes.  If we want to change any of the xml output, then we
    should do it in another API version.  I'm not sure if any of these
    changes in the XML were put in any documentation, but if they were, they
    should be changed back.
    
    DocImpact
    
    Change-Id: I983332ca9d620a9a539be26bc6655df9b2ef87ed

commit 6570c24d4516a27d95e5937919ac24a155a5cf94
Author: Chuck Thier <cthier@gmail.com>
Date:   Thu Aug 29 16:32:30 2013 +0000

Only set PYTHON_EGG_CACHE if it isn't already set
    
    Change-Id: I6503c80a236a1720562377d04313ee2688336c85

commit a8fcc946b650305ab3dc0324f2b7e799d817dc4d
Author: Samuel Merritt <sam@swiftstack.com>
Date:   Fri Aug 30 09:47:11 2013 -0700

Small test cleanups from b63089d.
    
    There was nothing there worth holding up the review over, but there
    were a couple little things in the test that confused me.
    
    Change-Id: I85c9c1bdb12de210c198d0d38404a1cf9f73adbb

commit 62ae88b368c6c27a03a8d1adf507df78a4c0cae7
Author: Alex Gaynor <alex.gaynor@gmail.com>
Date:   Fri Aug 30 07:38:57 2013 -0700

Defer computation of available methods until needed.
    
    In practice this data is almost only ever needed if we're throwing an
    HTTPMethodNotAllowed error, so we can avoid computing on every single
    request to the proxy server.
    
    Change-Id: I2abafa3aeb63ca6974dfcf6f31c864c049681f37

commit b63089daae2a9e2b5644b9e720b8dd2f6333e5d8
Author: Peter Portante <peter.portante@redhat.com>
Date:   Wed Aug 28 18:21:45 2013 -0400

Refactor DiskFile init logic into multiple methods
    
    The DiskFile constructor needs to grow a bit in the future it is was
    already considered "large" by some. This breaks the logic up into
    three main parts:
    
      1. Determine if the object exists, fetching the on-disk file(s) to
         use
      2. Finish the object construction when we have a tombstone file
      3. Finish the object construction when we have a regular data file,
         and optionally a fast-POST metadata file
    
    We added a number of assertions as a way of documenting the expected
    behavior of the loop that figures out the on-disk file to use, and a
    set of unit tests to verify the loop works as expected.
    
    Change-Id: I1503b7561e280ded3222a09514edc59ac26dbc65
    Signed-off-by: Peter Portante <peter.portante@redhat.com>

commit 27c7abfb696a558a577a21054d77e474417e2bcb
Author: Clay Gerrard <clay.gerrard@gmail.com>
Date:   Wed Aug 28 23:29:18 2013 -0700

Add constructor args to swob.Request.blank
    
    I knew webob.Request.blank could take most of the attributes on the class as
    kwargs to blank, so I went and looked how.  It seems to work ok and is pretty
    nice.
    
    Change-Id: I72fae7c28f81c97768ee98b8ebcd69789a4c2e84

commit da42fe01d99eaeb0cda5387efefeb9e284c6e20b
Author: Yuan Zhou <yuan.zhou@intel.com>
Date:   Fri Aug 16 16:55:39 2013 +0800

Clean up utf8ness quote
    
    Move common codes on utf8ness and quote to common/utils.py
    
    Change-Id: I91d98a06fa94ac608119a0d70adedc6d73337c64

commit 0fc0fba98680495b87a6194312b002fdceef3ab2
Author: Peter Portante <peter.portante@redhat.com>
Date:   Wed Aug 28 12:51:09 2013 -0400

Remove duplicate code for update_put_timestamp().
    
    Change-Id: I035283f30257961d6efb98f1bb65d27bf6a5bbf5
    Signed-off-by: Peter Portante <peter.portante@redhat.com>

commit e19cf5c0475fcc69cda1a883f29181247c4c8f54
Author: Samuel Merritt <sam@swiftstack.com>
Date:   Mon Aug 26 17:41:59 2013 -0700

Clean up some of the proxy unit tests.
    
    Since we've got machinery to catch HTTPException errors that happens
    after the proxy server's __call__ method but before the individual
    GET()/POST()/etc. methods, we shouldn't just call
    GET()/POST()/etc. Instead, we should use swob.Request's get_response()
    method, or otherwise actually call the proxy application. That way we
    can have the code raise HTTPBadRequest, but the test just gets a
    plain old response with a 400 status code, just like you'd expect.
    
    Earlier changes fixed some of the tests; this commit fixes some
    more. In the interest of not having a gigantic diff, this only gets
    about half the proxy's unit tests, and the others will be done in a
    subsequent commit.
    
    Change-Id: Iaca32e74a8f4499440e655c207a1012e14486c74

commit 092ca2d2df1107b78d0d7c7c53d3f13d8ad53fe6
Author: Chuck Thier <cthier@gmail.com>
Date:   Wed Aug 28 21:26:08 2013 +0000

Added test for timeout regression
    
    Change-Id: Ia0e5f1595d1b1f58e6c415d7397f33496576c944

commit 9dbe7cad064ee6ce5cd3f4c5f56a01dfab82df05
Author: Chuck Thier <cthier@gmail.com>
Date:   Wed Aug 28 15:40:16 2013 +0000

Fix regression that causes Timeouts to return 500
    
    When the hacking changes were put in, one of the exceptions that were
    fixed didn't also add a catch for Timeout, which causes certain timeouts
    to cause a 500 return
    
    Change-Id: I7a9db200528303868d8a9ab05f35a70d99bfb118

commit 3947a4fbd0223a40e6247df9477af9e0a47db753
Author: Peter Portante <peter.portante@redhat.com>
Date:   Wed Aug 28 00:39:48 2013 -0400

Don't use float() when normalize_timestamp() does it.
    
    And add a comment to note for the curious when the current timestamp
    will bump to 11 digits of seconds (which implies that one would no
    longer be able to use string compare on timestamps of differing
    length).
    
    Change-Id: I823e7659c41a54915d7c9c64c9161a29e79cd48c
    Signed-off-by: Peter Portante <peter.portante@redhat.com>

commit 61a8a9af18f501e2b7940666b4131dfaf713d6a1
Author: Samuel Merritt <sam@swiftstack.com>
Date:   Fri Aug 16 16:24:00 2013 -0700

Refactor path splitting and validation.
    
    The account, container, and object servers all do the same thing at
    the top of GET/PUT/etc.: they split the path, validate that the first
    two components are a good (device, partition) pair, and return a 400
    if they're not. The object server already had a module-local helper
    function for this, but the account and container servers just had
    duplicate boilerplate code. Now it's all in one common helper.
    
    Change-Id: I9a20d37fc9e1a68b10149a7aa78cb9691fc04ea9

commit 0c505d4456057ca3d401e519916748ed4c9ee822
Author: Samuel Merritt <sam@swiftstack.com>
Date:   Mon Aug 26 19:00:46 2013 -0700

Fix setdefault() for swob's HeaderKeyDict.
    
    You'd think this would just work, given that HeaderKeyDict inherits
    from dict and overrides the usual __thingy__ methods, but it
    doesn't. It would work if you title-cased the key, but the whole point
    of HeaderKeyDict is to do that for you.
    
    Change-Id: If5c22df0690a245d1dd02fa3a52fa135235fe60d

commit 82ab838ffa950f3a92e98466834ab925e68848e5
Author: Clay Gerrard <clay.gerrard@gmail.com>
Date:   Mon Aug 26 14:23:22 2013 -0700

remove unused reference to resellers_conf
    
    Change-Id: I956f25f44b20f65c72beddb2cd49225bdecc8430

commit bdba16035dd03310259203f7baf576fc6fcb530b
Author: Zap Chang <zapchang@gmail.com>
Date:   Wed Aug 21 22:13:16 2013 +0800

Improve swift-bench performance
    
    Single-host bench controller should be monkey-patched to achieve
    maximum concurrency.
    
    Change-Id: I77c63773edc7427d06c2fbfb67ededf36d9913d3

commit dd9eda60d8334e30772e81ffbd286569d36192b0
Author: Peter Portante <peter.portante@redhat.com>
Date:   Fri Jul 12 13:27:56 2013 -0400

get rid of unlinkold
    
    We currently have two methods for cleaning up object directories - unlinkold in
    the object server and hash_suffix in the replicator.
    
    This modifies the object server to reuse the more complete hash_suffix logic.
    
    Change-Id: Id2fd83caad979bfd446d52a775a1fa373a670cf1
    Signed-off-by: Peter Portante <peter.portante@redhat.com>

commit a4f371439bbe3195da22daac621ca331e02a9ae9
Author: Samuel Merritt <sam@swiftstack.com>
Date:   Wed Aug 14 11:55:15 2013 -0700

Refactor how we pick listings' content type.
    
    There were a few different places where we had some repeated code to
    figure out what format an account or container listing response should
    be in (text, JSON, or XML). Now that's been pulled into a single
    function.
    
    As part of this, you can now raise HTTPException subclasses in proxy
    controllers instead of laboriously plumbing error responses all the
    way back up to swift.proxy.server.Application.handle_request(). This
    lets us avoid certain ugly patterns, like the one where a method
    returns a tuple of (x, y, z, error) and the caller has to see if it
    got (value, value, value, None) or (None, None, None, errorvalue). Now
    we can just raise the error.
    
    Change-Id: I316873df289160d526487ad116f6fbb9a575e3de

commit 908e5fcb70e82cbd38e49768cff524ba3cc1571b
Author: Kun Huang <academicgareth@gmail.com>
Date:   Tue Jul 16 16:28:32 2013 +0800

[tempauth] update token if its value changes or expires
    
    In current codes, tempauth token will be updated only when it expires. In
    token's lifetime, any changes in tempauth config file doesn't update token,
    which means if you update your config file and restart proxy server, the
    role in runtime might not change. So a good fix is updating token when
    it changes or expires.
    
    fixes bug #1201678
    
    Change-Id: Ieb080e87268e16d0a2e5896431aa41fcedc7a900

commit d3e2cb35db5c9d90c35ec6440a9ffd18f2d33aa4
Author: Samuel Merritt <sam@swiftstack.com>
Date:   Tue Aug 13 14:54:16 2013 -0700

Fix unbalanced rings after initial creation.
    
    Sometimes you'd make a ring that should have a good balance (~= 1),
    but it would come out with a poor balance (>= 90). Rebalancing a
    couple times would fix it, but it shouldn't be necessary to do that.
    
    This was due to a bug in the ring builder's rebalance() method where,
    when placing a replica for a partition, it would pick *one of* the
    tiers with the lowest number of replicas, but the choice was made
    without regard to parts_wanted.  Now rebalancing pays attention to the
    drives' parts_wanted attributes, so the tier with the emptiest drive
    gets the replica.
    
    (Also fixed a typo in a completely unrelated comment.)
    
    Change-Id: I1821f15170f0ed212b0ad7c81b97031540ab4460

commit 600e9c86722301cbe881f8a7e8a9706bb41ffacc
Author: Ksenia Demina <kdemina@mirantis.com>
Date:   Fri Jul 19 14:32:55 2013 +0400

Add delay in swift-bench
    
    With enable write affinity, it's necessary to wait until
    replication has moved things to their proper homes before
    running delete request. With write affinity turned on, only
    nodes in local region will get the object right after PUT request.
    
    Fix bug #1198926
    
    Change-Id: I3aa8933d45c47a010ae05561e12176479e7c9bcc

Thierry Carrez (ttx) on 2013-09-19

Changed in swift:
status:	Fix Released → Fix Committed

Thierry Carrez (ttx) on 2013-10-09

Changed in swift:
milestone:	none → 1.10.0-rc1
status:	Fix Committed → Fix Released

Thierry Carrez (ttx) on 2013-10-17

Changed in swift:
milestone:	1.10.0-rc1 → 1.10.0

Ksenia Svechnikova (kdemina) on 2015-09-17

no longer affects:

python-swiftclient

Report a bug

This report contains Public information

Everyone can see this information.

You are

Subscribing...

Edit bug mail

Other bug subscribers

Remote bug watches

Bug watches keep track of this bug in other bug trackers.