Reviewed:  https://review.openstack.org/126595
Committed: https://git.openstack.org/cgit/openstack/swift/commit/?id=06800cbe446ce4c937a57b69517b55c3bba9b6e1
Submitter: Jenkins
Branch:    feature/ec

commit 7528f2b22169e90fe8ddd19b7ef7d46ecff5d231
Author: Christian Schwede <email address hidden>
Date:   Mon Oct 6 10:01:03 2014 +0000

    Fix minor typo
    
    Fixes minor typo in one method and adds missing parameter in other
    method. Only checked swift/container/reconciler.py for now.
    
    Change-Id: I5c648010f09b6e4b1fb0380bc97b266e680602f8

commit 94fd95ba30c72fbcb03367aaa8da407a408948d5
Author: OpenStack Proposal Bot <email address hidden>
Date:   Sat Oct 4 06:07:47 2014 +0000

    Imported Translations from Transifex
    
    Change-Id: I31b5e6b0f2922150902e1bfa52144302ee0c7a8e

commit d6a827792619f3343af07fc2519f4253fbdc67f7
Author: John Dickinson <email address hidden>
Date:   Fri Oct 3 10:17:00 2014 -0400

    updated AUTHORS and CHANGELOG for 2.2.0
    
    Change-Id: I6c0bc1570f6a48439de5a029a86f1b582f30f8a6

commit 5b2c27a5874c2b5b0a333e4955b03544f6a8119f
Author: Richard (Rick) Hawkins <email address hidden>
Date:   Wed Oct 1 09:37:47 2014 -0400

    Fix metadata overall limits bug
    
    Currently metadata limits are checked on a per request basis. If
    multiple requests are sent within the per request limits, it is
    possible to exceed the overall limits.  This patch adds an overall
    metadata check to ensure that multiple requests to add metadata to
    an account/container will check overall limits before adding
    the additional metadata.
    
    Change-Id: Ib9401a4ee05a9cb737939541bd9b84e8dc239c70
    Closes-Bug: 1365350

commit 301a96f664d58b4ccad8e3cbf5d5a889cc76790f
Author: Jay S. Bryant <email address hidden>
Date:   Tue Sep 30 15:08:59 2014 -0500

    Ensure sys.exit called in fork_child after exception
    
    Currently, the fork_child() function in auditor.py does not
    handle the case where run_audit() encounters an exception
    properly.
    
    A simple case is where the /srv directory is set
    with permissions such that the 'swift' user cannot access it.
    Such a situation causes a os.listdir() to return an OSError
    exception.  When this happens the fork_child() process does not
    run to completion and sys.exit() is not executed.  The process
    that was forked off continues to run as a result.  Execution goes
    back up to the audit_loop function which restarts the whole process.  The
    end result is an increasing number of processes on the system
    until the parent is terminated.  This can quickly exhaust the
    process descriptors on a system.
    
    This change wraps run_audit() in a try block and adds an
    exception handler that prints what exception was encountered.
    The sys.exit() was moved to a finally: block so that it will
    always be run, avoiding the creation of zombies.
    
    Change-Id: I89d7cd27112445893852e62df857c3d5262c27b3
    Closes-bug: 1375348

commit 6d49cc3092168de6d22378557b2c37ea4063beeb
Author: Samuel Merritt <email address hidden>
Date:   Thu Oct 2 17:14:58 2014 -0400

    Fix ring-builder crash.
    
    If you adjust the replica count after adding devices but before the
    first rebalance, the ring builder would crash.
    
    To reproduce:
    
        $ swift-ring-builder crashy.builder create 8 3 1
        $ swift-ring-builder crashy.builder add r1z1-10.1.1.1:6000/sda1 1
        $ swift-ring-builder crashy.builder set_replicas 5
        $ swift-ring-builder crashy.builder rebalance
    
    Change-Id: Id91cf8680961ccf6e3db153577e99cad545bee9d

commit 702d88bc89f56b96f9e47dc0923f66cbc7cbd5f6
Author: Samuel Merritt <email address hidden>
Date:   Wed Oct 1 14:03:51 2014 -0400

    fix a comment
    
    Change-Id: I6310390f756bc0d326e56fceef32b630f0fe1ca1

commit 64aa8062bc96799993deafbdf184adb72c509f20
Author: YummyBian <email address hidden>
Date:   Sat Sep 27 23:49:38 2014 +0800

    Some statements are evaluated twice in the setUp of the
    TestObjectReplicator
    
    Remove the duplicated statements.
    
    Closes-Bug: #1374783
    
    Change-Id: If2b55e864fea497d7a7b218adf11eb7749c27765

commit 20e9ad538b6ba90674a75310d9cc184162ba9398
Author: Christian Schwede <email address hidden>
Date:   Sun Sep 14 20:48:32 2014 +0000

    Limit partition movement when adding a new tier
    
    When adding a new tier (region, zone, node, device) to an existing,
    already balanced ring all existing partitions in the existing tiers of
    the same level are gathered for reassigning, even when there is
    not enough space in the new tier. This will create a lot of unnecessary
    replication traffic in the backend network.
    
    For example, when only one region exists in the ring and a new region is
    added, all existing parts are selected to reassign, even when the new
    region has a total weight of 0. Same for zones, nodes and devices.
    
    This patch limits the number of partitions that are choosen to reassign
    by checking for devices on other tiers that are asking for more
    partitions.
    
    Failed devices are not considered when applying the limit.
    
    Co-Authored By: Florent Flament <email address hidden>
    Change-Id: I6178452e47492da4677a8ffe4fb24917b5968cd9
    Closes-Bug: 1367826

commit 2ce6341b527e81a43513adef6f5c54b4e09e27a4
Author: Pete Zaitcev <email address hidden>
Date:   Mon Sep 29 14:49:42 2014 -0600

    Fix up the return value of launch()
    
    This is "obviously" the right thing to do, except of course it's
    pure sugar: the return value is not used anywhere. Except if someone
    has a script that imports the whole thing somehow and then does
    isinstance(dict). Because that is so much easier than submitting
    a patch, I can imagine someone, somewhere doing that.
    
    The fix came in a patch by dfg, see review 121851.
    
    Change-Id: I39ddf70dc2d027b7db6e31097400248dc66eb137

commit e2285e46db11dd59edbe076540ac0db24ef6f3f2
Author: OpenStack Proposal Bot <email address hidden>
Date:   Sun Sep 28 06:08:55 2014 +0000

    Imported Translations from Transifex
    
    Change-Id: I59b68ecb6f07e2b71e4655699d0165007d764718

commit 041cf67c36988e3aee0eab0e597d78cec8688df0
Author: YummyBian <email address hidden>
Date:   Sat Sep 27 23:10:48 2014 +0800

    Incorrect variable name in the diskfile.py
    
    You should use the suffix variable instead of the hash_ variable when
    updates the hashes dictionary in the get_hashes().
    
    Fixes Bug1374777
    
    Change-Id: If9a7ab5b6d271385e4fdc79ff9c216047a2f4d7e

commit 0f93fff46ad098a360f719a762dde4d17e852fbc
Author: Keshava Bharadwaj <email address hidden>
Date:   Thu Sep 25 00:48:49 2014 +0530

    Fixes unit tests to clean up temporary directories
    
    This patch fixes the unit tests to remove the temporary directories
    created during run of unit tests. Some of unit tests did not tear down
    correctly, whatever it had set it up for running. This would over period
    of time bloat up the tmp directory. As on date, there were around 49 tmp
    directories left uncleared per round of unit tests. This patch fixes it.
    
    Change-Id: If591375ca9cc87d52c7c9c6dc16c9fb4b49e99fc

commit 38ba5790fb527967c2fcbaf094e76a73f4b94d38
Author: Keshava Bharadwaj <email address hidden>
Date:   Thu Sep 18 00:45:35 2014 +0530

    Provides proper error handling on builder unpickle
    
    This patch provides the necessary error handling while unpickling
    a builder file. Earlier if a builder file is empty/invalid/corrupted,
    the stacktrace was shown to user with an exit code of 1. This fixes it
    to show a user-friendly message and also returns the exit code of 2,
    indicating there was a failure.
    
    Change-Id: I51eb24702c422299629f8053d4591dd10f5863f8
    Closes-Bug: #1370680

commit e567722c4ea125f4607093112396e16ae6c6dd42
Author: John Dickinson <email address hidden>
Date:   Mon Sep 22 09:46:34 2014 -0700

    updated hacking rules
    
    1) Added comment for H231, which we were already enforcing. H231
    is for Python 3.x compatible except statements.
    
    2) Added check for H201, which we were enforcing in reviews
    but waiting on hacking checks to be updated. H201 is for bare
    except statements, and the update in upstream hacking is to
    support the "  # noqa" flag on it.
    
    The H201 check catches some existing bare excepts that are fixed.
    
    Change-Id: I68638aa9ea925ef62f9035a426548c2c804911a8

commit 781a1620048edbbc057b15c94da390847b34d9df
Author: Michael Barton <email address hidden>
Date:   Thu Sep 25 16:29:57 2014 +0000

    use get_container_info in ratelimit
    
    Everyone else cooperates to store the container_info cache in the env the
    first time it's requested.  This should save a duplicate memcache hit on
    requests that ratelimiting looks at.
    
    Change-Id: Ic6411d4619db6b53fafe9fdbf1d0a370d1258c38

commit 23c2d95e98ee5edf317da55e3ee4c303a18991b6
Author: Samuel Merritt <email address hidden>
Date:   Tue Sep 23 12:32:42 2014 -0700

    Remove unused option from ObjectReplicator.__init__
    
    disk_chunk_size is unused by the ObjectReplicator class. Note that it
    *is* used by the object replicator process; it's just that
    DiskFileManager is the one that pulls it out of the conf and pays
    attention to it. Keeping it as an attribute on ObjectReplicator is
    unnecessary.
    
    Change-Id: I1eeef7b33873b4c8bb269ca02dcb067098b6fded

commit 6354e2da57a8d487caf3605d4005134f584cf935
Author: Gerry Drudy <email address hidden>
Date:   Mon Sep 15 11:52:14 2014 +0100

    direct_client not passing args between some functions
    
    The call to _get_direct_account_container in direct_get_account
    has several of its args =None instead of set to the value passed
    to direct_get_account.
    
    The same applies to _get_direct_account_container in
    direct_get_container.
    
    The direct_get_container is only called by the account-reaper
    and this bug will have limited impact on it. The marker,
    maintained in reap_container, is ignored by direct_get_container.
    This is not as bad as it sounds, if the account-reaper successfully
    deletes the first 10K objects, assuming the container has > 10K
    objects, the next call to direct_get_container will in fact return
    the next 10K objects even though it sets marker=None (assuming the
    first 10K objects were successfully deleted).
    
    This patch also updates test_direct_get_account and
    test_direct_get_container to ensure the appropriate
    args are included in the connection query_string.
    
    Closes-Bug: #1369558
    Change-Id: If1c8aa1240d38354ebc9b1ebca92dc1c8c36cb5f

commit c767ec9a37faa4414702aa9e39231afe0ba3b097
Author: Clay Gerrard <email address hidden>
Date:   Wed Jul 9 16:04:20 2014 -0700

    Let eventlet.wsgi.server log tracebacks when eventlet_debug is enabled
    
    The "logging" available in eventlet.wsgi.server/BaseHTTPServer doesn't
    generally suite our needs, so it should be bypassed using a NullLogger in
    production.  But in development it can be useful if tracebacks generated from
    inside eventlet.wsgi (say a NameError in DiskFile.__iter__) end up in logs.
    Since we already have eventlet_debug parsed inside of run_server we can skip
    the NullLogger bypass and let stuff blast out to STDERR when configured for
    development/debug logging.
    
    Change-Id: I20a9e82c7fed8948bf649f1f8571b4145fca201d