CHANGELOG

swift (1.13.1)

    * Change the behavior of CORS responses to better match the spec

      A new proxy config variable (strict_cors_mode, default to True)
      has been added. Setting it to False keeps the old behavior. For
      an overview of old versus new behavior, please see
      https://review.openstack.org/#/c/69419/

    * Invert the responsibility of the two instances of proxy-logging in
      the proxy pipeline

      The first proxy_logging middleware instance to receive a request
      in the pipeline marks that request as handling it. So now, the
      left most proxy_logging middleware handles logging for all
      client requests, and the right most proxy_logging middleware
      handles all other requests initiated from within the pipeline to
      its left. This fixes logging related to large object
      requests not properly recording bandwidth.

    * Added swift-container-info and swift-account-info tools

    * Allow specification of object devices for audit

    * Dynamic large object COPY requests with ?multipart-manifest=get
      now work as expected

    * When a client is downloading a large object and one of the segment
      reads gets bad data, Swift will now immediately abort the request.

    * Fix ring-builder crash when a ring partition was assigned to a
      deleted device, zero-weighted device, and normal device

    * Make probetests work with conf.d configs

    * Various other minor bug fixes and improvements.

swift (1.13.0)

    * Account-level ACLs and ACL format v2

      Accounts now have a new privileged header to represent ACLs or
      any other form of account-level access control. The value of
      the header is a JSON dictionary string to be interpreted by the
      auth system. A reference implementation is given in TempAuth.
      Please see the full docs at
      http://swift.openstack.org/overview_auth.html

    * Added a WSGI environment flag to stop swob from always using
      absolute location. This is useful if middleware needs to use
      out-of-spec Location headers in a response.

    * Container sync proxies now support simple load balancing

    * Config option to lower the timeout for recoverable object GETs

    * Add a way to ratelimit all writes to an account

    * Allow multiple storage_domain values in cname_lookup middleware

    * Moved all DLO functionality into middleware

      The proxy will automatically insert the dlo middleware at an
      appropriate place in the pipeline the same way it does with the
      gatekeeper middleware. Clusters will still support DLOs after upgrade
      even with an old config file that doesn't mention dlo at all.

    * Remove python-swiftclient dependency

    * Add secondary groups to process user during privilege escalation

    * When logging request headers, it is now possible to specify
      specifically which headers should be logged

    * Added log_requests config parameter to account and container servers
      to match the parameter in the object server. This allows a deployer
      to turn off log messages for these processes.

    * Ensure swift.source is set for DLO/SLO requests

    * Fixed an issue where overwriting segments in a dynamic manifest
      could cause issues on pipelined requests.

    * Properly handle COPY verb in container quota middleware

    * Improved StaticWeb 404 error message on web-listings and index

    * Various other minor bug fixes and improvements.

swift (1.12.0)

    * Several important pieces of information have been added to /info:

       - Configured constraints are included and allow a client to discover
         the limits on names and object sizes that the cluster supports.

       - The supported tempurl methods are now included.

       - Static large object constraints are now included.

    * The Last-Modified header value returned will now be the object's
      timestamp rounded up to the next second. This allows subsequent
      requests with If-[un]modified-Since to use the Last-Modified
      value as expected.

    * Non-integer values for if-delete-at headers will now properly
      report a 400 error instead of a 503.

    * Fix object versioning with non-ASCII container names.

    * Bulk delete with POST now works properly.

    * Generic means for persisting system metadata

      Swift now supports system-level metadata on accounts and
      containers. System metadata provides a means to store internal
      custom metadata with associated Swift resources in a safe and
      secure fashion without actually having to plumb custom metadata
      through the core swift servers. The new gatekeeper middleware
      prevents this system metadata from leaking into the request or
      being set by a client.

    * catch_errors and gatekeeper middleware are now forced into the proxy
      pipeline if not explicitly referenced.

    * New container sync configuration option, separating the end user
      from knowing the required end point and adding more secure
      signed requests. See
      http://swift.openstack.org/overview_container_sync.html for full
      information.

    * bulk middleware now can be configured to retry deleting containers.

    * The default yield_frequency used to keep client connections alive
      during slow bulk requests was reduced from 60 seconds to 10 seconds.
      While this is a change to a default, it should not affect deployments
      and there is no migration process needed.

    * Swift processes will attempt to set RLIMIT_NPROC to 8192.

    * Server processes will now exit with a non-zero error code on config
      errors.

    * Warn if read_affinity is configured but not enabled.

    * Fix checkmount error parsing in swift-recon.

    * Log at warn level when an object is quarantined.

    * Fixed CVE-2014-0006 to avoid a potential timing attack with tempurl.

    * Various other minor bug fixes and improvements.


swift (1.11.0)

    * Added discoverable capabilities

      A Swift proxy server now by default (although it can be turned off)
      will respond to requests to /info. The response to these requests
      include information about the cluster and can be used by clients to
      determine which features are supported in the cluster.

    * Object replication ssync (an rsync alternative)

      A Swift storage node can now be configured to use Swift primitives
      for replication transport instead of rsync. This is an experimental
      feature that is not yet considered production ready.

    * If a source times out on an object server read, try another one
      of them with a modified range.

    * The proxy now responds to many types of requests as soon as it
      has a quorum. This can help speed up responses (without
      changing the results), especially when one node is acting up.
      There is a post_quorum_timeout config value that can tune how
      long to wait for requests to finish after a quorum has been
      established.

    * Add accurate timestamps in proxy log lines for the start and
      end of a request. These are added as new fields on the end of
      the existing log lines, and therefore should not break
      existing, well-behaved log processors.

    * Add an "inline" query parameter to tempurl

      By default, temporary URLs add a "Content-Disposition" header
      that forces many clients to download the object. Now, temporary
      URLs support an optional "inline" query parameter that will
      force a "Content-Disposition: inline" header to be added to the
      response, overriding the default.

    * Use TCP_NODELAY for created sockets. This can dramatically
      lower latency for small object workloads.

    * DiskFile API, with reference implementation

      The DiskFile abstraction for talking to data on disk has been
      refactored to allow alternate implementations to be developed.
      Included in the codebase is an in-memory reference
      implementation. For full documentation, please see the developer
      documentation. The DiskFile API is still a work in progress and
      is not yet finalized.

    * Removal of swift-bench

      The included benchmarking tool swift-bench has been extracted
      from the codebase and is now in its own repository at
      https://github.com/openstack/swift-bench. New swift-bench
      binaries and packages may be found on PyPI at
      https://pypi.python.org/pypi/swift-bench

    * Bulk delete now also supports the POST verb, in addition to DELETE

    * Added functionality to the swift-ring-builder to support
      limited recreation of ring builder files from the ring file itself.

    * HEAD on account now returns 410 if account was deleted and
      not yet reaped. The old behavior was to return a 404.

    * Fixed a bug introduced since the 1.10.0 release that
      prevented expired objects from being removed from the system.
      This resulted in orphaned expired objects taking up space on
      the system but inaccessible to the API. This regression and
      fix are only important if you have deployed code since the
      1.10.0 release. For a full discussion, including a script that
      can be used to clean up orphaned objects, see
      https://bugs.launchpad.net/swift/+bug/1257330

    * Tie socket write buffer size to server chunk size parameter. This
      pairs the underlying network buffer size with the size of data
      that Swift attempts to read from the connection, thereby
      improving efficiency and throughput on connections.

    * Fix 500 from account-quota middleware. If a user had set
      X-Account-Meta-Quota-Bytes to something non-integer prior to
      the installation of the account-quota middleware, then the
      quota check would choke on it. Now a non-integer value is
      treated as "no quota".

    * Quarantine objects with busted metadata. Before, if you
      encountered an object with corrupt or missing xattrs, the
      object server would return a 500 on GET, and wouldn't quarantine
      anything. Now the object server returns a 404 for that GET and
      the corrupted file is quarantined, thus giving replication a
      chance to fix it.

    * Fix quarantine and error counts in audit logs

    * Report transaction ID in failure exception logs

    * Make pbr a build-time only dependency

    * Worked around a bug in eventlet 0.9.16 where the size of the
      memcache connection pools would grow unbounded.

    * Tempurl keys are now properly stored as utf8

    * Fixed an issue where concurrent PUT requests to accounts or
      containers may result in errors due to locked databases.

    * Handle copy requests in account and container quota middleware

    * Now ensure that a WWW-Authenticate header is on all 401 responses

    * Various other bug fixes and improvements

swift (1.10.0)

    * Added support for pooling memcache connections

    * Added support to replicating handoff partitions first in object
      replication. Can also configure how many remote nodes a storage node
      must talk to before removing a local handoff partition.

    * Fixed bug where memcache entries would not expire

    * Much faster calculation for choosing handoff nodes

    * Added container listing ratelimiting

    * Fixed issue where the proxy would continue to read from a storage
      server even after a client had disconnected

    * Added support for headers that are only visible to the owner of a Swift
      account

    * Fixed ranged GET with If-None-Match

    * Fixed an issue where rings may not be balanced after initial creation

    * Fixed internationalization support

    * Return the correct etag for a static large object on the PUT response

    * Allow users to extract archives to containers with ACLs set

    * Fix support for range requests against static large objects

    * Now logs x-copy-from header in a useful place

    * Reverted back to old XML output of account and container listings to
      ensure older clients do not break

    * Account quotas now appropriately handle copy requests

    * Fix issue with UTF-8 handling in versioned writes

    * Various other bug fixes and improvements, including support for running
      Swift under Pypy and continuing work to support storage policies

swift (1.9.1)

    * Disallow PUT, POST, and DELETE requests from creating older tombstone
      files, preventing the possibility of filling up the disk and removing
      unnecessary container updates.

    * Set default wsgi workers to cpu_count

      Change the default value of wsgi workers from 1 to auto. The new
      default value for workers in the proxy, container, account & object
      wsgi servers will spawn as many workers per process as you have cpu
      cores. This will not be ideal for some configurations, but it's much
      more likely to produce a successful out of the box deployment.

    * Added reveal_sensitive_prefix config setting to filter the auth token
      logged by the proxy server.

    * Ensure Keystone's reseller prefix ends with an underscore. Previously
      this was a recommendation--now it is enforced.

    * Added log_file_pattern config to swift-drive-audit for drive errors

    * Add support for telling Swift to detect a content type on a request.

    * Additional object stats are now logged in the object auditor

    * Moved the DiskFile interface into its own module

    * Ensure the SQLite cursors are closed when creating functions

    * Better support for valid Accept headers

    * In Keystone, don't allow users to delete their own account

    * Return a UTC timezone designator in container listings

    * Ensure that users can't remove their account quotas

    * Allow floating point value for dispersion coverage

    * Fix incorrect error page handling in staticweb

    * Add utf-8 charset to multipart-manifest=get response.

    * Allow dispersion tools to use keystone server with insecure certificate

    * Ensure that files are always closed in tests

    * Use OpenStack's "Hacking" guidelines for code formatting

    * Various other minor bug fixes and improvements

swift (1.9.0)

    * Global clusters support

      The "region" concept introduced in Swift 1.8.0 has been augmented with
      support for using a separate replication network and configuring read
      and write affinity. These features combine to offer support for a single
      Swift cluster spanning wide geographic area.

    * Disk performance

      The object server now can be configured to use threadpools to increase
      performance and smooth out latency throughout the system. Also, many
      disk operations were reordered to increase reliability and improve
      performance.

    * Added config file conf.d support

      Allow Swift daemons and servers to optionally accept a directory as the
      configuration parameter. This allows different parts of the config file
      to be managed separately, eg each middleware could use a separate file
      for its particular config settings.

    * Allow two TempURL keys per account

      By adding a second key, a user can safely rotate keys and prevent URLs
      already in use from becoming invalid. TempURL middlware has also been
      updated to allow a configuable set of allowed methods and to prevent a
      bugrelated to content-disposition names.

    * Added crossdomain.xml middleware. See
      http://docs.openstack.org/developer/swift/crossdomain.html for details

    * Added rsync bandwidth limit setting for object replicator

    * Transaction ID updated to include the time and an optional suffix

    * Added x-remove-versions-location header to disable versioned writes

    * Improvements to support for Keystone ACLs

    * Added parallelism to object expirer daemon

    * Added support for ring hash prefix in addition to the existing suffix

    * Allow all headers requested for CORS

    * Stop getting useless bytes on manifest Range requests

    * Improved container-sync resiliency

    * Added example Apache config files. See
      http://docs.openstack.org/developer/swift/apache_deployment_guide.html
      for more info

    * If an account is marked as deleted but hasn't been reaped and is still
      on disk, responses will include an "X-Account-Status" header

    * Fix 503 on account/container HEAD with invalid format

    * Added extra safety on account-level DELETE when using bulk deletes

    * Made colons quote-safe in logs (mainly for IPv6)

    * Fixed bug with bulk delete max items

    * Fixed static large object manifest range requests

    * Prevent static large objects from containing other static large objects

    * Fixed issue with use of delimiter in container queries where some
      objects would not be listed

    * Various other minor bug fixes and improvements

swift (1.8.0)

    * Make rings' replica count adjustable

    * Added a region tier to the ring above zones

    * Added timing-based sorting of object servers on read requests

    * Added support for auto-extract archive uploads

    * Added support for bulk delete requests

    * Added support for large objects with static manifests

    * Added list_endpoints middleware to provide an API for determining where
      the ring places data

    * proxy-logging middleware can now handle logging for other middleware

      proxy-logging should be used twice in the proxy pipeline. The first
      handles middleware logs for requests that never made it all the way
      to the server. The last handles requests that do make it to the server.

      This is a change that may require an update to your proxy server
      config file or custom middleware that you may be using. See the full
      docs at http://docs.openstack.org/developer/swift/misc.html#module-swift.common.middleware.proxy_logging.

    * Changed the default sample rate for a few high-traffic requests.

      Added log_statsd_sample_rate_factor to globally tune the StatsD
      sample rate. This tunable can be used to reduce StatsD traffic
      proportionally for all metrics and is intended to replace
      log_statsd_default_sample_rate, which is left alone for
      backward-compatibility, should anyone be using it.

    * Added swift_hash_path_prefix option to swift.conf

      New deployments are advised to set this value to a random secret
      to protect against hash collisions

    * Added user-managed container quotas

    * Added support for account-level quotas managed by an auth reseller

    * Added --run-dir option to swift-init

    * Added more options to swift-bench

    * Added support for CORS "actual requests"

    * Added fallocate_reserve option to protect against full drives

    * Allow ring rebalance to take a seed

    * Ring serialization will now produce the same gzip file (Py2.7)

    * Added support to swift-drive-audit for handling rotated logs

    * Added first-byte latency timings for GET requests

    * Added per disk PUT timing monitoring support

    * Added speed limit options for DB auditor

    * Force log entries to be one line

    * Ensure that fsync is used and not just fdatasync

    * Improved handoff node selection

    * Deprecated keystone is_admin feature

    * Fix large objects with unicode in the segment names

    * Update Swift's MemcacheRing to provide API compatibility with
      standard Python memcache libraries

    * Various other minor bug fixes and improvements

swift (1.7.6)

    * Better tempauth storage URL guessing

    * Added --top option to swift-recon -d

    * Allow optional, temporary healthcheck failure

    * keystoneauth middleware now supports cross-tenant ACLs

    * Add dispersion report flags to limit reports

    * Add config option to turn eventlet debug on/off

    * Added override option for swift-init's KILL_WAIT

    * Added oldest and most recent replication pass to swift-recon

    * Fixed 500 error response when GETing a many-segment manifest

    * Memcached keys now use a delta timeout when possible

    * Refactor DiskFile to hide temp file names and exts

    * Remove IP-based container-sync ACLs from auth middlewares

    * Fixed bug in deleting memcached account info data

    * Fixed lazy-listing of object manifest segments

    * Fixed bug where a ? in the object name caused an error

    * Swift now returns 406 if it can't satisfy Accept

    * Fix infinite recursion bug in object replicator

    * Swift will now reject names with NULL characters

    * Fixed object-auditor logging to use a minimum of unix sockets

    * Various other minor bug fixes and improvements

swift (1.7.5)

    * Support OPTIONS verb, including CORS preflight requests

    * Added support for custom log handlers

    * Range support is extended to support GET requests with multiple ranges.
      Multi-range GETs are not yet supported against large-object manifests.

    * Cluster constraints are now settable by config

    * Replicators can now run against specific devices or partitions

    * swift-bench now supports running on multiple cores and multiple servers

    * Added partition option to swift-get-nodes

    * Allow underscores in account and user in tempauth via base64 encodings

    * New option to the dispersion report to output the missing partitions

    * Changed storage server StatsD metrics to report timings instead of
      counts for errors. See the admin guide for the updated metric names.

    * Removed a dependency on WebOb and replaced it with an internal module

    * Fixed config parsing in swift-bench -x

    * Fixed sample_rate in StatsD logging

    * Track unlinks of async_pendings with StatsD

    * Remove double GET on range requests

    * Allow unsetting of X-Container-Sync-To and ACL headers

    * DB reclamation now removes empty suffix directories

    * Fix non-standard 100-continue behavior

    * Allow object-expirer to delete the last copy of a versioned object

    * Only set TCP_KEEPIDLE on systems where it is supported

    * Fix stdin flush and fdatasync issues on BSD platforms

    * Allow object-expirer to delete the last version of an object

    * Various other minor bug fixes and improvements

swift (1.7.4)

    * Fix issue where early client disconnects may have caused a memory leak

swift (1.7.2)

    * Fix issue where memcache serialization was not properly loading
      the config value

swift (1.7.0)

    * Use custom encoding for ring data instead of pickle

      Serialize RingData in a versioned, custom format which is a combination
      of a JSON-encoded header and .tostring() dumps of the
      replica2part2dev_id arrays.  This format deserializes hundreds of times
      faster than rings serialized with Python 2.7's pickle (a significant
      performance regression for ring loading between Python 2.6 and Python
      2.7).  Fixes bug 1031954.

      The new implementation is backward-compatible; if a ring
      does not begin with a new-style magic string, it is assumed to be an
      old-style pickle-dumped ring and is handled as before.  So new Swift
      code can read old rings, but old Swift code will not be able to read
      newly-serialized rings.

    * Do not use pickle for serialization in memcache, but JSON

      To avoid issues on upgrades (unability to read pickled values, and cache
      poisoning for old servers not understanding JSON), we add a
      memcache_serialization_support configuration option, with the following
      values:

       0 = older, insecure pickle serialization
       1 = json serialization but pickles can still be read (still insecure)
       2 = json serialization only (secure and the default)

      To avoid an instant full cache flush, existing installations should
      upgrade with 0, then set to 1 and reload, then after some time (24
      hours) set to 2 and reload. Support for 0 and 1 will be removed in
      future versions.

    * Update proxy-server StatsD logging. This is a significant change to the
      existing StatsD intigration. Docs for this feature can be found in
      doc/source/admin_guide.rst.

    * Improved swift-bench to allow random object sizes and better usability

    * Updated probe tests

    * Replicator removal metrics are now generated on a per-device basis

    * Made object replicator locking more optimistic

    * Split proxy-server code into separate modules

    * Fixed bug where swift-recon would not report all unmounted drives

    * Fixed issue where a LockTimeout may have caused a file descriptor to
      not be closed properly

    * Fixed a bug where an error may have caused the proxy to stop returning
      data to a client

    * Fixed bug where expirer would get confused by odd deletion times

    * Fixed a bug where auto-creating accounts would return an error if they
      were recreated after being deleted

    * Fix when rate_limit_after_segment kicks in

    * fallocate() failures properly return HTTPInsufficientStorage from
      object-server before reading from wsgi.input, allowing the proxy
      server to quickly error_limit that node

    * Fixed error with large object manifests and x-newest headers on GET

    * Various other minor bug fixes and improvements

swift (1.6.0)

    * Removed bin/swift and swift/common/client.py from the swift repo. These
      tools are now managed in the python-swiftclient project. The
      python-swiftclient project is a second deliverable of the openstack
      swift project.

    * Moved swift_auth (openstack keystone) middleware from keystone project
      into swift project

    * Made dispersion report work with any replica count other than 3. This
      substantially affects the JSON output of the dispersion report, and any
      tools written to consume this output will need to be updated.

    * Added Solaris (Illumos) compatibility

    * Added -a option to swift-get-nodes to show all handoffs

    * Add UDP protocol support for logger

    * Added config options for rate limiting of large object downloads.

    * Added config option `log_handoffs` (defaults to True) to proxy server
      to log and update statsd with information about when a handoff node is
      used. This is helpful to track the health of the cluster.

    * swift-bench can now use auth 2.0

    * Support forbidding substrings based on a regexp in name_filter
      middleware

    * Hardened internal server processes so only authorized methods can be
      called.

    * Made ranged requests on large objects work correctly when size of
      manifest file is not 0 byte

    * Added option to dispersion report to print 404s to stdout

    * Fix object replication on older rsync versions when using ipv4

    * Fixed bug with container reclaim/report race

    * Make object server's caching more configurable.

    * Check disk failure before syncing for each partition

    * Allow special characters to be referenced by manifest objects

    * Validate devices and partitions to avoid directory traversals

    * Support WebOb 1.2

    * Ensure that accessing the ring devs reloads the ring if necessary.
      Specifically, this allows replication to work when it has been started
      with an empty ring.

    * Various other minor bug fixes and improvements

swift (1.5.0)

    * New option to toggle SQLite database preallocation with account
      and container servers.

      IMPORTANT:
      The default for database preallocation is now off when before
      it was always on. This will affect performance on clusters that
      use standard drives with shared account, container, object
      servers. Such deployments will need to update their
      configurations to turn database preallocation back on (see
      account-server.conf-sample and container-server.conf.sample
      files).

      If you are using dedicated account and container servers with
      SSDs, you should defragment your file systems after upgrade and
      should notice dramatically less disk usage.

    * swift3 middleware removed and moved to http://github.com/fujita/swift3.
      This will require a config change in the proxy server and adds a new
      dependency for deployers using this middleware.

    * Moved proxy server logging to middleware. This requires a config change
      in the proxy server.

    * Added object versioning feature. (See docs for full description)

    * Add statsd logging throughout the system (beta, some event names may
      change)

    * Expanded swift-recon middleware support

    * The ring builder now supports as-unique-as-possible partition
      placement, unified balancing methods, and can work on more than one
      device at a time.

    * Numerous bug fixes to StaticWeb (previously unusable at scale).

    * Bug fixes to all middleware to allow passthrough requests under various
      conditions and to share pre-authed request code (which previously had
      differing behaviors and interaction bugs).

    * Bug fix to object expirer that could cause infinite looping.

    * Added optional delay to account reaping.

    * Async-pending write optimization.

    * Dispersion tools now support multiple auth versions

    * Updated man pages

    * Proxy server can now deny requests to particular hostnames

    * Updated docs for domain remap middleware

    * Updated docs for cname lookup middleware

    * Made swift CLI binary easier to wrap

    * Proxy will now also return X-Timestamp header

    * Added associated projects doc as a place to track ecosystem projects

    * end_marker made consistent across both object and container listings

    * Various other minor bug fixes and improvements

swift (1.4.8)

    * Added optional max_containers_per_account restriction

    * Added alternate metadata header removal method

    * Added optional name_check middleware filter

    * Added support for venv-based test runs with tox

    * StaticWeb behavior change with X-Web-Mode: true and
      non-StaticWeb-enabled containers (immediately 404s instead of passing
      the request on down the WSGI pipeline).

    * Fixed typo in swift-dispersion-report JSON output.

    * Swift-Recon-related fix to create temporary files on the same disk as
      their final destinations.

    * Updated return codes in swift3 middleware

    * Fixed swift3 middleware to allow Content-Range header in response

    * Updated swift.common.client and swift CLI tool with auth 2.0 changes

    * Swift CLI tool now supports common openstack auth args

    * Body of HTTP responses now included in error messages of swift CLI tool

    * Refactored some ring building functions for clarity and simplicity

swift (1.4.7)

    * Improvements to account and container replication.

    * Fix for account servers allowing .pending to exist before .db.

    * Fixed possible key-guessing exploit in formpost.

    * Fixed bug in ring builder when removing a large percentage of devices.

    * Swift CLI tool now supports openstack-standard CLI flags.

    * New JSON output option for swift-dispersion-report.

    * Removed old stats tools.

    * Other bug fixes and documentation updates.

swift (1.4.6)

    * TempURL and FormPost middleware added

    * Added memcache.conf option

    * Dropped eval-based json parser fallback

    * Properly lose all groups when dropping privileges

    * Fix permissions when creating files

    * Fixed bug regarding negative Content-Length in requests

    * Consistent formatting on Last-Modified response header

    * Added timeout option to swift-recon

    * Allow arguments to be passed to nosetest

    * Removed tools/rfc.sh

    * Other minor bug fixes

swift (1.4.5)

    * New swift-orphans and swift-oldies command line tools to detect
      orphaned Swift processes and long running processes.

    * Command line tool "swift" now supports marker queries.

    * StaticWeb middleware improved to save an extra request when
      possible.

    * Updated swift-init to support swift-object-expirer.

    * Fixed object replicator timeout handling [bug 814263].

    * Fixed accept header 503 vs. 400 [bug 891247].

    * More exception handling for auditors.

    * Doc updates for PPA [bug 905608].

    * Doc updates to explain replication more clearly [bug 906976].

    * Updated SAIO instructions to no longer mention ~/swift/trunk.

    * Fixed docstrings in the ring code.

    * PEP8 Updates.

swift (1.4.4)

    * Fixes to prevent socket hoarding (memory leak)

    * Add sockstat info to recon.

    * Fixed leak from SegmentedIterable.

    * Fixed bufferedhttp to deref socks and fps.

    * Add support for OS Auth API version 2.

    * Make Eventlet's WSGI server log differently.

    * Updated TimeoutError and except Exception refs.

    * Fixed time-sensitive tests.

    * Fixed object manifest etags.

    * Fixes for swift-recon disk usage distribution graph.

    * Adding new manpages for configuration files.

    * Change bzr to swift in getting_started doc.

    * Fixes the HTTPConflict import.

    * Expiring Objects Support.

    * Fixing bug with x-trans-id.

    * Requote the source when doing a COPY.

    * Add documentation for Swift Recon.

    * Make drive audit regexes detect 4-letter drives.

    * Adding what acc/cont/obj into the ratelimit error messages.

    * Query only specific zone via swift-recon.

swift (1.4.3)

    * Additional quarantine catching code.

    * Added client_ip to all proxy log lines not otherwise containing it.

    * Content-Type is now application/xml for "GET services/bucket" swift3
      middleware requests.

    * Alpha release of the Swift Recon Experiment

    * Fix last modified date for swift3 middleware.

    * Fix to clear account/container metadata on account/container deletion.

    * Fix for corner case regarding X-Newest.

    * Fix for object auditor running out of file descriptors.

    * Fix to return all proper headers for manifest objects.

    * Fix to the swift tool to strip any leading slashes on file names when
      uploading.

swift (1.4.2)

    * Removed stats/logging code from Swift [now in separate slogging project].

    * Container Synchronization Feature - First Edition

    * Fix swift3 authentication bug about the Date and X-Amz-Date handling.

    * Changing ratelimiting so that it only limits PUTs/DELETEs.

    * Object POSTs are implemented as COPYs now by default (you can revert to
      previous implementation with conf object_post_as_copy = false)

    * You can specify X-Newest: true on GETs and HEADs to indicate you want
      Swift to query all backend copies and return the newest version
      retrieved.

    * Object COPY requests now always copy the newest object they can find.

    * Account and container GETs and HEADs now shuffle the nodes they use to
      balance load.

    * Fixed the infinite charset: utf-8 bug

    * This fixes the bug that drop_buffer_cache() doesn't work on systems where
      off_t isn't 64 bits.

swift (1.4.1)

    * st renamed to swift

    * swauth was separated froms swift. It is now its own project and can be
      found at https://github.com/gholt/swauth.

    * tempauth middleware added as an extremely limited auth system for dev
      work.

    * Account and container listings now properly labeled UTF-8 (previously the
      label was "utf8").

    * Accounts are auto-created if an auth token is valid when the
      account_autocreate proxy config parameter is set to true.

swift (1.4.0)

    * swift-bench now cleans up containers it creates.

    * WSGI servers now load WSGI filters and applications after forking for
      better plugin support.

    * swauth-cleanup-tokens now handles 404s on token containers and tokens
      better.

    * Proxy logs the remote IP address as the client IP in the absence of
      X-Forwarded-For and X-Cluster-Client-IP headers instead of - like it did
      before.

    * Swift3 WSGI middleware added support for param-signed URLs.

    * swauth- scripts now exit with proper exit codes.

    * Fixed a bug where allowed_headers weren't honored for HEAD requests.

    * Double quarantining of corrupted sqlite3 databases now works.

    * Fix for Object replicator breaking when running object replicator with no
      objects on the server.

    * Added the Accept-Ranges header to GET and HEAD requests.

    * When a single object has multiple async pending updates on a single
      device, only latest async pending is now sent.

    * Fixed issue of Swift3 WSGI middleware not working correctly with '/' in
      object names.

    * Renamed swift-stats-* to swift-dispersion-* to avoid confusion with log
      stats stuff.

    * Added X-Trans-Id transaction id header to every response.

    * Fixed a Python 2.7 compatibility problem.

    * Now using bracketed notation for ip literals in rsync calls, so
      compressed ipv6 literals work.

    * Added a container stats collector and refactoring some of the stats code.

    * Changed subdir nodes in XML formatted object listings to align with
      object nodes. Now: <subdir name="foo"><name>foo</name></subdir> Before:
      <subdir name="foo" />.

    * Fixed bug in Swauth to support for multiple swauth instances.

    * swift-ring-builder: Added list_parts command which shows common
      partitions for a given list of devices.

    * Object auditor now shows better statistics updates in the logs.

    * Stats uploaders now allow overrides for source_filename_pattern and
      new_log_cutoff values.