-------------------------
Riak 0.? Release Notes
-------------------------
Bugs Fixed
----------
0520 - vnode throws error during handoff
-------------------------
Riak 0.14.2 Release Notes
-------------------------
Riak 0.14.2 is a patch-level release for the 0.14 series and as such restricted
to bug fixes.
Bugs Fixed
----------
0977 - riak_kv_vnode:syntactic_put_merge creates different merged timestamps if
clock skewed
1006 - Objects not found in map-reduce are no longer retried on other vnodes in preflist
1009 - Empty js files in js_source_dir caused crash on start
1015 - Queued javascript map phases are processed even after a map reduce job
has failed (e.g. client receives timeout)
1090 - bitcask_nifs_keydir_itr() has a race condition wrt. thread creation
1094 - Read repair introduces 'undefined' timestamps
1095 - result non-determinstic with allow_mult=false and siblings have same timestamp
1104 - MapReduce queries return results from other MapReduce queries
1112 - Map phase retries on not_found causes "500 Internal Server Error" response when
input contains the same bucket/key pair multiple times
1113 - Map phase retries on not_found causes "500 Internal Server Error" response when
nodes are down
1114 - Map phase retries on not_found causes "500 Internal Server Error" response
(function_clause) when nodes are down
All bug and issue numbers reference https://issues.basho.com.
-------------------------
Riak 0.14.1 Release Notes
-------------------------
Riak 0.14.1 is a patch-level release for the 0.14 series and as such restricted
to bug fixes.
Bugs Fixed
----------
0969 - Map/reduce cache fails to prune
0971 - Latency of "/stats" request steadily increases over time
0982 - riak-admin js_reload does not work
0987 - cluster_info formatting needs truncation limits
1000 - Ensure Riak and Riak EE use latest riak-erlang-client library
All bug and issue numbers reference https://issues.basho.com.
-------------------------
Riak 0.14.0 Release Notes
-------------------------
Riak 0.14.0, aka "Dakota," is officially tagged and ready for downloading. We
made a lot of strides in both feature robustness and overall stability.
Here are the highlights:
1) Map/Reduce performance has been further improved in three areas. The first,
Key Filters, will enable you to build metadata into your keys so that you can
filter a set of inputs for a given criterion before being processed by the
Map/Reduce job. This reduces the number of keys passed to the Map/Reduce job and
improves performance for many queries.
The second major enhancement is a rework of the Map/Reduce Query Planner. The
new planner takes a more sophisticated approach to analyzing and scheduling
bucket/key pairs around the cluster. Using batches of 50 bucket/key pairs, the
new code schedules bucket/key pairs onto common vnodes in a single message, thus
reducing chattiness and improving overall throughput.
Finally, 0.14 supports three different JavaScript VM pools to reduce overall
contention. Users can tailor the size of each pool to their particular needs via
configuration values that regulate the VMs allocated to map functions, reduce
functions, and pre-commit hooks (the three major points of JavaScript
invocation).
2) Cluster and node debugging facilities have been added in the form of
cluster_info and riak_err. The former gathers environment, configuration, and
runtime statistics and writes them to a single file for admin or developer
review. The latter, riak_err, limits the amount of RAM that is used while
processing event log messages. There is some new documentation on the wiki that
further describes how to use the cluster_info feature. See the bottom of
http://wiki.basho.com/Command-Line-Tools.html
3) Windowed Merges in Bitcask - Bitcask performs periodic merges over all
non-active files to compact the space being occupied by old versions of stored
data. In certain situations this can cause some memory and CPU spikes on the
Riak node where the merge is taking place. To that end, we've added the ability
to specify when Bitcask will perform merges. This can be set here:
https://github.com/basho/bitcask/blob/master/ebin/bitcask.app#L38
4) Other noteworthy enhancements include:
* Support for HTTPS and multiple HTTP IPs (thanks Benjamin Black)
* Packaging scripts for building debs, rpms and Solaris packages
* The ability to list buckets through the REST API.
Here's the full list of enhancements and bug fixes:
Enhancements
------------
047 - Riak should provide a configuration URL
078 - List buckets from REST API
571 - Provide "riak-admin remove node@host" command to remove nodes from a cluster
743 - Allow multiple mochiweb processes
744 - Allow binding multiple HTTP/HTTPS IPs and ports
745 - Allow binding multiple HTTP/HTTPS IPs and ports
765 - riak-erlang-client unit test fails due to updated error message
777 - Bundle redbug into all release packages
781 - OTP default error_logger handler isn't super-awesome
785 - bitcask_merge_worker leaks ports if the merge crashes
856 - Calls to unexported ets:next/1 in riak_kv_js_manager.erl
857 - Call to unexported riak_kv_keylister_master:start_keylist/6 in lk.erl
871 - Link walking busted in refactored MapReduce branch
873 - riak_err UTC time formatting is broken
881 - Add riak_err app to all Riak* product packaging
882 - Add cluster_info app to all Riak* packaging
884 - If-Match and If-None-Match should support a list of ETag values
893 - riak_err mis-handles error_report with badmatch in riak_err_handler:format_event/2
902 - Bitcask should only perform merges during acceptable timeframes
916 - Dialyzer errors for pre-Dakota release (cluster_info)
932 - Fix "riak-admin transfers" regression: "No transfers active\n" message missing
950 - riak_err sends progress reports to console
Bugs Fixed
----------
107 - ensure correct ETag format
210 - Cannot setup multiple backends using riak_kv_bitcask_backend
508 - Updating riak_object values can lose metadata
537 - riakc_pb_socket function calls are all subject to the default 5 second
gen_server timeout
787 - fail gracefully in bitcask:fold if last entry is truncated
804 - Riak KV handoff sender badmatch when using Innostore backend
813 - riak-admin js_reload results in error
827 - Webmachine only reads the first chunk of an incoming chunked request
835 - Bitcask worker_ready state not updated correctly on merge
852 - Peer address should be pulled from X-Forwarded-For for all RFC1918 space
862 - riak-admin js_reload exits with {badrecord,vm_state}
877 - List keys causes erl_drv_rwlock_runlock() error on Solaris when using
the Bitcask backend
879 - Etag If-Match bug in H7
888 - Bitcask errors if the writer process is restarted
891 - Using persistent HTTP connections seems to increase the latency per request
892 - Summary plots fail to generate if ""0"" duration results are present
901 - riak-erlang-client server_closes_socket_test unit tests fails
905 - New HTTP/HTTPS binding config is not backwards compatible
913 - Reference to non-existent gmt_config_svr
933 - Map phases break for non-existent keys
937 - riak_client:filter_keys exits with bad_match
938 - riak_kv_js_manager times out reserving JS VM
All bug and issue numbers reference https://issues.basho.com.
