Skip to content

HTTPS clone URL

Subversion checkout URL

You can clone with
or
.
Download ZIP
Commits on Jun 24, 2014
  1. @borshop

    Merge pull request #405 from basho/rz/clear-tree-stats

    borshop authored
    When clearing a tree clear its stats too
    
    Reviewed-by: coderoshi
  2. @borshop

    Merge pull request #404 from basho/er/ensemble-delete

    borshop authored
    Add support for ensemble tombstone value
    
    Reviewed-by: andrewjstone
Commits on Jun 23, 2014
  1. When clearing a tree clear its stats too

    authored
    When a tree is cleared (erased from disk) the corresponding build
    status should also be reset.  That is, the `riak-admin search
    aae-status` command should display `--` for the last time the tree was
    built since it is _not_ built.
Commits on Jun 20, 2014
  1. @coderoshi
Commits on Jun 4, 2014
  1. @borshop

    Merge pull request #396 from basho/bugfix/387

    borshop authored
    Correctly set Jetty temp directory
    
    Reviewed-by: coderoshi
  2. @borshop

    Merge pull request #397 from basho/er/list-working-indexes-only

    borshop authored
    Filter indexes that do not exist on Solr
    
    Reviewed-by: rzezeski
  3. Correctly set Jetty temp directory

    authored
    A custom `solr-jetty-context.xml` must be used so that the
    `tempDirectory` is set properly.  `jetty.home` is a system property,
    not a regular property.
Commits on Jun 3, 2014
  1. @borshop

    Merge pull request #400 from basho/rz/remove-comments

    borshop authored
    Remove obsolete comment
    
    Reviewed-by: coderoshi
  2. @coderoshi
  3. Remove obsolete comment

    authored
    This comment is no longer valid.  The reason AAE was being triggered
    after node restarts was because Riak Test was hard killing (kill -9)
    the cluster causing corruption and thus causing AAE to go nuts.
  4. @borshop

    Merge pull request #398 from basho/bug/rdb/gh-kv967-embedded-cntr

    borshop authored
    Use the embedded counter inside a Map
    
    Reviewed-by: seancribbs
  5. @russelldb

    Use the embedded counter inside a Map

    russelldb authored
    Use the embedded mod_map inside the dt extractor
Commits on May 30, 2014
  1. @borshop

    Merge pull request #394 from basho/bugfix/383

    borshop authored
    Change `/search/Index` to `/search/query/Index`
    
    Reviewed-by: coderoshi
  2. @borshop

    Merge pull request #393 from basho/bugfix/211

    borshop authored
    Verify _yz fields are not multivalued
    
    Reviewed-by: coderoshi
Commits on May 28, 2014
  1. Change `/search/Index` to `/search/query/Index`

    authored
    The routes for querying and index administration are ambiguous if
    there is an index named `index`.  The user will not be able to query
    because Yokozuna will interpret `GET /search/index?q=...` as a request
    to the index admin resource.  This commit changes the query resource
    to `/search/query/Index` so there is no ambiguity.
Commits on May 27, 2014
  1. Verify _yz fields are not multivalued

    authored
    The _yz fields should never be multivalued.  Prevent users from
    storing schemas that declare them as such.
Commits on May 23, 2014
  1. @borshop

    Merge pull request #388 from basho/feature/er/hide-update-handler

    borshop authored
    Hide solr path under internal_solr
    
    Reviewed-by: rzezeski
  2. @borshop

    Merge pull request #390 from basho/bugfix/update-pb-msg

    borshop authored
    Update invalid_name msg in PB index create
    
    Reviewed-by: coderoshi
  3. @borshop

    Merge pull request #386 from basho/bugfix/356

    borshop authored
    Fix sort param in `search_fold` to guarantee unique values
    
    Reviewed-by: coderoshi
Commits on May 22, 2014
  1. @coderoshi
  2. @borshop

    Merge pull request #368 from basho/rz/bb-json-obj

    borshop authored
    Add JSON Object Value Generator
    
    Reviewed-by: lordnull
  3. Update invalid_name msg in PB index create

    authored
    Recently there was a change to only allow ASCII chars in the index name.
    This updates the PB interface error message when an index name is
    invalid.  The HTTP resource was already updated in the original commit
    bb44381.
Commits on May 21, 2014
  1. @borshop

    Merge pull request #381 from basho/jrd-ascii-indexes

    borshop authored
    Constrain index names to ASCII to duck latin-1/UTF-8 conflicts
    
    Reviewed-by: coderoshi
Commits on May 20, 2014
  1. Add JSON Object Value Generator

    authored
    Add `json_obj_valgen` to the Yokozuna BB driver. I used this to test
    indexing "large" JSON objects while investigating #358.  The config
    line to use this value generator looks like so:
    
    ```
    {value_generator, {function, yz_driver, json_obj_valgen,
    [<NumFields>]}}.
    ```
    
    This will generate a static JSON object (i.e. generated once at start
    up) which contains `<NumFields>` integer fields.  All workers will
    index the identical object.  Here is an example config from one of my
    testing runs:
    
    ```
    {mode, max}.
    {concurrent, 32}.
    {driver, yz_driver}.
    {code_paths, ["/root/work/yz-bb/misc/bench"]}.
    {secure, false}.
    {bucket, {<<"data">>, <<"largeobj">>}}.
    {index, <<"largeobj">>}.
    {pb_conns, [{"10.0.1.201", 8087}]}.
    {http_conns, []}.
    {duration, infinity}.
    {key_generator, {to_binstr, "~B", {partitioned_sequential_int, 0,
    1000}}}.
    {value_generator, {function, yz_driver, json_obj_valgen, [4000]}}.
    {operations, [{load_pb, 1}]}.
    ```
    
    This uses 32 workers to write the same JSON object containing 4000
    integer fields over the key space 0 to 999.
  2. Fix sort param in `search_fold` to guarantee unique values

    authored
    The sort parameter must be type/bucket/key, as described in issue
    duplicate values.  Other modifications include:
    
    * Make sure to pass the new accumulator after calling `F` in
      `search_fold/9`.
    
    * Increase the page size to 100 so less HTTP requests are made.  This
      should result in lower latency.
    
    * Modify `yz_mapreduce` to use 1000 keys so since the page size was
      increased to 100.
    
    * Add a test case to `yz_mapreduce` to verify that all 1000 unique
      results are returned.
    
    * Add `yz_rt:load_module/2` to be used for loading a module's BEAM
      onto a set of nodes.  This is needed by the new test to have the HOF
      passed to `search_fold` available on the server process.
  3. @borshop

    Merge pull request #385 from basho/bugfix/376

    borshop authored
    Fix deletion for many-to-one indexing
    
    Reviewed-by: coderoshi
Commits on May 19, 2014
  1. @borshop

    Merge pull request #375 from basho/rz/time-wait-load

    borshop authored
    Reduce TIME-WAIT load
    
    Reviewed-by: coderoshi
  2. Fix deletion for many-to-one indexing

    authored
    It is possible to point multiple types and buckets to the same index.
    The object key alone is not a unique identifier as the same key could
    be repeated across types and buckets.
    
    This patch fixes the deletion code to delete on the actual unique key
    which is type + bucket + key.  Other changes made include:
    
    * Expand the essential test to verify deletion and sibling cleanup
      work correctly when using the same key across multiple types and
      buckets.  Add unicode and special chars to additional verification.
    
    * Update the AAE test to to pass the full key when making remote call
      to `yz_solr:delete`.
    
    * Use the ?DEFAULT_TYPE macro in case the default type is changed
      before 2.0.0.
    
    * Change `yz_rt:random_keys` to return binary keys.
    
    * Handle legacy objects (pre 2.0.0) in `encode_delete`.
    
    * Make sure to pass the lager transform when compiling the Yokozuna
      driver and make the `build_bb_driver` function more robust in
      general.
Commits on May 15, 2014
  1. @macintux
  2. @borshop

    Merge pull request #351 from basho/refactor/normalize-riak-test-execu…

    borshop authored
    …tion
    
    Normalize riak_test execution process
    
    Reviewed-by: coderoshi
Commits on May 14, 2014
  1. @macintux
  2. @kellymclaughlin
  3. @kellymclaughlin
  4. @kellymclaughlin
  5. @kellymclaughlin

    Test updates to avoid nondeterminism in riak_test execution

    kellymclaughlin authored
    Changes to avoid some problems with nondeterminism with execution
    of the yokozuna riak_test modules.
Something went wrong with that request. Please try again.