Items with no specified field now defaults to all the possible fields from the
document source. Previously, we had required 'fields' to be specified either
as a top level parameter or for each item. The default behavior is now similar
to the MLT API.
Closes#7382
This commit changes the way how files are selected for retransmission
on recovery / restore. Today this happens on a per-file basis where the
rather weak checksum and the file length in bytes is compared to check if
a file is identical. This is prone to fail in the case of a checksum collision
which can happen under certain circumstances.
The changes in this commit move the identity comparsion to a per-commit / per-segment
level where files are only treated as identical iff all the other files in the
commit / segment are the same. This "all or nothing" strategy is reducing the chance for
a collision dramatically since we also use a strong hash to identify commits / segments
based on the content of the ".si" / "segments.N" file.
Closes#7351
resetting the clients on each test (in after test) makes the tests running, especially in network mode, much slower, since transport client needs to be created each time when randmized to be used. Also, on OSX, the excessive connections causes bind exceptions eventually which makes running the network tests much harder on it.
closes#7329
Today we always use the latest commit point to return the metadata from
the store. This might cause problems for snapshot and restore since in
contrast to recovery it won't prevent concurrent flushes (lucene commits).
This can lead to all kinds of interesting effects if we are snapshotting
while flushing. This change uses the IndexCommit to open the metadata snapshot
from the store which is consistent with what we snapshot.
Closes#7376
Changes the serialisation of pre and post offset to use Long instead of VLong so that negative values are supported. This actually only showed up in the case where minDocCount=0 as the rounding is only serialised in this case.
Closes#7312
The term vector API can now generate term vectors on the fly, if the terms are
not already stored in the index. This commit exploits this new functionality
for the MLT query. Now the terms are directly retrieved using multi-
termvectors API, instead of generating them from the texts retrieved using the
multi-get API.
Closes#7014
Today we have very confusing naming since some methods names claim to
read binary but in fact read utf-16 and convert to utf-8 under the hood.
This commit clarifies the interfaces and adds additional documentation.
Closes#7367
These options are not used anymore. Instead numeric values can now contain dups
and it is the responsibility of the aggregation to deal with it (eg. terms).
And otherwise all values sources are now sorted, which is the contract of the
interfaces that they implement.
Close#7276
A request that relates to indices (`IndicesRequest` or `CompositeIndicesRequest`) might be converted to some other internal request(s) (e.g. shard level request) that get distributed over the cluster. Those requests contain the concrete index they refer to, but it is not known which indices (or aliases or expressions) the original request related to.
This commit makes sure that the original indices are available as part of the shard level requests and makes them implement `IndicesRequest` as well.
Also every internal request should be created passing in the original request, so that the original headers, together with the eventual original indices and options, get copied to it. Corrected some places where this information was lost.
NOTE: As for the bulk api and other multi items api (e.g. multi_get), their shard level requests won't keep around the whole set of original indices, but only the ones that related to the bulk items sent to each shard, the important bit is that we keep the original names though, not only the concrete ones.
Closes#7319
This change fixes the creation circle shapes o it calculates it correctly instead of essentially using the diameter as the radius. The radius has to be converted into degrees but calculating the ratio of the desired radius to the circumference of the earth and then multiplying it by 360 (number of degrees around the earths circumference). This issue here was that it was only multiplied by 180 making the result out by a factor of 2. Also made the test for circles actually check to make sure it has the correct centre and radius.
Closes#7301
Recent test failures triggered by #7289 were caused by this, simply because internal node settings (transport type key) that are not supported by the external older nodes were copied to them by mistake.
* Removed & refactored unused module code
* Allowed to set transports programmatically
* Allow to set the source of the changed transport
Note: The current implementation breaks BWC as you need to specify a concrete
transport now instead of a module if you want to use a different
Transport or HttpServerTransport
Closes#7289
TransportShardSingleOperationAction is currently subclassed by different transport actions. Some of them are internal only, meaning that their execution will take place only in the same node where their parent execution took place. That means that their main transport handler doesn't need to be registered, the only transport handler that's needed is the shard level one.
Added `isSubAction` method (defaults to false) to the parent class that tells whether the action is a main one or a subaction, used to decide whether we need to register the main transport handler.
Closes#7285
These two aggregators basically do exactly the same thing, they just interpret
bytes differently. This refactoring found an (unreleased) bug in the long terms
aggregator which didn't work correctly with duplicate values.
Close#7279
This was causing too much work e.g. when pulling node stats or when
opening a new reader, because the least_used distributor would
unnecessarily check free disk space on all path.data entires every
time we try to open a file for reading or check its length.
Closes#7306Closes#7323
This issue has been fixed in commons-cli:1.3 project which sadly has not been released yet.
See https://issues.apache.org/jira/browse/CLI-183
This patch builds another list of options with no selected groups by default.
When commons-cli:1.3 will be released, we need to remove this patch.
Closes#7282.
Switch management threads to a fixed thread pool with up to 5 threads, and queue size of 100 by default, after which excess incoming requests are rejected.
Closes#7318Closes#7320
The score is explained already, it should not be again explained per function.
Also, remove explanation from parameter list of ScoreFunction#explainScore()
and leave only the score.
This also removes ExplainableSearchScript which is not used anywhere and
was the only reason to have the Explanation in the parameter anyway.
closes#7245
The terms and histogram aggregations always have an order. So it would make the
response easier to consume to return the buckets as a list instead of a
collection in order to make it easier to do things like getting the first/last
buckets.
Close#7275
This change means that the default settings for expand_wildcards are only applied if the expand_wildcards parameter is not specified rather than being set upfront. It also adds the none and all options to the parameter to allow the user to specify no expansion and expansion to all indexes (equivalent to 'open,closed')
Closes#7258
Added @Nullable to:
- IndicesService.indexService
- IndexService.shard
- IndexService.shardInjector
This change doesn't try to do anything smart but just makes sure that a
*MissingException is thrown instead of a NullPointerException when the requested
object doesn't exist.
Close#7251
Also replaced int,String pair with ShardId that holds the same info and serializes it the same way.
Replaced shardId and index getters in BroadcastOperationRequest with a single ShardId getter.
Closes#7255
The geohash grid it 8 cells wide and 4 cells tall. GeoHashUtils.neighbor(String,int,int.int) set the limit of the number of cells in y to < 3 rather than <= 3 resulting in it either not finding all neighbours or incorrectly searching for a neighbour in a different parent cell.
Closes#7226
This change stores the index creation time in the index metadata when an index is created. The creation time cannot be changed but can be set as part of the create index request to allow for correct creation times for historical data.
Closes#7119
In order to have the possibility of debugging on the command line, the user
now can either set the es.cli.debug system property
which results in stack traces being written to to the terminal.
Closes#7222
A range filter on a date field with a numeric `from`/`to` value is **not** cached by default:
DELETE /test
PUT /test/t/1
{
"date": "2014-01-01"
}
GET /_validate/query?explain
{
"query": {
"filtered": {
"filter": {
"range": {
"date": {
"from": 0
}
}
}
}
}
}
Returns:
"explanation": "ConstantScore(no_cache(date:[0 TO *]))"
This patch fixes as well not caching `from`/`to` when using `now` value not rounded.
Previously, a query like:
GET /_validate/query?explain
{
"query": {
"filtered": {
"filter": {
"range": {
"date": {
"from": "now"
"to": "now/d+1"
}
}
}
}
}
}
was cached.
Also, this patch does not cache anymore `now` even if the user asked for caching it.
As it won't be cached at all by definition.
Added as well tests for all possible combinations.
Closes#7114.
When installing a bin only plugin, it is identified as a site plugin.
A current workaround would be to create in the zip file another empty dir. So if you have:
* `bin/myfile.sh`
* `empty/empty.txt`
the `bin` content will be extracted as expected.
Closes#7152.
indexRandom will try to delete bogus documents multiple times since they get tracked by indexOrAlias/id, and after the actual deletion any other attempt throws error and fails the test
An anti-pattern that we have in our code, noticeable for java API users, is that we modify incoming requests by replacing the index or alias with the concrete index. This way not only the request has changed, but all following communications that use that request will lose the information on whether the original request was performed against an alias or an index.
Refactored the following base classes: `TransportShardReplicationOperationAction`, `TransportShardSingleOperationAction`, `TransportSingleCustomOperationAction`, `TransportInstanceSingleOperationAction` and all subclasses by introduced an InternalRequest object that contains the original request plus additional info (e.g. the concrete index). This internal request doesn't get sent over the transport but rebuilt on each node on demand (not different to what currently happens anyway, as concrete index gets set on each node). When the request becomes a shard level request, instead of using the only int shardId we serialize the ShardId that contains both concrete index name (which might then differ ffrom the original one within the request) and shard id.
Using this pattern we can move get, multi_get, explain, analyze, term_vector, multi_term_vector, index, delete, update, bulk to not replace the index name with the concrete one within the request. The index name within the original request will stay the same.
Made it also clearer within the different transport actions when the index needs to be resolved and when that's not needed (e.g. shard level request), by exposing `resolveIndex` method. Moved check block methods to parent classes as their content was always the same on every subclass.
Improved existing tests by randomly introducing the use of an alias, and verifying that the responses always contain the concrete index name and not the original one, as that's the expected behaviour.
Added backwards compatibility tests to make sure that the change is applied in a backwards compatible manner.
Closes#7223
If a geo_shape had edges which either ran vertically along the dateline or touched the date line but did not cross it they would fail to parse. This is because the code which splits a polygon along the dateline did not take into account the case where the polygon touched but did not cross the dateline. This PR fixes those issues and provides tests for them.
Close#7016
The `.percolator` type is a hidden type and therefor the types from the delete mapping request should passed down to the delete by query request, otherwise the percolator type gets ignored and the percolator queries don't get deleted from disk (only unregistered).
Closes#7087
internal cluster communication.
See CorruptedCompressorTests for details on how this bug can be hit.
This change also removes the ability to use the unsafe variant of
ChunkedEncoder, removing support for the compress.lzf.decoder setting.
If a dynamic mapping for a geo_point field is defined and the first document specifies the value of the field as a geo_point array, the dynamic mapping throws an error as the array is broken into individual number before consulting the dynamic mapping configuration. This change adds a check of the dynamic mapping before the array is split into individual numbers.
Closes#6939
This test assumed that the `num` field was mapped as an integer on all shards
and thus that all of them should fail when providing a timezone. However, since
it used dynamic mappings, some shards might have this field not mapped, as a
consequence they didn't fail.
TransportSingleCustomOperationAction is subclassed by two similar, yet different transport action: TransportAnalyzeAction and TransportGetFieldMappingsAction. Made their difference and similarities more explicit by sharing common code and moving specific code to subclasses:
- moved index field to the parent SingleCustomOperationAction class
- moved the common check blocks code to the parent transport action class
- moved the main transport handler to the TransportAnalyzeAction subclass as it is only used to receive external requests through clients. In the case of the TransportGetFieldMappingsIndexAction instead, the action is internal and executed only locally as part of the user facing TransportGetFieldMappingsAction. The corresponding request gets sent over the transport though as part of the related shard request
- removed the get field mappings index action from the action names mapping as it is not a transport handler anymore. It was before although never used.
Closes#7214
TransportIndexReplicationAction is always executed locally, as an internal action that is part of either delete by query or delete (when routing is required but not specified). Only the corresponding shard level requests get sent over the transport, hence no transport endpoint is needed for the index version, nor the index request itself is supposed to be sent over the transport.
Moved classes from org.elasticsearch.action.delete.index to org.elasticsearch.action.delete and adjusted visibility so that internal requests are not public anymore.
Also removed serialization code from IndexDeleteResponse as it never gets sent over transport either.
Closes#7211
Index, type and id were returned as part of the REST explain api response, but not through java api. That info was read out of the request, relying on the fact that the index would get overridden with the concrete one within that same request.
Closes#7201
Until now, IP addresses were only checked for four dots, which
allowed invalid values like 127.0.0.111111
This adds an additional check for validation.
Closes#7131
- The context enables setting arbitrary transient data on the message (this data is not serialized with the request)
- Changed header accessors/mutators so header manipulation will be done directly on the request (to void NPE with transport message headers when dealing with maps that can potentially be null)
It also didn't follow the setter convention that we adopted for request builders.
Fixed also javadocs warning caused byt missing descriptions for tag.
Closes#7186
The second internal node, when present, wasn't able to join the existing cluster due ti misconfigured unicast hosts, thus it would form its own cluster.
we need that in order for refresh to be effective and actually refresh in the second round of indexing, otherwise, it caches a 0 docs shard and a refresh won't expire anything there
A request level flag, defaults to be unset, to control the query cache. When not set, it defaults to the index level settings, when explicitly set, will override the index level setting
closes#7167
In the case of inserts the UpdateHelper class will now allow the script used to apply updates to run on the upsert doc provided by clients. This allows the logic for managing the internal state of the data item to be managed by the script and is not reliant on clients performing the initialisation of data structures managed by the script.
Closes#7143
The query cache allow to cache the (binary serialized) response of the shard level query phase execution based on the actual request as the key. The cache is fully coherent with the semantics of NRT, with a refresh (that actually ended up refreshing) causing previous cached entries on the relevant shard to be invalidated and eventually evicted.
This change enables query caching as an opt in index level setting, called `index.cache.query.enable` and defaults to `false`. The setting can be changed dynamically on an index. The cache is only enabled for search requests with search_type count.
The indices query cache is a node level query cache. The `indices.cache.query.size` controls what is the size (bytes wise) the cache will take, and defaults to `1%` of the heap. Note, this cache is very effective with small values in it already. There is also the advanced option to set `indices.cache.query.expire` that allow to control after a certain time of inaccessibility the cache will be evicted.
Note, the request takes the search "body" as is (bytes), and uses it as the key. This means same JSON but with different key order will constitute different cache entries.
This change includes basic stats (shard level, index/indices level, and node level) for the query cache, showing how much is used and eviction rates.
While this is a good first step, and the goal is to get it in, there are a few things that would be great additions to this work, but they can be done as additional pull requests:
- More stats, specifically cache hit and cache miss, per shard.
- Request level flag, defaults to "not set" (inheriting what the setting is).
- Allowing to change the cache size using the cluster update settings API
- Consider enabling the cache to query phase also when asking hits are involved, note, this will only include the "top docs", not the actual hits.
- See if there is a performant manner to solve the "out of order" of keys in the JSON case.
- Maybe introduce a filter element, that is outside of the request, that is checked, and if it matches all docs in a shard, will not be used as part of the key. This will help with time based indices and moving windows for shards that fall "inside" the window to be more effective caching wise.
- Add a more infra level support in search context that allows for any element to mark the search as non deterministic (on top of the support for "now"), and use it to not cache search responses.
closes#7161
This adds support to return the "Access-Control-Allow-Credentials" header
if needed, so CORS will work flawlessly with authenticated applications.
Closes#6380
It's now possible to define the additional customesettings for transport clients by extending `transportClientSettings` callback method on `ElasticsearchIntegrationTest`.
Filters and Queries now supports `time_zone` parameter which defines which time zone should be applied to the query or filter to convert it to UTC time based value.
When applied on `date` fields the `range` filter and queries accept also a `time_zone` parameter.
The `time_zone` parameter will be applied to your input lower and upper bounds and will move them to UTC time based date:
[source,js]
--------------------------------------------------
{
"constant_score": {
"filter": {
"range" : {
"born" : {
"gte": "2012-01-01",
"lte": "now",
"time_zone": "+1:00"
}
}
}
}
}
{
"range" : {
"born" : {
"gte": "2012-01-01",
"lte": "now",
"time_zone": "+1:00"
}
}
}
--------------------------------------------------
In the above examples, `gte` will be actually moved to `2011-12-31T23:00:00` UTC date.
NOTE: if you give a date with a timezone explicitly defined and use the `time_zone` parameter, `time_zone` will be
ignored. For example, setting `from` to `2012-01-01T00:00:00+01:00` with `"time_zone":"+10:00"` will still use `+01:00` time zone.
Closes#3729.
Our transport relies on action names that tell what we need to do with each message received and sent on any node, together with the content of the request itself.
The action names could use a better categorization and more consistent naming though, the following are the categories introduced with this commit:
- indices: for all the apis that execute against indices
- admin: for the apis that allow to perform administration tasks against indices
- data: for the apis that are about data
- read: apis that read data
- write: apis that write data
- benchmark: apis that run benchmarks
- cluster: for all the cluster apis
- admin: for the cluster apis that allow to perform administration tasks
- monitor: for the cluster apis that allow to monitor the system
- internal: for all the internal actions that are used from node to node but not directly exposed to users
The change is applied in a backwards compatible manner: we keep the mapping old-to-new action name around, and when receiving a message, depending on the version of the node we receive it from, we use the received action name or we convert it to the previous version (old to new if version < 1.4). When sending a message, depending on the version of the node we talk to, we use the updated action or we convert it to the previous version (new to old if version < 1.4).
For the cases where we don't know the version of the node we talk to, namely unicast ping, transport client nodes info and transport client sniff mode (which calls cluster state), we just use a lower bound for the version, thus we will always use the old action name, which can be understood by both old nodes and new nodes.
Added test that enforces known updated categories for transport action names and test that verifies all action names have a pre 1.4 version for bw compatibility
Added backwards compatibility tests for unicast and transport client in sniff mode, the one for the ordinary transport client (which calls nodes info) is implicit as it's used all the time in our bw comp tests.
Added also backwards comp test that sends an empty message to any of the registered transport handler exposed by older nodes and verifies that what gets back is not ActionNotFoundTransportException, which would mean that there is a problem in the actions mappings.
Added TestCluster#getClusterName abstract method and allow to retrieve externalTransportAddress and internalCluster from CompositeTestCluster.
Closes#7105
If simultaneous create & delete operations arrive against the same id,
it's possible that primary and replica see those operations in
different orders, which may result in replica throwing
DocumentAlreadyExistsException when the primary didn't which would
lead to replica being inconsistent (missing a document that primary
had indexed).
This push fixes the issue, by never throwing DAEE from the replica on
create.
Closes#7146#7142
Made sure that the routing required check is performed against the concrete index, added use of aliases to existing routing tests.
Taken the change to unify the failure message as well to this form: routing is required for [" + index + "]/[" + type + "]/[" + id + "]
Closes#7145
Before the index reader used by the percolator didn't allow to register a CoreCloseListener, but now it does, making it safe to cache index field data cache entries.
Creating field data structures is relatively expensive and caching them can save a lot of noise if many queries are evaluated in a percolator call.
Closes#6806Closes#7081
Fields of type `token_count`, `murmur3`, `_all` and `_field_names` are generated only when indexing.
If a GET requests accesses the transaction log (because no refresh
between indexing and GET request) then these fields cannot be retrieved at all.
Before the behavior was so:
`_all, _field_names`: The field was siletly ignored
`murmur3, token_count`: `NumberFormatException` because GET tried to parse the values from the source.
In addition, if these fields were not stored, the same behavior occured if the fields were
retrieved with GET after a `refresh()` because here also the source was used to get the fields.
Now, GET accepts a parameter `ignore_errors_on_generated_fields` which has
the following effect:
- Throw exception with meaningful error message explaining the problem if set to false (default)
- Ignore the field if set to true
- Always ignore the field if it was not set to stored
This changes the behavior for `_all` and `_field_names` as now an Exception is thrown if a user
tries to GET them before a `refresh()`.
closes#6676closes#6973
Allow to set the value default to network.tcp.no_delay and network.tcp.keep_alive so they won't be set at all, since on solaris, setting tcpNoDelay can actually cause failure
relates to #7115
CliTool is a base class for command-line interface tools (such as the plugin manager and potentially others). It supports the following:
- single or multi command tool
- help printing infrastructure (based on help files)
- consistent mechanism of parsing arguments (based on commons-cli lib)
- separation of argument parsing and command execution (for easier unit testing)
- terminal abstraction (will use System.console() when available)
A multi-bucket aggregation where multiple filters can be defined (each filter defines a bucket). The buckets will collect all the documents that match their associated filter.
This aggregation can be very useful when one wants to compare analytics between different criterias. It can also be accomplished using multiple definitions of the single filter aggregation, but here, the user will only need to define the sub-aggregations only once.
Closes#6118
Now that we have explicit support for aliases when creating indices and as part of index templates, we may remove support for aliases (only names) as part of index settings. This is partially breaking as the following calls:
curl -XPUT localhost:9200/index -d '{
"settings" : {
"aliases" : [ "alias1"]
}
}
and
curl -XPUT localhost:9200/index -d '{
"settings" : {
"index.aliases" : [ "alias1"]
}
}
were previously supported and will need to be replaced with
curl -XPUT localhost:9200/index -d '{
"aliases" : {
"alias1": {}
}
}
Closes#5545
The histogram reduce method can run into an infinite loop if the
Rounding.nextRoundingValue value is buggy, which happened to be the case for
DayTimeZoneRoundingFloor.
DayTimeZoneRoundingFloor is fixed, and the histogram reduce method has been
changed to fail instead of running into an infinite loop in case of a buffy
nextRoundingValue impl.
Close#6965
Today, `copy_to` always copies a field to the current document, which is often
wrong in the case of nested documents. For example, if you have a nested field
called `n` which has a sub-field `n.source` whose content should be copied to
`target`, then the latter field should be created in the root document instead
of the nested one, since it doesn't have `n.` as a prefix. On the contrary, if
you configure the destination field to be `n.target`, then it should go to the
nested document.
Close#6701
Implements a new Exists API allowing users to do fast exists check on any matched documents for a given query.
This API should be faster then using the Count API as it will:
- early terminate the search execution once any document is found to exist
- return the response as soon as the first shard reports matched documents
closes#6995
Index process fails when having `_timestamp` enabled and `path` option is set.
It fails with a `TimestampParsingException[failed to parse timestamp [null]]` message.
Reproduction:
```
DELETE test
PUT test
{
"mappings": {
"test": {
"_timestamp" : {
"enabled" : "yes",
"path" : "post_date"
}
}
}
}
PUT test/test/1
{
"foo": "bar"
}
```
You can define a default value for when timestamp is not provided
within the index request or in the `_source` document.
By default, the default value is `now` which means the date the document was processed by the indexing chain.
You can disable that default value by setting `default` to `null`. It means that `timestamp` is mandatory:
```
{
"tweet" : {
"_timestamp" : {
"enabled" : true,
"default" : null
}
}
}
```
If you don't provide any timestamp value, indexation will fail.
You can also set the default value to any date respecting timestamp format:
```
{
"tweet" : {
"_timestamp" : {
"enabled" : true,
"format" : "YYYY-MM-dd",
"default" : "1970-01-01"
}
}
}
```
If you don't provide any timestamp value, indexation will fail.
Closes#4718.
Closes#7036.
when there is a cluster block (like no master yet discovered), the bulk action doesn't properly catch the exception of inner execute to notify the listener, causing the bulk operation to hang
closes#7086
The used -p option could result in accidentally deleting more directories
than /var/lib/elasticsearch - so this option was removed
Note: This only happens if the directories are empty, but still isnt needed.
Relates #5770
This commit effectively reverts e1aa91d , as it is not needed anymore to add the original listed nodes. The cluster state local call made will in fact always return at least the local node (see #6811).
There were a couple of downsides caused by putting the original listed nodes among the connected nodes:
1) in the following retries, they weren't seen as listed nodes anymore, thus the light connect wasn't used
2) among the connected nodes some were "bad" duplicates as they are already there and don't contain all needed info for each node. This was causing serialization problems for instance given that the node version was missing on the `DiscoveryNode` object.
Closes#7067
The RetryListener was notified twice for each single failure, which caused some additional retries, but more importantly was making the client reach the maximum number of retries (number of connected nodes) too quickly, meanwhile ongoing retries which could succeed were not completed yet.
The TransportService used to throw ConnectTransportException due to throwConnectException set to true, and also notify the listener of any exception received from a separate thread through the request holder.
Simplified exception handling by just removing the throwConnectException option from the TransportService, used only in the transport client. The transport client now relies solely on the request holder to notify of failures and eventually retry.
Closes#6829
Adds additional version checks in NodeStats for older versions
When using an external cluster (backwards compatibility tests), the act
of checking the request breaker requires a network buffer, which
increments the breaker. This change only checks the request breaker in
InternalTestCluster and uses Guice to retrieve it instead of
a (possible) network request.
Also removed the now unused InternalCircuitBreakerService class
Also improved filter chain tests to not rely on execution time, and made filter chain tests look more similar to what happens in reality by removing multiple threads creation in testTooManyContinueProcessing (something we don't support anyway, makes little sense to test it).
Closes#7021
With commit 07c632a2d4dbefe44e8f25dc4ded6cf143d60e41, we now have a new Lucene.parseVersionLenient(String, Version) method which tries to find an existing Lucene version based on the two first digits X.Y of X.Y.Z String.
Pull Request #7055 fixed Version parsing for bugfix releases
causing problems with minor version in segments files. Even though
we never release anything with lucene in alpha / beta status this
commit fixes lenient parsing for these cases.
Relates to #7055
We parse the version that is shipped with the Lucene segments in order
to find the version of lucene that wrote a particular segment. Yet, some lucene
version ie:
* 4.3.1 (Elasticsearch 0.90.2)
* 4.5.1 (Elasticsearch 0.90.7)
* 3.6.1 (pre Elasticsearch 0.90.0)
wrote illegal strings containing the minor version which causes IAE exceptions
being thrown from lucenes parsing method.
Closes#7055
Adds a breaker for request BigArrays, which are used for parent/child
queries as well as some aggregations. Certain operations like Netty HTTP
responses and transport responses increment the breaker, but will not
trip.
This also changes the output of the nodes' stats endpoint to show the
parent breaker as well as the fielddata and request breakers.
There are a number of new settings for breakers now:
`indices.breaker.total.limit`: starting limit for all memory-use breaker,
defaults to 70%
`indices.breaker.fielddata.limit`: starting limit for fielddata breaker,
defaults to 60%
`indices.breaker.fielddata.overhead`: overhead for fielddata breaker
estimations, defaults to 1.03
(the fielddata breaker settings also use the backwards-compatible
setting `indices.fielddata.breaker.limit` and
`indices.fielddata.breaker.overhead`)
`indices.breaker.request.limit`: starting limit for request breaker,
defaults to 40%
`indices.breaker.request.overhead`: request breaker estimation overhead,
defaults to 1.0
The breaker service infrastructure is now generic and opens the path to
adding additional circuit breakers in the future.
Fixes#6129
Conflicts:
src/main/java/org/elasticsearch/index/fielddata/IndexFieldData.java
src/main/java/org/elasticsearch/index/fielddata/IndexFieldDataService.java
src/main/java/org/elasticsearch/index/fielddata/RamAccountingTermsEnum.java
src/main/java/org/elasticsearch/index/fielddata/ordinals/GlobalOrdinalsBuilder.java
src/main/java/org/elasticsearch/index/fielddata/ordinals/InternalGlobalOrdinalsBuilder.java
src/main/java/org/elasticsearch/index/fielddata/plain/AbstractIndexOrdinalsFieldData.java
src/main/java/org/elasticsearch/index/fielddata/plain/DisabledIndexFieldData.java
src/main/java/org/elasticsearch/index/fielddata/plain/IndexIndexFieldData.java
src/main/java/org/elasticsearch/index/fielddata/plain/NonEstimatingEstimator.java
src/main/java/org/elasticsearch/index/fielddata/plain/PackedArrayIndexFieldData.java
src/main/java/org/elasticsearch/index/fielddata/plain/ParentChildIndexFieldData.java
src/main/java/org/elasticsearch/index/fielddata/plain/SortedSetDVOrdinalsIndexFieldData.java
src/main/java/org/elasticsearch/node/internal/InternalNode.java
src/test/java/org/elasticsearch/index/aliases/IndexAliasesServiceTests.java
src/test/java/org/elasticsearch/index/codec/CodecTests.java
src/test/java/org/elasticsearch/index/fielddata/AbstractFieldDataTests.java
src/test/java/org/elasticsearch/index/fielddata/IndexFieldDataServiceTests.java
src/test/java/org/elasticsearch/index/mapper/MapperTestUtils.java
src/test/java/org/elasticsearch/index/query/IndexQueryParserFilterCachingTests.java
src/test/java/org/elasticsearch/index/query/SimpleIndexQueryParserTests.java
src/test/java/org/elasticsearch/index/query/guice/IndexQueryParserModuleTests.java
src/test/java/org/elasticsearch/index/search/FieldDataTermsFilterTests.java
src/test/java/org/elasticsearch/index/search/child/ChildrenConstantScoreQueryTests.java
src/test/java/org/elasticsearch/index/similarity/SimilarityTests.java
No that we are using the index created version to make index-time decisions,
assuming that the version is the current version when settings are null is
very error-prone. Instead we should ensure that settings are always non-null
and contain the version when the index was created.
Close#7032
In context of mapper attachment and other mapper plugins, when dealing with multi fields, sub fields never get the `externalValue` although it was set.
Here is a full script which reproduce the issue when used with mapper attachment plugin:
```
DELETE /test
PUT /test
{
"mappings": {
"test": {
"properties": {
"f": {
"type": "attachment",
"fields": {
"f": {
"analyzer": "english",
"fields": {
"no_stemming": {
"type": "string",
"store": "yes",
"analyzer": "standard"
}
}
}
}
}
}
}
}
}
PUT /test/test/1
{
"f": "VGhlIHF1aWNrIGJyb3duIGZveGVz"
}
GET /test/_search
{
"query": {
"match": {
"f": "quick"
}
}
}
GET /test/_search
{
"query": {
"match": {
"f.no_stemming": "quick"
}
}
}
GET /test/test/1?fields=f.no_stemming
```
Related to https://github.com/elasticsearch/elasticsearch-mapper-attachments/issues/57Closes#5402.
This is only applicable when the order is set to _count. The upper bound of the error in the doc count is calculated by summing the doc count of the last term on each shard which did not return the term. The implementation calculates the error by summing the doc count for the last term on each shard for which the term IS returned and then subtracts this value from the sum of the doc counts for the last term from ALL shards.
Closes#6696
This commit adds regular expression support for the allow-origin
header depending on the value of the request `Origin` header.
The existing HttpRequestBuilder is also extended to support the
OPTIONS HTTP method.
Relates #5601Closes#6891
The HTTP client implementation used by the Elasticsearch REST tests is
backed by apache http client instead of a self written helper class,
that uses HttpUrlConnection. This commit removes the old simple HttpClient
class and uses the more powerful and reliable one for all tests.
It also fixes a minor bug, that when sending a 301 redirect, a Location
header needs to be added as well, which was uncovered by the switching
to the new client.
Closes#7003
The bug reproduces when the point under test for the placement of the hole of the polygon has an x coordinate which only intersects with the ends of edges in the main polygon. The previous code threw out these cases as not relevant but an intersect at 1.0 of the distance from the start to the end of an edge is just as valid as an intersect at any other point along the edge. The fix corrects this and adds a test.
Closes#5773
This commit adds the ability to force blocking on the flush operaition
to make sure all files have been written and synced to disk. Without
this option a flush might be executing at the same time causing the
current flush to fail and return before all files being synced.
Closes#6996
Added two new interfaces:
1) IndicesRequest that allows to retrieve the indices the request relates to in a generic manner, together with the indices options that tell how they are going to get resolved and expanded
2) CompositeIndicesRequest for compound requests that hold multiple indices request like MultiSearchRequest, MultiGetRequest, MultiTermVectorsRequest, BulkRequest, BenchmarkRequest, PercolateRequest, MultiPercolateRequest and MoreLikeThisRequest
Taken the chance to streamline the indices options and add them to every request where it makes sense (although they can't be changed from the outside), rather than leaving them implicit in the related TransportAction when indices get expanded (tipycally MetaData#concreteIndices or MetaData#concreteSingleIndex). Added IndicesOptions parameter to MetaData#concreteSingleIndex to make sure it is taken from the request, where the information belongs, instead of hardcoded within MetaData. The concreteSingleIndex method remains but it's just a utility method that returns a single index instead of an array and complains otherwise.
Also made sure NPE is never thrown when setting indices(null) to IndicesAliasesRequest, similar to what SearchRequest does.
Closes#6933
The benchmark indexes 200 unique full-width longs. For uninverted field data
we try to use the most memory-efficient storage, and in that case it would use
two arrays: one for the doc->ordinals mapping and one for the ordinal->value
mapping. Which is slower than what doc values do by storing directly the
mapping from docs to values.
Before this change each aggregation had to output an object field with its name and write its JSON inside that object. This allowed for badly behaved aggregations which could write JSON content in the root of the 'aggs' object. this change move the writing of the aggregation name to a level above the aggregation itself, ensuring that aggregations can only write within there own scope in the JSON output.
Closes#7004
Today if the user supplies a custom missing value for a string sort,
we do it in an extremely slow way, not using ordinals but dereferencing
bytes for every document. Ordinals are only used if the missing value
is _first or _last.
Instead, use ordinals with custom missing values too.
Closes#7005
Single index operations to use the newly added IndexClosedException introduced with #6475. This way we can also fail faster when we are trying to execute operations on closed indices and their use is not allowed (depending on indices options). Indices blocks are still checked but we can already throw error while resolving indices (MetaData#concreteIndices).
Effectively this change also affects what we return when using one of the following apis: analyze, bulk, index, update, delete, explain, get, multi_get, mlt, term vector, multi_term vector. We now return `{"error":"IndexClosedException[[test] closed]","status":403}` instead of `{"error":"ClusterBlockException[blocked by: [FORBIDDEN/4/index closed];]","status":403}`.
Closes#6988
The default shard size in the terms aggregation now uses BucketUtils.suggestShardSideQueueSize() to set the shard size if the user does not specify it as a parameter.
Closes#6857
Allow users to control document collection termination, if a specified terminate_after number is
set. Upon setting the newly added parameter, the response will include a boolean terminated_early
flag, indicating if the document collection for any shard terminated early.
closes#6876
instead of a custom encoding in BINARY.
In low level benchmarks this is 2x to 5x faster: its also optimized
for the common case where fields actually only contain at most one
value for each document.
Additionally SORTED_NUMERIC doesn't lose values if they appear more
than once, so mathematical computations such as averages are correct.
Closes#6967
The recycling happening in facets is done manually and arrays are sometimes not
released. Aggregations do it in a less error-prone way by registering on to the
SearchContext.
This commit removes custom comparators in favor of the ones that are in Lucene.
The major change is for nested documents: instead of having a comparator wrapper
that deals with nested documents, this is done at the fielddata level by having
a selector that returns the value to use for comparison.
Sorting with custom missing string values might be slower since it is using
TermValComparator since Lucene's TermOrdValComparator only supports sorting
missing values first or last. But other than this particular case, this change
will allow us to benefit from improvements on comparators from the Lucene side.
Close#5980
This change just changes the default for index.codec.bloom.load to
false: with recent performance improvements to ID lookup, such as
#6298, bloom filters don't give much of a performance gain anymore,
and they can consume non-trivial RAM when there are many tiny
documents.
For now, we still index the bloom filters, so if a given app wants
them back, it can just update the index.codec.bloom.load to true.
Closes#6959
GET only returned null even when stored if requested with GET like this:
`curl -XGET "http://localhost:9200/test/test/1?fields=_all"`
Instead, it should simply behave like a String field and return the
concatenated fields as String.
closes#6924
A new option `prune` has been added to allow users to control phrase suggestion pruning when `collate`
is set. If the new option is set, the phrase suggestion option will contain a boolean `collate_match`
indicating whether the respective result had hits in collation.
CLoses#6927
Today we only do count searches to ensure sane results are returned
after upgrading etc. This change adds sorting to the picture asserting
on simple numeric sorting that uses field data etc. after upgrading.
Relates to #6967
In unicast discovery, we try to reuse existing discovery nodes based on the node address they have. If we find an existing node based on its address, and for some reason its not connected, don't add it to the list of nodes to disconnect from, as that (full) connection is useful down the road
closes#6966
Looking at the connect code, if 2 threads at the same time try and connect to a node, and both enter sequentially the connectLock code block, the second one would try and put the connection in the map, and close the replaced channels, which will cause the existing connection to close as well (since it removes the node from the connectedNodes map)
To fix this, simply make sure we properly check the existence of the connection within the connectionLock block, so there won't be concurrent connections going on.
While doing this, also went over all the mutation code that handles disconnections, and made sure they are properly done only within a connection lock.
closes#6964
This commits removes BytesValues/LongValues/DoubleValues/... and tries to use
Lucene's APIs such as NumericDocValues or RandomAccessOrds instead whenever
possible.
The next step would be to take advantage of the fact that APIs are the same in
Lucene and Elasticsearch in order to remove our custom comparators and use
Lucene's.
There are a few side-effects to this change:
- GeoDistanceComparator has been removed, DoubleValuesComparator is used instead
on top of dynamically computed values (was easier than migrating
GeoDistanceComparator).
- SortedNumericDocValues doesn't guarantee uniqueness so long/double terms
aggregators have been updated to make sure a document cannot fall twice in
the same bucket.
- Sorting by maximum value of a field or running a `max` aggregation is
potentially significantly faster thanks to the random-access API.
Our aggs and p/c aggregations benchmarks don't report differences with this
change on uninverted field data. However the fact that doc values don't need
to be wrapped anymore seems to help a lot. For example
TermsAggregationSearchBenchmark reports ~30% faster terms aggregations on doc
values on string fields with this change, which are now only ~18% slower than
uninverted field data although stored on disk.
Close#6908
The assertion on binary equality for streamable serialization
sometimes fails due to the usage of identify hashmaps inside
the InternalSearchHits serialization. This only happens if
the number of shards the result set is composed of is very high.
This commit makes the serialziation deterministic and removes
the need to serialize the ordinal due to in-order serialization.
The change introduced in #6949 (do not serialize the cluster state) also means master now responds with an empty response rather then a JoinResponse. However, sendJoinRequestBlocking still expected a JoinRequest.
At the moment we serialize the cluster state in JoinResponse and ValidateJoinRequest. However this state is not used anywhere and can be removed to save on network overhead
Closes#6949
This test deletes and updates using upserts documents over several threads in a
tight loop. It counts the number of responses and verifies that the versions at
the end are correct.
today if a snapshot is corrupted the restore operation
never terminates. Yet, if the snapshot is corrupted there
is no way to restore it anyway. If such a snapshot is restored
today the only way to cancle it is to delete the entire index which
might cause dataloss. This commit also fixes an issue in InternalEngine
where a deadlock can occur if a corruption is detected during flush
since the InternalEngine#snapshotIndex aqcuires a topLevel read lock
which prevents closing the engine.
Closes#6938
The `index.fail_on_corruption` was not updateable via the index settings
API. This commit also fixed the setting prefix to be consistent with other
setting on the engine. Yet, this feature is unreleased so this won't break anything.
Closes#6941
Small refactorings to make the MessageChannelHandler more extensible.
Also allowed access to the different netty pipelines
This is the fix after the first version had problems with the HTTP
transport due to wrong reusing channel handlers, which is the reason
why tests failed.
Relates #6889Closes#6915
It's now possible to inject action filters from plugins via `ActionModule#registerFilter` through the following code:
```
public void onModule(ActionModule actionModule) {
actionModule.registerFilter(MyFilter.class);
}
```
Also made `TransportAction#execute` methods final to enforce the execution of the filter chain. By default the chain is empty though.
Note that the action filter chain is executed right after the request validation, as the filters might rely on a valid request to do their work.
Closes#6921
Today when we start a `TransportClient` we use the given transport
addresses and create a `DiscoveryNode` from it without knowing the
actual nodes version. We just use the `Version.CURRENT` which is an
upper bound. Yet, the other node might be a version less than the
currently running and serialisation of the nodes info might break. We
should rather use a lower bound here which is the version of the first
release with the same major version as `Version.CURRENT` since this is
what we officially support.
This commit moves to use the minimum major version or an RC / Snapshot
if the current version is a snapshot.
Closes#6894
Adds the ability to the Term Vector API to generate term vectors for some
chosen fields, even though they haven't been explicitely stored in the index.
Relates to #5184Closes#6567
The bulk processor tries to acquire all leases for the semaphore to wait
for all pending requests. Yet, we should release them afterwards again to
ensure we don't ever deadlock if there is a bug in the processor.
This commit also adds a testcase for this method
It's now possible to register watchers along with a specified check frequency. There are three frequencies: low, medium, high. Each one is associated with a check interval that determines how frequent the watchers will check for changes and notify listeners if needed. By default, the intervals are 5s, 30s and 60s respectively, but they can also be customized in the settings. also:
- Added the WatcherHandle construct by which one can stop it (remove it) and resume it (re add it). Also provices access to the watchers itself and the frequency by which it's checked
- Change the default frequency to 30 seconds interval (used to be 60 seconds). The only watcher that is currently effected by this is the script watcher (now auto-loading scripts will auto-load every 30 seconds if changed)
This is to prevent a rare racing condition where the very same shard gets allocated to the node after our sanity check that the cluster state didn't check and the actual deletion of the files.
Closes#6902
This results in unstable tests, most likely due to Channels being mixed
up by wrongly creating the pipelines. Needs investigation and a test.
This reverts commit db7f0d36af.
The IndexingMemoryController determines the amount of indexing buffer size and translog buffer size each shard should have. It takes memory from inactive shards (indexing wise) and assigns it to other shards. To do so it needs to know about the addition and closing of shards. The current implementation hooks into the indicesService.indicesLifecycle() mechanism to receive call backs, such shard entered the POST_RECOVERY state. Those call backs are typically run on the thread that actually made the change. A mutex was used to synchronize those callbacks with IndexingMemoryController's background thread, which updates the internal engines memory usage on a regular interval. This introduced a dependency between those threads and the locks of the internal engines hosted on the node. In a *very* rare situation (two tests runs locally) this can cause recovery time outs where two nodes are recovering replicas from each other.
This commit introduces a a lock free approach that updates the internal data structures during iterations in the background thread.
Closes#6892
FieldMapper has two methods
`Filter termsFilter(List values, @Nullable QueryParseContext)` which is supposed
to work on the inverted index and
`Filter termsFilter(QueryParseContext, List, QueryParseContext)` which is
supposed to work on field data. Let's rename the second one to
`fieldDataTermsFilter` and remove the unused `QueryParseContext`.
Close#6888
If you call `bin/plugin --remove es-plugin` the plugin got removed but the file `bin/plugin` itself was also deleted.
We now don't allow the following plugin names:
* elasticsearch
* plugin
* elasticsearch.bat
* plugin.bat
* elasticsearch.in.sh
* service.bat
Closes#6745
With the removal of setNextScore in #6864, script engines must use
the Scorer to find the score of a document. The DocLookup is updated
appropriately to do this, but most script engines require a Number to be
bound for numeric variables. Groovy already had an encapsulation for
this funtionality, and this moves it out to be shared with other script
engines.
closes#6898
Simon Willnauer
Alex Ksikes
Adrien Grand
javanna
Martijn van Groningen
Shay Banon
Igor Motov
Britta Weber
Colin Goodheart-Smithe
uboness
Alexander Reelsen
mikemccand
David Pilato
Robert Muir
Ryan Ernst
Lee Hinman
Thomas Peuss
olivier bourgain
Areek Zillur
markharwood
Skye Book
Brian Murphy
Itamar Syn-Hershko
Clinton Gormley
Nik Everett
Boaz Leskes
matt-preston