Commit Graph

4456 Commits

Author SHA1 Message Date
Zachary Tong 20d67720aa
Refactor Percentiles/Ranks aggregation builders and factories (#51887) (#54537)
- Consolidates HDR/TDigest factories into a single factory
- Consolidates most HDR/TDigest builder into an abstract builder
- Deprecates method(), compression(), numSigFig() in favor of a new
unified PercentileConfig object
- Disallows setting algo options that don't apply to current algo

The unified config method carries both the method and algo-specific
setting. This provides a mechanism to reject settings that apply
to the wrong algorithm.  For BWC the old methods are retained
but marked as deprecated, and can be removed in future versions.

Co-authored-by: Mark Tozzi <mark.tozzi@gmail.com>

Co-authored-by: Mark Tozzi <mark.tozzi@gmail.com>
2020-04-02 10:39:41 -04:00
Nik Everett a5adac0d1e
Fix pipeline agg serialization for ccs (backport of #54282) (#54468)
This fixes pipeline aggregations used in cross cluster search from an older
version of Elasticsearch to a newer version of Elasticsearch. I broke
this in #53730 when I was too aggressive in shutting off serialization
of pipeline aggs. In particular, this comes up when the coordinating
node is pre-7.8.0 and the gateway node is on or after 7.8.0.

The fix is another step down the line to remove pipeline aggregators
from the aggregation tree. Sort of. It create a new
`List<PipelineAggregator>` member in `InternalAggregation` *but* it is
only used for bwc serialization and it is fed by the mechanism
established in #53730 to read the pipelines from the
2020-04-02 10:35:40 -04:00
Nik Everett b4feda84e8
Add scroll info to search task description (backport of #54606) (#54612)
Right now you can't tell from the task description whether or not the
search is a scroll. This adds that information to the description which
is super useful if you are trying to debug a cluster that is running out
of scroll contexts.
2020-04-02 09:04:49 -04:00
Jason Tedor 18b602280c
Add validation to the usage service (#54617)
Today the usage service can let in some issues, such as handlers that do
not have a name, where the errors do not manifest until later (calling
the usage API), or conflicting handlers with the same name. This commit
addresses this by adding some validation to the usage service.
2020-04-02 08:56:28 -04:00
Andy Bristol eb14635f1f
add tests to StatsAggregatorTests (#53768)
Adds tests for supported ValuesSourceTypes, unmapped fields, scripting,
and the missing param. The tests for unmapped fields and scripting are
migrated from the StatsIT integration test
2020-04-01 17:07:51 -07:00
Andy Bristol c87b830d06
migrate tests from MissingIT to agg tests (#53448)
Move the remaining tests for the missing aggregation into its
AggregatorTestCase out of its integration test and remove the IT
2020-04-01 17:05:44 -07:00
Andy Bristol ec76e7306e
supported field type tests for max agg (#53701)
Adds test hooks for testing supported ValuesSource types for the max
aggregation
2020-04-01 15:24:53 -07:00
Andy Bristol 5d0351ea00
add tests to SumAggregatorTests (#53568)
This adds tests for supported ValuesSourceTypes, unmapped fields,
scripting, and the missing param. The tests for unmapped fields and
scripting are migrated from the SumIT integration test
2020-04-01 15:24:21 -07:00
Andy Bristol 62a52465fc
aggregator and yaml tests for missing agg (#53214)
Tests for unmapped fields, the missing parameter, scripting, and correct
ValuesSource types in MissingAggregatorTests. Basic yaml tests for the
missing agg

For #42949
2020-04-01 15:23:08 -07:00
William Brafford 958e9d1b78
Refactor nodes stats request builders to match requests (#54363) (#54604)
* Refactor nodes stats request builders to match requests (#54363)

* Remove hard-coded setters from NodesInfoRequestBuilder

* Remove hard-coded setters from NodesStatsRequest

* Use static imports to reduce clutter

* Remove uses of old info APIs
2020-04-01 17:03:04 -04:00
Gordon Brown f0cb8a56a9
Handle -1 gc_threshold settings explicitly (#54546)
Because -1 is technically a valid TimeValue (as a sentinel value), that is now
explicitly checked for when validating gc_thresholds. The tests are also
adjusted to test this case separately from other negative values.
2020-04-01 13:56:50 -06:00
Mayya Sharipova bf4857d9e0
Search hit refactoring (#41656) (#54584)
Refactor SearchHit to have separate document and meta fields.
This is a part of bigger refactoring of issue #24422 to remove
dependency on MapperService to check if a field is metafield.

Relates to PR: #38373
Relates to issue #24422

Co-authored-by: sandmannn <bohdanpukalskyi@gmail.com>
2020-04-01 15:19:00 -04:00
jimczi 7787603d56 Add 7.6.3 version 2020-04-01 16:23:28 +02:00
David Turner 6d976e1468 Resolve some coordination-layer TODOs (#54511)
This commit removes a handful of TODO comments in the cluster coordination
layer that no longer apply.

Relates #32006
2020-04-01 12:36:18 +01:00
David Turner 5e3b6ab82b Use VotingConfiguration#of where possible (#54507)
This resolves a longstanding TODO in the cluster coordination subsystem.

Relates #32006
2020-04-01 09:30:42 +01:00
Nhat Nguyen c2506af8a6 Enable engine debug log for testMaybeFlush
Relates #52223
2020-03-31 23:40:14 -04:00
Jason Tedor 63e5f2b765
Rename META_DATA to METADATA
This is a follow up to a previous commit that renamed MetaData to
Metadata in all of the places. In that commit in master, we renamed
META_DATA to METADATA, but lost this on the backport. This commit
addresses that.
2020-03-31 17:30:51 -04:00
Jason Tedor 5fcda57b37
Rename MetaData to Metadata in all of the places (#54519)
This is a simple naming change PR, to fix the fact that "metadata" is a
single English word, and for too long we have not followed general
naming conventions for it. We are also not consistent about it, for
example, METADATA instead of META_DATA if we were trying to be
consistent with MetaData (although METADATA is correct when considered
in the context of "metadata"). This was a simple find and replace across
the code base, only taking a few minutes to fix this naming issue
forever.
2020-03-31 17:24:38 -04:00
Zachary Tong c9db2de41d
[7.x] Comprehensively test supported/unsupported field type:agg combinations (#54451)
* Comprehensively test supported/unsupported field type:agg combinations (#52493)

This adds a test to AggregatorTestCase that allows us to programmatically
verify that an aggregator supports or does not support a particular
field type.  It fetches the list of registered field type parsers,
creates a MappedFieldType from the parser and then attempts to run
a basic agg against the field.

A supplied list of supported VSTypes are then compared against the
output (success or exception) and suceeds or fails the test accordingly.

Co-Authored-By: Mark Tozzi <mark.tozzi@gmail.com>
* Skip fields that are not aggregatable

* Use newIndexSearcher() to avoid incompatible readers (#52723)

Lucene's `newSearcher()` can generate readers like ParallelCompositeReader
which we can't use.  We need to instead use our helper `newIndexSearcher`
2020-03-31 14:35:03 -04:00
Jake Landis 9b1fe93363
[7.x] introduce 6.8.9 as a version (#53817) 2020-03-31 13:03:28 -05:00
Armin Braun c38e125425
Remove Redundant Documentation on SnapshotsService (#54482) (#54505)
The docs here add nothing compared to those in the package. If anything
they are somewhat confusing since they don't give all necessary details to understand the snapshot process.
=> remove them and link to the complete docs at the package level
2020-03-31 17:07:48 +02:00
Yannick Welsch 597dfa8481 Avoid holding onto bulk items until all completed (#54407)
Bulk requests currently keep a reference to all bulk item requests until every one of them has
completed. There is no need to do so, however, and, in case of large bulks, can mean
unnecessary holding onto memory that might be better used elsewhere. More so as different
shard-level bulks can complete at different speeds, and one slow shard-level request should
not require holding onto every other shard-level request.
2020-03-31 16:19:07 +02:00
Dan Hermann 2ede8662e1
Bump multi-release JARs to Java 11 2020-03-31 06:48:46 -05:00
Tim Brooks 915435bbe4
Fix issue with pipeline releasing bytes early (#54474)
Currently there is an issue with the InboundPipeline releasing bytes
earlier than appropriate. This can lead to the bytes being reused before
the message is handled. This commit fixes that issue and adds a test to
detect when it is occurring.
2020-03-30 22:39:15 -06:00
Lee Hinman a3d1945254
[7.x] Add warnings/errors when V2 templates would match same i… (#54449)
* Add warnings/errors when V2 templates would match same indices… (#54367)

* Add warnings/errors when V2 templates would match same indices as V1

With the introduction of V2 index templates, we want to warn users that templates they put in place
might not take precedence (because v2 templates are going to "win"). This adds this validation at
`PUT` time for both V1 and V2 templates with the following rules:

** When creating or updating a V2 template
- If the v2 template would match indices for an existing v1 template or templates, provide a
warning (through the deprecation logging so it shows up to the client) as well as logging the
warning

The v2 warning looks like:

```
index template [my-v2-template] has index patterns [foo-*] matching patterns from existing older
templates [old-v1-template,match-all-template] with patterns (old-v1-template =>
[foo*],match-all-template => [*]); this template [my-v2-template] will take
precedence during new index creation
```

** When creating a V1 template
- If the v1 template is for index patterns of `"*"` and a v2 template exists, warn that the v2
template may take precedence
- If the v1 template is for index patterns other than all indices, and a v2 template exists that
would match, throw an error preventing creation of the v1 template

** When updating a V1 template (without changing its existing `index_patterns`!)
- If the v1 template is for index patterns that would match an existing v2 template, warn that the
v2 template may take precedence.

The v1 warning looks like:

```
template [my-v1-template] has index patterns [*] matching patterns from existing index templates
[existing-v2-template] with patterns (existing-v2-template => [foo*]); this template [my-v1-template] may be ignored in favor of an index template at index creation time
```

And the v1 error looks like:

```
template [my-v1-template] has index patterns [foo*] matching patterns from existing index templates
[existing-v2-template] with patterns (existing-v2-template => [f*]), use index templates (/_index_template) instead
```

Relates to #53101

* Remove v2 index and component templates when cleaning up tests

* Finish half-finished comment sentence

* Guard template removal and ignore for earlier versions of ES

Co-authored-by: Elastic Machine <elasticmachine@users.noreply.github.com>

* Also ignore 500 errors when clearing index template v2 templates

Co-authored-by: Elastic Machine <elasticmachine@users.noreply.github.com>
2020-03-30 13:25:50 -06:00
Mark Tozzi 529622d4f4
Unit tests for Range and DateRange aggs (#52380) (#54455) 2020-03-30 15:07:43 -04:00
Mark Tozzi 10e0e59561
Tests for agg missing values (#51068) (#54452) 2020-03-30 15:05:38 -04:00
Ryan Ernst 3a24fe9d37
Move keystore-cli to its own tools project (#40787) (#54294)
This commit moves the keystore cli into its own project, so that the
test dependencies can be isolated from the rest of server.
2020-03-30 11:20:07 -07:00
Nik Everett 56047f74be
Fix auto_date_histogram serialization bug (#54447)
This fixes a serialization bug in `auto_date_histogram` that comes up in
a cluster mixed between pre-7.3.0 and post-7.3.0.

Includes #54429 to keep 7.x looking like master for simpler backports.

Closes #54382
2020-03-30 13:49:38 -04:00
Nik Everett e58ad9fed3
Clean up how pipeline aggs check for multi-bucket (backport of #54161) (#54379)
Pipeline aggregations like `stats_bucket`, `sum_bucket`, and
`percentiles_bucket` only operate on buckets that have multiple buckets.
This adds support for those aggregations to `geo_distance`, `ip_range`,
`auto_date_histogram`, and `rare_terms`.

This all happened because we used a marker interface to mark compatible
aggs, `MultiBucketAggregationBuilder` and it was fairly easy to forget
to implement the interface.

This replaces the marker interface with an abstract method in
`AggregationBuilder`, `bucketCardinality` which makes you return `NONE`,
`ONE`, or `MANY`. The `bucket` aggregations can check for `MANY`. At
this point `ONE` and `NONE` amount to about the same thing, but I
suspect that'll be a useful distinction when validating bucket sorts.

Closes #53215
2020-03-30 10:44:55 -04:00
Andrei Dan d5320d9d29
Read the index.number_of_replicas from template so that wait_for_active_shards is interpreted correctly (#54231) (#54413)
This commit takes into account the index.number_of_replicas (defaults to
0 - no replicas- ) value when setting an index template. This change
enables the index.wait_for_active_shards value to be interpreted
correctly

(cherry picked from commit 07026ac3d56dc9fae69467adfda7eaed7ea3ca00)
Signed-off-by: Andrei Dan <andrei.dan@elastic.co>

Co-authored-by: tninokehoe <62655306+tninokehoe@users.noreply.github.com>
2020-03-30 14:34:49 +01:00
Armin Braun 9392fca36a
Improve Snapshot Abort Behavior (#54256) (#54410)
This commit improves the behavior of aborting snapshots and by that fixes
some extremely rare test failures.

Improvements:
1. When aborting a snapshot while it is in the `INIT` stage we do not need
to ever delete anything from the repository because nothing is written to the
repo during INIT any more (in the past running deletes for these snapshots made
sense because we were writing `snap-` and `meta-` blobs during the `INIT` step).
2. Do not try to finalize snapshots that never moved past `INIT`. Same reason as
with the first step. If we never moved past `INIT` no data was written to the repo
so no need to now write a useless entry for the aborted snapshot to `index-N`.
This is especially true, since the reason the snapshot was aborted during `INIT` was
a delete call so the useless empty snapshot just added to `index-N` would be removed
by the subsequent delete that is still waiting anyway.
3. if after aborting a snapshot we wait for it to finish we should not try deleting it
if it failed. If the snapshot failed it means it did not become part of the most recent
`RepositoryData` so a delete for it will needlessly fail with a confusing message about
that snapshot being missing or concurrent repository modification. I moved to throw the snapshot missing exception here because that seems the most user friendly. This allows the user to simply ignore `404` returns from the delete API when using it to make sure a snapshot is aborted+deleted.

Marking this as a non-issue since it doesn't have any negative repercussions other than confusing exceptions on some snapshot aborts.

Closes #52843
2020-03-30 15:08:18 +02:00
Jim Ferenczi 12cfdc24b0 Fixed rewrite of time zone without DST (#54398)
We try to rewrite time zones to fixed offsets in the date histogram aggregation
if the data in the shard is within a single transition.
However this optimization is not applied on time zones that don't apply daylight saving changes
but had some random transitions in the past (e.g. Australia/Brisbane or Asia/Katmandu).
This changes fixes the rewrite of such time zones to fixed offsets.
2020-03-30 13:18:57 +02:00
Martijn van Groningen 4b4fbc160d
Refactor AliasOrIndex abstraction. (#54394)
Backport of #53982

In order to prepare the `AliasOrIndex` abstraction for the introduction of data streams,
the abstraction needs to be made more flexible, because currently it really can be only
an alias or an index.

* Renamed `AliasOrIndex` to `IndexAbstraction`.
* Introduced a `IndexAbstraction.Type` enum to indicate what a `IndexAbstraction` instance is.
* Replaced the `isAlias()` method that returns a boolean with the `getType()` method that returns the new Type enum.
* Moved `getWriteIndex()` up from the `IndexAbstraction.Alias` to the `IndexAbstraction` interface.
* Moved `getAliasName()` up from the `IndexAbstraction.Alias` to the `IndexAbstraction` interface and renamed it to `getName()`.
* Removed unnecessary casting to `IndexAbstraction.Alias` by just checking the `getType()` method.

Relates to #53100
2020-03-30 10:12:16 +02:00
Nhat Nguyen 6e025c12f0 Add debug logging for testRunningTasksCount
Relates #53594
2020-03-29 18:34:41 -04:00
Jason Tedor f0033783db
Deprecate node local storage setting (#54374)
This setting is not documented and has dubious value since it means
there can be nodes in the cluster (non-data and non-master nodes) that
do not have persistent node IDs. This does not have any use cases so
this commit removes the setting.
2020-03-28 14:36:41 -04:00
Jason Tedor 60437b474d
Fix line-length violation in DiscoveryNodeRole
This commit fixes a line-length checkstyle violation in
DiscoveryNodeRole.java.
2020-03-28 13:06:20 -04:00
Jason Tedor 03cab96b2d
Fix imports in discovery node classes
This commit fixes some imports that were leftover after resolving some
merge conflicts on a backport.
2020-03-28 12:56:22 -04:00
Jason Tedor c3be3206ce
Decouple environment from DiscoveryNode (#54373)
Today Environment is coupled to DiscoveryNode via the node.local_storage
setting. This commit decouples Environment from this setting.
2020-03-28 12:52:47 -04:00
Jason Tedor 37b59a357f
Ensure that the output of node roles are sorted (#54376)
This commit ensures that node roles are sorted by node role name, which
makes the output easier to consume, and also makes it easier to rely on
the behavior of the output in assertions.
2020-03-28 12:51:21 -04:00
Tim Brooks 2ccddbfa88
Move transport decoding and aggregation to server (#54360)
Currently all of our transport protocol decoding and aggregation occurs
in the individual transport modules. This means that each implementation
(test, netty, nio) must implement this logic. Additionally, it means
that the entire message has been read from the network before the server
package receives it.

This commit creates a pipeline in server which can be passed arbitrary
bytes to handle. Internally, the pipeline will decode, decompress, and
aggregate the messages. Additionally, this allows us to run many
megabytes of bytes through the pipeline in tests to ensure that the
logic works.

This work will enable future work:

Circuit breaking or backoff logic based on message type and byte
in the content aggregator.
Sharing bytes with the application layer using the ref counted
releasable network bytes.
Improved network monitoring based specifically on channels.
Finally, this fixes the bug where we do not circuit break on the correct
message size when compression is enabled.
2020-03-27 14:13:10 -06:00
Stuart Tettemer 1630de4a42
Scripting: stats per context in nodes stats (#54008) (#54357)
Adds script cache stats to `_node/stats`.
If using the general cache:
```
      "script_cache": {
        "sum": {
          "compilations": 12,
          "cache_evictions": 9,
          "compilation_limit_triggered": 5
        }
      }

```
If using context caches:
```
      "script_cache": {
        "sum": {
          "compilations": 13,
          "cache_evictions": 9,
          "compilation_limit_triggered": 5
        },
        "contexts": [
          {
            "context": "aggregation_selector",
            "compilations": 8,
            "cache_evictions": 6,
            "compilation_limit_triggered": 3
          },
          {
            "context": "aggs",
            "compilations": 5,
            "cache_evictions": 3,
            "compilation_limit_triggered": 2
          },
```
Backport of: 32f46f2
Refs: #50152
2020-03-27 12:26:00 -06:00
Tim Brooks f5b4020819
Remove netty BytesReference implementations (#54355)
Elasticsearch has a number of different BytesReference implementations.
These implementations can all implement the interface in different ways
with subtly different behavior and performance characteristics. On the
other-hand, the JVM only represents bytes as an array or a direct byte
buffer. This commit deletes the specialized Netty implementations and
moves to using a generic ByteBuffer reference type. This will allow us
to focus on standardizing performance and behave around a smaller number
of implementations that can be used by all components in Elasticsearch.
2020-03-27 11:01:33 -06:00
Lee Hinman f2cc2b1127
[7.x] Add REST APIs for IndexTemplateV2Metadata CRUD (#54039) (#54347)
* Add REST APIs for IndexTemplateV2Metadata CRUD (#54039)

* Add REST APIs for IndexTemplateV2Metadata CRUD

This commit adds the get/put/delete APIs for interacting with the now v2 versions of index
templates.

These APIs are behind the existing `es.itv2_feature_flag_registered` system property feature flag.

Relates to #53101

* Add exceptions for HLRC tests

* Add skips for 7.x versions

* Use index_template instead of template_v2 in action names

* Add test for MetaDataIndexTemplateService.addIndexTemplateV2

* Move removal to static method and add test

* Add unit tests for request classes (implement hashCode & equals)

Co-authored-by: Elastic Machine <elasticmachine@users.noreply.github.com>

* Fix compilation

Co-authored-by: Elastic Machine <elasticmachine@users.noreply.github.com>
2020-03-27 10:47:22 -06:00
Dan Hermann 1690e78646
Validation for data stream creation 2020-03-27 10:07:46 -05:00
Alan Woodward 461f1307d6 Add XContentHelper.childBytes() method (#54287)
We have a number of places where we want to read a fairly complex object from
XContent, but aren't interested in its contents; for example, mappings are often
serialized and deserialized between several objects before they are actually built
into a MappingMetaData object. This means that potentially large maps of maps
are constructed several times, only to immediately be re-serialized again.

This commit adds a new helper method to XContentHelper that reads the children
of an xcontent object directly to a BytesReference, serialized via the same xcontenttype
as the parent parser, avoiding the construction of intermediary maps or lists.
2020-03-27 14:21:56 +00:00
Armin Braun 14b5daad7c
Fix Snapshot Completion Listener Lost on Master Failover (#54286) (#54330)
* Fix Snapshot Completion Listener Lost on Master Failover

If master fails over before (or we run into any other exception) when removing
the snapshot from the CS we must still resolve all the completion listeners for
the snapshot.
2020-03-27 14:11:13 +01:00
Gordon Brown 0d30b48613
Disallow negative TimeValues (#53913)
This commit causes negative TimeValues, other than -1 which is sometimes used as
a sentinel value, to be rejected during parsing.

Also introduces a hack to allow ILM to load policies which were written to the
cluster state with a negative min_age, treating those values as 0, which should
match the behavior of prior versions.
2020-03-26 13:30:35 -06:00
William Brafford 14204f8381
Use set-based interface for NodesStatsRequest (#53637) (#54141)
The NodesStatsRequest class uses a set of strings for its internal
serialization. This commit updates the class's interface so that we
no longer use hard-coded getters and setters, but rather
methods that add strings directly. For example, the old way of
adding "os" metrics to a request would be to call request.os(true).
The new way of doing this is to call request.addMetric("os").

For the time being, the canonical list of metrics is an enum in
NodesStatsRequest. This will eventually be replaced with something
pluggable.
2020-03-26 14:41:49 -04:00
Christoph Büscher da404bbce2
HLRC: Don't send defaults for SubmitAsyncSearchRequest (#54200) (#54266)
Currently we set the defaults for ccsMinimizeRoundtrips, preFilterShardSize and
requestCache on the HLRC SubmitAsyncSearchRequest in the constructor. This is no
longer needed since we now only send the parameters along with the rest request
that are supported (omitting e.g. ccsMinimizeRoundtrips) and the correct
defaults are set on the client side. This change removes setting and sending
these defaults where possible, leaving only the overwrite of batchedReduceSize
with a default value of 5, since the default used in the vanilla SearchRequest
is 512. However, we don't need to send this value along as a request parameter
if its the default since the correct one will be set on the receiving end if no
value is specified.
Also adding tests for RestSubmitAsyncSearchAction that check the correct
defaults are set when parameters are missing on the server side.

Backport of #54200
2020-03-26 19:01:17 +01:00