* [ML] have DELETE analytics ignore stats failures and clean up unused stats (#60776)
When deleting an analytics configuration, the request MIGHT fail if
the .ml-stats index does not exist or is in strange state (shards unallocated).
Instead of making the request fail, we should log that we were unable to delete the stats docs and then
have them cleaned up in the 'delete_expire_data' janitorial process
When an exception is thrown during test inference we are
not including the cause message in our logging. This commit
addresses this issue.
Backport of #60749
* Merge test runner task into RestIntegTest (#60261)
* Merge test runner task into RestIntegTest
* Reorganizing Standalone runner and RestIntegTest task
* Rework general test task configuration and extension
* Fix merge issues
* use former 7.x common test configuration
This commit does three things:
* Removes all Copyright/license headers for the build.gradle files under x-pack. (implicit Apache license)
* Removes evaluationDependsOn(xpackModule('core')) from build.gradle files under x-pack
* Removes a place holder test in favor of disabling the test task (in the async plugin)
- Replace immediate task creations by using task avoidance api
- One step closer to #56610
- Still many tasks are created during configuration phase. Tackled in separate steps
Prior to this change ML memory estimation processes for a
given job would always use the same named pipe names. This
would often cause one of the processes to fail.
This change avoids this risk by adding an incrementing counter
value into the named pipe names used for memory estimation
processes.
Backport of #60395
In order to unify model inference and analytics results we
need to write the same fields.
prediction_probability and prediction_score are now written
for inference calls against classification models.
This sets up all indexing to one of our write aliases to require it actually be an alias.
This allows failures scenarios to be captured quickly, loudly, and then potentially recovered.
If a feature is created via a custom pre-processor,
we should return the importance for that feature.
This means we will not return the importance for the
original document field for custom processed features.
closes https://github.com/elastic/elasticsearch/issues/59330
Data frame analytics jobs that work with very large datasets
may produce bulk requests that are over the memory limit
for indexing. This commit adds a helper class that bundles
index requests in bulk requests that steer away from the
memory limit. We then use this class both from the results
joiner and the inference runner ensuring data frame analytics
jobs do not generate bulk requests that are too large.
Note the limit was implemented in #58885.
Backport of #60219
Previously the test was asserting the prediction on each document
was close 10.0 from the expected. It turned out that was not enough
as we occasionally saw the test failing by little.
Instead of relaxing that assertion, this commit changes it to
assert the mean prediction error is less than 10.0. This should
reduce the chances of the test failing significantly.
Fixes#60212
Backport of #60221
When the job is force-closed or shutting down due to a fatal error we clean
up all cancellable job operations. This includes cancelling the results processor.
However, this means that we might not persist objects that are written from the
process like stats, memory usage, etc.
In hindsight, we do not gain from cancelling the results processor in its
entirety. It makes more sense to skip row results and model chunks but keep
stats and instrumentation about the job as the latter may contain useful information
to understand what happened to the job.
Backport of #60113
Putting an ingest pipeline used to require that the user calling
it had permission to get nodes info as well as permission to
manage ingest. This was due to an internal implementaton detail
that was not visible to the end user.
This change alters the behaviour so that a user with the
manage_pipeline cluster privilege can put an ingest pipeline
regardless of whether they have the separate privilege to get
nodes info. The internal implementation detail now runs as
the internal _xpack user when security is enabled.
Backport of #60106
This commit continues on the work in #59801 and makes other
implementors of the LocalNodeMasterListener interface thread safe in
that they will no longer allow the callbacks to run on different
threads and possibly race each other. This also helps address other
issues where these events could be queued to wait for execution while
the service keeps moving forward thinking it is the master even when
that is not the case.
In order to accomplish this, the LocalNodeMasterListener no longer has
the executorName() method to prevent future uses that could encounter
this surprising behavior.
Each use was inspected and if the class was also a
ClusterStateListener, the implementation of LocalNodeMasterListener
was removed in favor of a single listener that combined the logic. A
single listener is used and there is currently no guarantee on execution
order between ClusterStateListeners and LocalNodeMasterListeners,
so a future change there could cause undesired consequences. For other
classes, the implementations of the callbacks were inspected and if the
operations were lightweight, the overriden executorName method was
removed to use the default, which runs on the same thread.
Backport of #59932
In #58877, when we switched test inference on java, we just
use the doc's `_source` as features. However, this could be
missing out on features that were used during training,
e.g. alias fields, etc.
This commit addresses this by extracting fields to use as
features during inference the same way they are extracted
in `DataFrameDataExtractor` when they are used for training.
Backport of #59963
* [ML] add new `custom` field to trained model processors (#59542)
This commit adds the new configurable field `custom`.
`custom` indicates if the preprocessor was submitted by a user or automatically created by the analytics job.
Eventually, this field will be used in calculating feature importance. When `custom` is true, the feature importance for
the processed fields is calculated. When `false` the current behavior is the same (we calculate the importance for the originating field/feature).
This also adds new required methods to the preprocessor interface. If users are to supply their own preprocessors
in the analytics job configuration, we need to know the input and output field names.
When an inference model is loaded it is accounted for in circuit breaker
and should not be released until there are no users of the model. Adds
a reference count to the model to track usage.
Backport of #59525 to 7.x branch.
* Actions are moved to xpack core.
* Transport and rest actions are moved the data-streams module.
* Removed data streams methods from Client interface.
* Adjusted tests to use client.execute(...) instead of data stream specific methods.
* only attempt to delete all data streams if xpack is installed in rest tests
* Now that ds apis are in xpack and ESIntegTestCase
no longers deletes all ds, do that in the MlNativeIntegTestCase
class for ml tests.
This commit adds a new api to track when gold+ features are used within
x-pack. The tracking is done internally whenever a feature is checked
against the current license. The output of the api is a list of each
used feature, which includes the name, license level, and last time it
was used. In addition to a unit test for the tracking, a rest test is
added which ensures starting up a default configured node does not
result in any features registering as used.
There are a couple features which currently do not work well with the
tracking, as they are checked in a manner that makes them look always
used. Those features will be fixed in followups, and in this PR they are
omitted from the feature usage output.
The `create_doc`, `create`, `write` and `index` privileges do not grant
the PutMapping action anymore. Apart from the `write` privilege, the other
three privileges also do NOT grant (auto) updating the mapping when ingesting
a document with unmapped fields, according to the templates.
In order to maintain the BWC in the 7.x releases, the above privileges will still grant
the Put and AutoPutMapping actions, but only when the "index" entity is an alias
or a concrete index, but not a data stream or a backing index of a data stream.
Renames and moves the cross validation splitter package.
First, the package and classes are renamed from using
"cross validation splitter" to "train test splitter".
Cross validation as a term is overloaded and encompasses
more concepts than what we are trying to do here.
Second, the package used to be under `process` but it does
not make sense to be there, it can be a top level package
under `dataframe`.
Backport of #59529
When a field is not included yet its type is unsupported, we currently
state that the reason the field is excluded is that it is not in the
includes list. However, this implies the user could include it but
if the user tried to do so, they would get a failure as they would
be including a field with unsupported type.
This commit improves this by stating the reason a not included field
with unsupported type is excluded is because of its type.
Backport of #59424
Since we have added checking the cardinality of the dependent_variable
for classification, we have introduced a bug where an NPE is thrown
if the dependent_variable is a missing field.
This commit is fixing this issue.
Backport of #59524
This makes the data_stream timestamp field specification optional when
defining a composable template.
When there isn't one specified it will default to `@timestamp`.
(cherry picked from commit 5609353c5d164e15a636c22019c9c17fa98aac30)
Signed-off-by: Andrei Dan <andrei.dan@elastic.co>
Now that we have per-partition categorization, the estimate for
the model memory limit required for a particular analysis config
needs to take into account whether categorization is operating
for the job as a whole or per-partition.
We have recently added internal metrics to monitor the amount of
indexing occurring on a node. These metrics introduce back pressure to
indexing when memory utilization is too high. This commit exposes these
stats through the node stats API.
Removes member variable `index` from `ExtractedFieldsDetector`
as it is not used.
Backport of #59395
Co-authored-by: Elastic Machine <elasticmachine@users.noreply.github.com>
Backport of #59293 to 7.x branch.
* Create new data-stream xpack module.
* Move TimestampFieldMapper to the new module,
this results in storing a composable index template
with data stream definition only to work with default
distribution. This way data streams can only be used
with default distribution, since a data stream can
currently only be created if a matching composable index
template exists with a data stream definition.
* Renamed `_timestamp` meta field mapper
to `_data_stream_timestamp` meta field mapper.
* Add logic to put composable index template api
to fail if `_data_stream_timestamp` meta field mapper
isn't registered. So that a more understandable
error is returned when attempting to store a template
with data stream definition via the oss distribution.
In a follow up the data stream transport and
rest actions can be moved to the xpack data-stream module.
With the introduction of per-partition categorization the old
logic for creating a job notification for categorization status
"warn" does not work. However, the C++ code is already writing
annotations for categorization status "warn" that take into
account whether per-partition categorization is being used and
which partition(s) the warnings relate to. Therefore, this
change alters the Java results processor to create notifications
based on the annotations the C++ writes. (It is arguable that
we don't need both annotations and notifications, but they show
up in different ways in the UI: only annotations are visible in
results and only notifications set the warning symbol in the
jobs list. This means it's best to have both.)
Backport of #59377
This adds a setting to data frame analytics jobs called
`max_number_threads`. The setting expects a positive integer.
When used the user specifies the max number of threads that may
be used by the analysis. Note that the actual number of threads
used is limited by the number of processors on the node where
the job is assigned. Also, the process may use a couple more threads
for operational functionality that is not the analysis itself.
This setting may also be updated for a stopped job.
More threads may reduce the time it takes to complete the job at the cost
of using more CPU.
Backport of #59254 and #57274
Since we are able to load the inference model
and perform inference in java, we no longer need
to rely on the analytics process to be performing
test inference on the docs that were not used for
training. The benefit is that we do not need to
send test docs and fit them in memory of the c++
process.
Backport of #58877
Co-authored-by: Dimitris Athanasiou <dimitris@elastic.co>
Co-authored-by: Benjamin Trent <ben.w.trent@gmail.com>
Backport of #59076 to 7.x branch.
The commit makes the following changes:
* The timestamp field of a data stream definition in a composable
index template can only be set to '@timestamp'.
* Removed custom data stream timestamp field validation and reuse the validation from `TimestampFieldMapper` and
instead only check that the _timestamp field mapping has been defined on a backing index of a data stream.
* Moved code that injects _timestamp meta field mapping from `MetadataCreateIndexService#applyCreateIndexRequestWithV2Template58956(...)` method
to `MetadataIndexTemplateService#collectMappings(...)` method.
* Fixed a bug (#58956) that cases timestamp field validation to be performed
for each template and instead of the final mappings that is created.
* only apply _timestamp meta field if index is created as part of a data stream or data stream rollover,
this fixes a docs test, where a regular index creation matches (logs-*) with a template with a data stream definition.
Relates to #58642
Relates to #53100Closes#58956Closes#58583
This removes the blocking model lookup from the `inference` aggregator's
builder by integrating it into the request rewrite process that loads
stuff asynchronously.
Co-authored-by: Elastic Machine <elasticmachine@users.noreply.github.com>
There have been a few test failures that are likely caused by tests
performing actions that use ML indices immediately after the actions
that create those ML indices. Currently this can result in attempts
to search the newly created index before its shards have initialized.
This change makes the method that creates the internal ML indices
that have been affected by this problem (state and stats) wait for
the shards to be initialized before returning.
Backport of #59027
This commit creates a new Gradle plugin to provide a separate task name
and source set for running YAML based REST tests. The only project
converted to use the new plugin in this PR is distribution/archives/integ-test-zip.
For which the testing has been moved to :rest-api-spec since it makes the most
sense and it avoids a small but awkward change to the distribution plugin.
The remaining cases in modules, plugins, and x-pack will be handled in followups.
This plugin is distinctly different from the plugin introduced in #55896 since
the YAML REST tests are intended to be black box tests over HTTP. As such they
should not (by default) have access to the classpath for that which they are testing.
The YAML based REST tests will be moved to separate source sets (yamlRestTest).
The which source is the target for the test resources is dependent on if this
new plugin is applied. If it is not applied, it will default to the test source
set.
Further, this introduces a breaking change for plugin developers that
use the YAML testing framework. They will now need to either use the new source set
and matching task, or configure the rest resources to use the old "test" source set that
matches the old integTest task. (The former should be preferred).
As part of this change (which is also breaking for plugin developers) the
rest resources plugin has been removed from the build plugin and now requires
either explicit application or application via the new YAML REST test plugin.
Plugin developers should be able to fix the breaking changes to the YAML tests
by adding apply plugin: 'elasticsearch.yaml-rest-test' and moving the YAML tests
under a yamlRestTest folder (instead of test)
Backport of #58582 to 7.x branch.
This commit adds a new metadata field mapper that validates,
that a document has exactly a single timestamp value in the data stream timestamp field and
that the timestamp field mapping only has `type`, `meta` or `format` attributes configured.
Other attributes can affect the guarantee that an index with this meta field mapper has a
useable timestamp field.
The MetadataCreateIndexService inserts a data stream timestamp field mapper whenever
a new backing index of a data stream is created.
Relates to #53100
Dry up tests that use a disruption that isolates the master from all other nodes.
Also, turn disruption types that have neither parameters nor state into constants
to make things a little clearer.
Working through a heap dump for an unrelated issue I found that we can easily rack up
tens of MBs of duplicate empty instances in some cases.
I moved to a static constructor to guard against that in all cases.
.ml-state-write is supposed to be an index alias, however by accident it can become an index. If
.ml-state-write is a concrete index instead of an alias ML stops working. This change improves error
handling by setting the job to failed and properly log and audit the problem. The user still has to
manually fix the problem. This change should lead to a quicker resolution of the problem.
fixes#58482
We already had code to ensure the config index mappings were
up-to-date before creating a new config. However, it's also
possible that an update to a config could add the latest
settings that require the latest mappings to index correctly.
This change checks that the latest config index mappings are
in place in the 3 update actions in the same way as the checks
are done in the 3 put actions.
Backport of #58916
* [ML] handles compressed model stream from native process (#58009)
This moves model storage from handling the fully parsed JSON string to handling two separate types of documents.
1. ModelSizeInfo which contains model size information
2. TrainedModelDefinitionChunk which contains a particular chunk of the compressed model definition string.
`model_size_info` is assumed to be handled first. This will generate the model_id and store the initial trained model config object. Then each chunk is assumed to be in correct order for concatenating the chunks to get a compressed definition.
Native side change: https://github.com/elastic/ml-cpp/pull/1349
The checks on the license state have a singular method, isAllowed, that
returns whether the given feature is allowed by the current license.
However, there are two classes of usages, one which intends to actually
use a feature, and another that intends to return in telemetry whether
the feature is allowed. When feature usage tracking is added, the latter
case should not count as a "usage", so this commit reworks the calls to
isAllowed into 2 methods, checkFeature, which will (eventually) both
check whether a feature is allowed, and keep track of the last usage
time, and isAllowed, which simply determines whether the feature is
allowed.
Note that I considered having a boolean flag on the current method, but
wanted the additional clarity that a different method name provides,
versus a boolean flag which is more easily copied without realizing what
the flag means since it is nameless in call sites.
This PR implements recursive mapping merging for composable index templates.
When creating an index, we perform the following:
* Add each component template mapping in order, merging each one in after the
last.
* Merge in the index template mappings (if present).
* Merge in the mappings on the index request itself (if present).
Some principles:
* All 'structural' changes are disallowed (but everything else is fine). An
object mapper can never be changed between `type: object` and `type: nested`. A
field mapper can never be changed to an object mapper, and vice versa.
* Generally, each section is merged recursively. This includes `object`
mappings, as well as root options like `dynamic_templates` and `meta`. Once we
reach 'leaf components' like field definitions, they always overwrite an
existing one instead of being merged.
Relates to #53101.
When per_partition_categorization.stop_on_warn is set for an analysis
config it is now passed through to the autodetect C++ process.
Also adds some end-to-end tests that exercise the functionality
added in elastic/ml-cpp#1356
Backport of #58632
* Replace compile configuration usage with api (#58451)
- Use java-library instead of plugin to allow api configuration usage
- Remove explicit references to runtime configurations in dependency declarations
- Make test runtime classpath input for testing convention
- required as java library will by default not have build jar file
- jar file is now explicit input of the task and gradle will ensure its properly build
* Fix compile usages in 7.x branch
Adds parsing of `status` and `memory_reestimate_bytes`
to data frame analytics `memory_usage`. When the training surpasses
the model memory limit, the status will be set to `hard_limit` and
`memory_reestimate_bytes` can be used to update the job's
limit in order to restart the job.
Backport of #58588
Today we have individual settings for configuring node roles such as
node.data and node.master. Additionally, roles are pluggable and we have
used this to introduce roles such as node.ml and node.voting_only. As
the number of roles is growing, managing these becomes harder for the
user. For example, to create a master-only node, today a user has to
configure:
- node.data: false
- node.ingest: false
- node.remote_cluster_client: false
- node.ml: false
at a minimum if they are relying on defaults, but also add:
- node.master: true
- node.transform: false
- node.voting_only: false
If they want to be explicit. This is also challenging in cases where a
user wants to have configure a coordinating-only node which requires
disabling all roles, a list which we are adding to, requiring the user
to keep checking whether a node has acquired any of these roles.
This commit addresses this by adding a list setting node.roles for which
a user has explicit control over the list of roles that a node has. If
the setting is configured, the node has exactly the roles in the list,
and not any additional roles. This means to configure a master-only
node, the setting is merely 'node.roles: [master]', and to configure a
coordinating-only node, the setting is merely: 'node.roles: []'.
With this change we deprecate the existing 'node.*' settings such as
'node.data'.
* [ML] make waiting for renormalization optional for internally flushing job (#58537)
When flushing, datafeeds only need the guaruntee that the latest bucket has been handled.
But, in addition to this, the typical call to flush waits for renormalization to complete. For large jobs, this can take a fair bit of time (even longer than a bucket length). This causes unnecessary delays in handling data.
This commit adds a new internal only flag that allows datafeeds (and forecasting) to skip waiting on renormalization.
closes#58395
This changes the default value for the results field of inference
applied on models that are trained via a data frame analytics job.
Previously, the results field default was `predicted_value`. This
commit makes it the same as in the training job itself. The new
default field is `<dependent_variable>_prediction`. Apart from
making inference consistent with the training job the model came
from, it is helpful to preserve the dependent variable name
by default as it provides some context to the user that may
avoid confusion as to which model results came from.
Backport of #58538
This commits allows data streams to be a valid source for analytics and transforms.
Data streams are fairly transparent and our `_search` and `_reindex` actions work without error.
For `_transforms` the check-pointing works as desired as well. Data streams are effectively treated as an `alias` and the backing index values are stored within checkpointing information.
Unlike `classification`, which is using a cross validation splitter
that produces training sets whose size is predictable and equal to
`training_percent * class_cardinality`, for regression we have been
using a random splitter that takes an independent decision for each
document. This means we cannot predict the exact size of the training
set. This poses a problem as we move towards performing test inference
on the java side as we need to be able to provide an accurate upper
bound of the training set size to the c++ process.
This commit replaces the random splitter we use for regression with
the same streaming-reservoir approach we do for `classification`.
Backport of #58331
There was a discrepancy in the implementation of flush
acknowledgements: most of the class was designed on the
basis that the "last finalized bucket time" could be null
but the wire serialization assumed that it was never
null. This works because, the C++ sends zero "last
finalized bucket time" when it is not known or not
relevant. But then the Java code will print that to
XContent as it is assuming null represents not known or
not relevant.
This change corrects the discrepancies. Internally within
the class null represents not known or not relevant, but
this is translated from/to 0 for communications from the
C++ and old nodes that have the bug.
Additionally I switched from Date to Instant for this
class and made the member variables final to modernise it
a bit.
Backport of #58413
Backporting #58096 to 7.x branch.
Relates to #53100
* use mapping source direcly instead of using mapper service to extract the relevant mapping details
* moved assertion to TimestampField class and added helper method for tests
* Improved logic that inserts timestamp field mapping into an mapping.
If the timestamp field path consisted out of object fields and
if the final mapping did not contain the parent field then an error
occurred, because the prior logic assumed that the object field existed.
When a local model is constructed, the cache hit miss count is incremented.
When a user calls _stats, we will include the sum cache hit miss count across ALL nodes. This statistic is important to in comparing against the inference_count. If the cache hit miss count is near the inference_count it indicates that the cache is overburdened, or inappropriately configured.
This commit bumps our JNA dependency from 4.5.1 to 5.5.0, so that we are
now on the latest maintained line, and pick up a large collection of bug
fixes that have accumulated.
The main improvement here is that the total expected
count of training rows in the test is calculated as the
sum of the training fraction times the cardinality of each
class (instead of the training fraction times the total doc count).
Also relaxes slightly the error bound on the uniformity test from 0.12
to 0.13.
Closes#54122
Backport of #58180
This commit adds an optional field, `description`, to all ingest processors
so that users can explain the purpose of the specific processor instance.
Closes#56000.
* Remove usage of deprecated testCompile configuration
* Replace testCompile usage by testImplementation
* Make testImplementation non transitive by default (as we did for testCompile)
* Update CONTRIBUTING about using testImplementation for test dependencies
* Fail on testCompile configuration usage
This type of result will store stats about how well categorization
is performing. When per-partition categorization is in use, separate
documents will be written for every partition so that it is possible
to see if categorization is working well for some partitions but not
others.
This PR is a minimal implementation to allow the C++ side changes to
be made. More Java side changes related to per-partition
categorization will be in followup PRs. However, even in the long
term I do not see a major benefit in introducing dedicated APIs for
querying categorizer stats. Like forecast request stats the
categorizer stats can be read directly from the job's results alias.
Backport of #57978
Adds support for reading in `model_size_info` objects.
These objects contain numeric values indicating the model definition size and complexity.
Additionally, these objects are not stored or serialized to any other node. They are to be used for calculating and storing model metadata. They are much smaller on heap than the true model definition and should help prevent the analytics process from using too much memory.
Co-authored-by: Elastic Machine <elasticmachine@users.noreply.github.com>
Now that annotations are part of the anomaly detection job results
the annotations index should be refreshed on flushing and closing
the job so that flush and close continue to fulfil their contracts
that immediately after returning all results the job generated up
to that point are searchable.
Benjamin Trent
Dimitris Athanasiou
Przemysław Witek
Rene Groeschke
Jake Landis
David Roberts
Jay Modi
David Kyle
Martijn van Groningen
Ryan Ernst
Albert Zaharovits
Andrei Dan
Tim Brooks
Nik Everett
Armin Braun
Hendrik Muhs
Lee Hinman
Julie Tibshirani
Jason Tedor
Luca Cavanna
Tal Levy