The test failure in #31916 revealed that updating
rules on a job was modifying the detectors list
in-place. That meant the old cluster state and the
updated cluster state had no difference and thus the
change was not propagated to non-master nodes.
This commit fixes that and also reviews all of ML
metadata in order to ensure immutability.
Closes#31916
A scroll holds a reference to the shard store. If the cluster is moving shards
around that reference can prevent a shard from relocating back to node it used
to be on, causing test failures.
Closes#31827
For historical reasons SQL restricts GROUP BY to only one field.
This commit removes the restriction and improves the test suite with
multi group by tests.
Close#31793
Register SQL as an xpackModule
Specify group for SQL QA to disambiguate projects (otherwise due to an
old Gradle bug (https://github.com/gradle/gradle/issues/847) any
subprojects under SQL QA will not be able to refer to SQL xpackModule
Co-authored-by: Alpar Torok <torokalpar@gmail.com>
Significantly improve the example snippets in the documentation.
The examples are part of the test suite and checked nightly.
To help readability, the existing dataset was extended (test_emp renamed
to emp plus library).
Improve output of JDBC tests to be consistent with the CLI
Add lenient flag to JDBC asserts to allow type widening (a long is
equivalent to a integer as long as the value is the same).
Some proxies require all requests to have paths starting with / since
there are no relative paths at the HTTP connection level. Elasticsearch
assumes paths are absolute. In order to run rest tests against a cluster
behind such a proxy, set the system property
tests.rest.client_path_prefix to /.
Make password hashing algorithm/cost configurable for the
stored passwords of users for the realms that this applies
(native, reserved). Replaces predefined choice of bcrypt with
cost factor 10.
This also introduces PBKDF2 with configurable cost
(number of iterations) as an algorithm option for password hashing
both for storing passwords and for the user cache.
Password hash validation algorithm selection takes into
consideration the stored hash prefix and only a specific number
of algorithnm and cost factor options for brypt and pbkdf2 are
whitelisted and can be selected in the relevant setting.
* Move to Gradle 4.8 RC1
* Use latest version of plugin
The current does not work with Gradle 4.8 RC1
* Switch to Gradle GA
* Add and configure build compare plugin
* add work-around for https://github.com/gradle/gradle/issues/5692
* work around https://github.com/gradle/gradle/issues/5696
* Make use of Gradle build compare with reference project
* Make the manifest more compare friendly
* Clear the manifest in compare friendly mode
* Remove animalsniffer from buildscript classpath
* Fix javadoc errors
* Fix doc issues
* reference Gradle issues in comments
* Conditionally configure build compare
* Fix some more doclint issues
* fix typo in build script
* Add sanity check to make sure the test task was replaced
Relates to #31324. It seems like Gradle has an inconsistent behavior and
the taks is not always replaced.
* Include number of non conforming tasks in the exception.
* No longer replace test task, create implicit instead
Closes#31324. The issue has full context in comments.
With this change the `test` task becomes nothing more than an alias for `utest`.
Some of the stand alone tests that had a `test` task now have `integTest`, and a
few of them that used to have `integTest` to run multiple tests now only
have `check`.
This will also help separarate unit/micro tests from integration tests.
* Revert "No longer replace test task, create implicit instead"
This reverts commit f1ebaf7d93e4a0a19e751109bf620477dc35023c.
* Fix replacement of the test task
Based on information from gradle/gradle#5730 replace the task taking
into account the task providres.
Closes#31324.
* Only apply build comapare plugin if needed
* Make sure test runs before integTest
* Fix doclint aftter merge
* PR review comments
* Switch to Gradle 4.8.1 and remove workaround
* PR review comments
* Consolidate task ordering
Merges the `query-builder-bwc` qa project into the
`full-cluster-restart` qa project, saving a cluster starts on every
build and *many* cluster starts on `./gradlew bwcTests`.
This creates a YAML test "features" that indices if the cluster being
tested has xpack installed (`xpack`) or if it does *not* have xpack
installed (`no_xpack`). It uses those features to centralize skipping
a few tests that fail if xpack is installed.
The plan is to use this in a followup to skip docs tests that require
xpack when xpack is not installed. We *plan* to use the declaration
of required license level on the docs page to generate the required
`skip`.
Closes#30933.
This pull request adds a full cluster restart test for a Rollup job.
The test creates and starts a Rollup job on the cluster and checks
that the job already exists and is correctly started on the upgraded
cluster.
This test allows to test that the persistent task state is correctly
parsed from the cluster state after the upgrade, as the status field
has been renamed to state in #31031.
The test undercovers a ClassCastException that can be thrown in
the RollupIndexer when the timestamp as a very low value that fits
into an integer. When it's the case, the value is parsed back as an
Integer instead of Long object and (long) position.get(rollupFieldName)
fails.
This adds an api to allow updating a filter:
POST _xpack/ml/filters/{filter_id}/_update
The request body may have:
- description: setting a new description
- add_items: a list of the items to add
- remove_items: a list of the items to remove
This commit also changes the PUT filter api to
error when the filter_id is already used. As
now there is an api for updating filters, the
put api should only be used to create new ones.
Also, updating a filter results into a notification
message auditing the change for every job that is
using that filter.
This pull request removes the relationship between the state
of persistent task (as stored in the cluster state) and the status
of the task (as reported by the Task APIs and used in various
places) that have been confusing for some time (#29608).
In order to do that, a new PersistentTaskState interface is added.
This interface represents the persisted state of a persistent task.
The methods used to update the state of persistent tasks are
renamed: updatePersistentStatus() becomes updatePersistentTaskState()
and now takes a PersistentTaskState as a parameter. The
Task.Status type as been changed to PersistentTaskState in all
places were it make sense (in persistent task customs in cluster
state and all other methods that deal with the state of an allocated
persistent task).
This adds a `description` to ML filters in order
to allow users to describe their filters in a human
readable form which is also editable (filter updates
to be added shortly).
We currently have a specific REST action to retrieve all aliaes, which
uses internally the get index API. This doesn't seem to be required
anymore though as the existing RestGetAliaesAction could as well take
the requests with no indices and aliases specified.
This commit removes the RestGetAllAliasesAction in favour of using
RestGetAliasesAction also for requests that don't specify indices nor
aliases. Similar to #31129.
Rules allow users to supply a detector with domain
knowledge that can improve the quality of the results.
The model detects statistically anomalous results but it
has no knowledge of the meaning of the values being modelled.
For example, a detector that performs a population analysis
over IP addresses could benefit from a list of IP addresses
that the user knows to be safe. Then anomalous results for
those IP addresses will not be created and will not affect
the quantiles either.
Another example would be a detector looking for anomalies
in the median value of CPU utilization. A user might want
to inform the detector that any results where the actual
value is less than 5 is not interesting.
This commit introduces a `custom_rules` field to the `Detector`.
A detector may have multiple rules which are combined with `or`.
A rule has 3 fields: `actions`, `scope` and `conditions`.
Actions is a list of what should happen when the rule applies.
The current options include `skip_result` and `skip_model_update`.
The default value for `actions` is the `skip_result` action.
Scope is optional and allows for applying filters on any of the
partition/over/by field. When not defined the rule applies to
all series. The `filter_id` needs to be specified to match the id
of the filter to be used. Optionally, the `filter_type` can be specified
as either `include` (default) or `exclude`. When set to `include`
the rule applies to entities that are in the filter. When set to
`exclude` the rule only applies to entities not in the filter.
There may be zero or more conditions. A condition requires `applies_to`,
`operator` and `value` to be specified. The `applies_to` value can be
either `actual`, `typical` or `diff_from_typical` and it specifies
the numerical value to which the condition applies. The `operator`
(`lt`, `lte`, `gt`, `gte`) and `value` complete the definition.
Conditions are combined with `and` and allow to specify numerical
conditions for when a rule applies.
A rule must either have a scope or one or more conditions. Finally,
a rule with scope and conditions applies when all of them apply.
This is in preparation of pushing the new
rules design in the `ml-cpp` side. These
tests will be switched on again after merging
in the new rules implementation.
This commit upgrades us to Netty 4.1.25. This upgrade is more
challenging than past upgrades, all because of a new object cleaner
thread that they have added. This thread requires an additional security
permission (set context class loader, needed to avoid leaks in certain
scenarios). Additionally, there is not a clean way to shutdown this
thread which means that the thread can fail thread leak control during
tests. As such, we have to filter this thread from thread leak control.
The core REST tests with security currently use a hardcoded username and
password. This is not amenable to running these tests in scenarios where
the user controls the creation of the cluster and owns the credentials
for this cluster. This commit enables running the core REST tests with
security with a custom username and password.
The goal of this commit is to address unknown licenses when producing
the dependencies info report. We have two different checks that we run
on licenses. The first check is whether or not we have stashed a copy of
the license text for a dependency in the repository. The second is to
map every dependency to a license type (e.g., BSD 3-clause). The problem
here is that the way we were handling licenses in the second check
differs from how we handle licenses in the first check. The first check
works by finding a license file with the name of the artifact followed
by the text -LICENSE.txt. Yet in some cases we allow mapping an artifact
name to another name used to check for the license (e.g., we map
lucene-.* to lucene, and opensaml-.* to shibboleth. The second check
understood the first way of looking for a license file but not the
second way. So in this commit we teach the second check about the
mappings from artifact names to license names. We do this by copying the
configuration from the dependencyLicenses task to the dependenciesInfo
task and then reusing the code from the first check in the second
check. There were some other challenges here though. For example,
dependenciesInfo was checking too many dependencies. For now, we should
only be checking direct dependencies and leaving transitive dependencies
from another org.elasticsearch artifact to that artifact (we want to do
this differently in a follow-up). We also want to disable
dependenciesInfo for projects that we do not publish, users only care
about licenses they might be exposed to if they use our assembled
products. With all of the changes in this commit we have eliminated all
unknown licenses. A follow-up will enforce that when we add a new
dependency it does not get mapped to unknown, these will be forbidden in
the future. Therefore, with this change and earlier changes are left
having no unknown licenses and two custom licenses; custom here means it
does not map to an SPDX license type. Those two licenses are xz and
ldapsdk. A future change will not allow additional custom licenses
unless they are explicitly whitelisted. This ensures that if a new
dependency is added it is mapped to an SPDX license or mapped to custom
because it does not have an SPDX license.
Use all running nodes as unicast seeds in the rolling restart tests to
avoid a race between pinging and the tests. Without this if the tests
are too fast then when a new node comes up and pings its single
configured seed node that node *might* not have a ping from the other
running node.
I pushed a test that `assertBusy`s for a whole hour accidentally. I was
testing something and forgot to revert my local hack but caught it on
backport. This removes it.
This is much more realistic and can find more issues. This causes the
"mixed cluster" tests to be run twice so I had to fix the tests to work
in that case. In most cases I did as little as possible to get them
working but in a few cases I went a little beyond that to make them
easier for me to debug while getting them to work. My test changes:
1. Remove the "basic indexing" tests and replace them with a copy of the
tests used in the OSS. We have no way of sharing code between these two
projects so for now I copy.
2. Skip the a few tests in the "one third" upgraded scenario:
* creating a scroll to be reused when the cluster is fully upgraded
* creating some ml data to be used when the cluster is fully ugpraded
3. Drop many "assert yellow and that the cluster has two nodes"
assertions. These assertions duplicate those made by the wait condition
and they fail now that we have three nodes.
4. Switch many "assert green and that the cluster has two nodes" to 3
nodes. These assertions are unique from the wait condition and, while
I imagine they aren't required in all cases, now is not the time to
find that out. Thus, I made them work.
5. Rework the index audit trail test so it is more obvious that it is
the same test expecting different numbers based on the shape of the
cluster. The conditions for which number are expected are fairly
complex because the index audit trail is shut down until the template
for it is upgraded and the template is upgraded when a master node is
elected that has the new version of the software.
6. Add some more information to debug the index audit trail test because
it helped me figure out what was going on.
I also dropped the `waitCondition` from the `rolling-upgrade-basic`
tests because it wasn't needed.
Closes#25336
In case an error is returned when calling search_shards on a remote
cluster, which will lead to throwing an exception in the coordinating
node, we should make sure that the status code returned by the
coordinating node is the same as the one returned by the remote
cluster. Up until now a 500 - Internal Server Error was always
returned. This commit changes this behaviour so that for instance if an
index is not found, which causes an 404, a 404 is also returned by the
coordinating node to the client.
Closes#27461
This modifies the high level rest client to allow calling code to
customize per request options for the bulk API. You do the actual
customization by passing a `RequestOptions` object to the API call
which is set on the `Request` that is generated by the high level
client. It also makes the `RequestOptions` a thing in the low level
rest client. For now that just means you use it to customize the
headers and the `httpAsyncResponseConsumerFactory` and we'll add
node selectors and per request timeouts in a follow up.
I only implemented this on the bulk API because it is the first one
in the list alphabetically and I wanted to keep the change small
enough to review. I'll convert the remaining APIs in a followup.
This commit removes the RequestBuilder generic type from Action. It was
needed to be used by the newRequest method, which in turn was used by
client.prepareExecute. Both of these methods are now removed, along with
the existing users of prepareExecute constructing the appropriate
builder directly.
Currently failures to compile a script usually lead to a ScriptException, which
inherits the 500 INTERNAL_SERVER_ERROR from ElasticsearchException if it does
not contain another root cause. Instead, this should be a 400 Bad Request error.
This PR changes this more generally for script compilation errors by changing
ScriptException to return 400 (bad request) as status code.
Closes#12315
Limits the scope of the runtime dependency on
BouncyCastle so that it can be eventually removed.
* Splits functionality related to reading and generating certificates
and keys in two utility classes so that reading certificates and
keys doesn't require BouncyCastle.
* Implements a class for parsing PEM Encoded key material (which also
adds support for reading PKCS8 encoded encrypted private keys).
* Removes BouncyCastle dependency for all of our test suites(except
for the tests that explicitly test certificate generation) by using
pre-generated keys/certificates/keystores.
Removes the last remaining server dependencies from jdbc client. In order to do that it introduces the new project sql-shared-proto that contains only XContent-serializable classes. HTTP Client and JDBC now depend only on sql-shared-proto. I had to keep the original sql-proto project since it is used as a dependency by sql-cli and security integration tests.
Relates #29856
Persistent tasks was moved from X-Pack to core in #28455.
However, registration of the named writables and named
X-content was left in X-Pack.
This change moves the registration of the named writables
and named X-content into core. Additionally, the persistent
task actions are no longer registered in the X-Pack client
plugin, as they are already registered in ActionModule.
This commit adds the ability to configure how a docvalue field should be
formatted, so that it would be possible eg. to return a date field
formatted as the number of milliseconds since Epoch.
Closes#27740
Adding headers rather than setting them all at once seems more
user-friendly and we already do it in a similar way for parameters
(see Request#addParameter).
Make all bool constructs use match/should (that is a query context) as
that is controlled and changed to a filter context by ES automatically
based on the sort order (_doc, field vs _sort) and trackScores.
Fix#29685
By default ML native processes are only allowed to use
30% of RAM, so the previous 2GB setting prevented the
test passing on VMs with only 4GB RAM. This change
reduces the limit to 1200MB, which means it can now
pass on VMs with 4GB RAM.
Meta plugins existed only for a short time, in order to enable breaking
up x-pack into multiple plugins. However, now that x-pack is no longer
installed as a plugin, the need for them has disappeared. This commit
removes the meta plugins infrastructure.
As the first record is random, there's a chance it will
be aligned on a bucket start. Thus we need to check the
bucket count is in [23, 24].
Closes#30715
These tests aim to check the set model memory limit is
respected. Additionally, it was asserting counts of
partition, by, over fields in an attempt to check that
the used memory is spent meaningfully. However, this
made the tests fragile, as changes in the ml-cpp could
lead to CI failures.
This commit removes those assertions. We are working on
adding tests in ml-cpp that will compensate.
diskspace and creates a subfolder for storing data outside of Lucene
indexes, but as part of the ES data paths.
Details:
- tmp storage is managed and does not allow allocation if disk space is
below a threshold (5GB at the moment)
- tmp storage is supposed to be managed by the native component but in
case this fails cleanup is provided:
- on job close
- on process crash
- after node crash, on restart
- available space is re-checked for every forecast call (the native
component has to check again before writing)
Note: The 1st path that has enough space is chosen on job open (job
close/reopen triggers a new search)
It is possible for state documents to be
left behind in the state index. This may be
because of bugs or uncontrollable scenarios.
In any case, those documents may take up quite
some disk space when they add up. This commit
adds a step in the expired data deletion that
is part of the daily maintenance service. The
new step searches for state documents that
do not belong to any of the current jobs and
deletes them.
Closes#30551
The part of the history template responsible for slack attachments had a
dynamic mapping configured which could lead to problems, when a string
value looking like a date was configured in the value field of an
attachment.
This commit fixes the template by setting this field always to text.
This also requires a change in the template numbering to be sure this
will be applied properly when starting watcher.
This commit removes xpack from being a meta-plugin-as-a-module.
It also fixes a couple tests which were missing task dependencies, which
failed once the gradle execution order changed.
This commit changes the default out-of-the-box configuration for the
number of shards from five to one. We think this will help address a
common problem of oversharding. For users with time-based indices that
need a different default, this can be managed with index templates. For
users with non-time-based indices that find they need to re-shard with
the split API in place they no longer need to resort only to
reindexing.
Since this has the impact of changing the default number of shards used
in REST tests, we want to ensure that we still have coverage for issues
that could arise from multiple shards. As such, we randomize (rarely)
the default number of shards in REST tests to two. This is managed via a
global index template. However, some tests check the templates that are
in the cluster state during the test. Since this template is randomly
there, we need a way for tests to skip adding the template used to set
the number of shards to two. For this we add the default_shards feature
skip. To avoid having to write our docs in a complicated way because
sometimes they might be behind one shard, and sometimes they might be
behind two shards we apply the default_shards feature skip to all docs
tests. That is, these tests will always run with the default number of
shards (one).
This commit adds the ability to specify a plugin from maven for a
test cluster to use. Currently, only local projects may be used as
plugins, except when testing bwc, where the coordinates of the project
are used. However, that assumes all projects always keep the same
coordinates, or are even still plugins, which is no longer the case for
x-pack. The full cluster and rolling restart tests are changed to use
this new method when pulling x-pack versions before 6.3.0.
Modifies the SQL tests to use the new `Request` object flavored methods
introduced onto the `RestClient` in #29623. We'd like to remove the old
methods eventually so we should stop using them.
Tweak the return data, in particular with regards for ODBC columns to
better align with the spec
Fix order for SYS TYPES and TABLES according to the JDBC/ODBC spec
Fix#30386Fix#30521
This commit adds a general state listener to the SecurityIndexManager,
and replaces the existing health and up-to-date listeners with that. It
also moves helper methods relating to health to SecurityIndexManager
from SecurityLifecycleService.
This commit moves the generated-resources directory to be within
the build directory for the openldap-tests and saml-idp-tests
projects. Both projects create a generated-resources directory that
should have been in the build directory but were instead at the same
level as the build directory.
This commit renames IndexLifecycleManager to SecurityIndexManager as it
is not actually a general purpose class, but specific to security. It
also removes indirection in code calling the lifecycle service, instead
calling the security index manager directly.
Today when processing a request for a URL path for which we can not find
a handler we send back a plain-text response. Yet, we have the accept
header in our hand and can respect the accepted media type of the
request. This commit addresses this.
This commit updates the multi cluster search test with security so that
the user that is simply performing a multi cluster search does not have
any cluster permissions. This is done as none are needed by this user
and excess privileges could mask a behavior change.
QA tests that use security need to use a trial license instead of a
basic license. Basic licenses do not enable security so these tests are
not running in the expected configuration. This can also lead to issues
that seem completely unrelated such as authentication failures right
after cluster formation.
The authentication failure right after cluster formation happens since
a request makes it past the authentication license checks and the
request starts the authentication process. During authentication the
cluster forms and the license is updated to a basic license, which
disables all realms. The request that is being authenticated then tries
to iterate over the realms, but the realms are empty and the request
cannot be authenticated. This results in a authentication failure even
though the credentials provided are correct.
Closes#30306
When dealing with filtering, a composite aggregation might return empty
buckets (which have been filtered) which gets sent as is to the client.
Unfortunately this interprets the response as no more data instead of
retrying.
This now has changed and the listener keeps retrying until either the
query has ended or data passes the filter.
Fix#30292
This commit fixes an issue with the data diagnostics were
empty buckets are not reported even though they should. Once
a job is reopened, the diagnostics do not get initialized from
the current data counts (especially the latest record timestamp).
The result is that if the data that is sent have a time gap compared
to the previous ones, that gap is not accounted for in the empty bucket
count.
This commit fixes that by initializing the diagnostics with the current
data counts.
Closes#30080
Each test now has its own watch id that is being used.
This ensures there are no old history entries, which can potentially
lead to broken test assertions.
The current implementation starts/stops watcher using an executor. This
can result in our of order operations.
This commit reduces those executor calls to an absolute minimum in order
to be able to do state changes within the cluster state listener method,
which runs in sequence.
When a state change occurs that forces the watcher service to pause
(like no watcher index, no master node, no local shards), the service is
now in a paused state.
Pausing is a super lightweight operation, which marks the
ExecutionService as paused and waits for the currently executing watches
to finish in the background via an executor. The same applies for
stopping, the potentially long running operation is outsourced in to an
executor, as waiting for executed watches is decoupled from the current
state.
The only other long running operation is starting, where watches need to
be loaded. This is also done via an executor, but has an additional
protection by checking the cluster state version it was started with. If
another cluster state version was trying to load the watches, then this
loading will not take effect.
This PR also cleans up some unused states, like the a simple boolean in
the HistoryStore/TriggeredWatchStore marking it as started or stopped,
as this can now be caught in the execution service.
Another advantage of this approach is the fact, that now only triggered
watches are not getting executed, while watches that are run via the
Execute Watch API will still be executed regardless if watcher is
stopped or not.
Lastly the TickerScheduleTriggerEngine thread now only starts on data nodes.
Many tests are added with a version check so that they do not run against a
version that doesn't have the feature yet. Master is 7.0, so all tests that
do not run against 6.0+ can be removed and the version check can be removed
on all tests that always run on 6.0+.
The elasticsearch-users utility had various messages that were
outdated or incorrect. This commit updates the output from this
command to reflect current terminology and configuration.
This commit increases the logging for authentication in the x-pack
multi-node qa test project. This is needed to assist in debugging HTTP
authorization failures in waiting for the second node in these tests.
See #30306
We had a number of awaitsFix links that weren't updated after the xpack
merge.
Where possible I changed the links to the new locations, but in some
circumstances the original ticket was closed (suggesting the awaitsfix
should be removed) or was otherwise unclear the status.
This commit moves the repository-s3 fixture test added in #29296 in a
new `repository-s3/qa/amazon-s3` project. This new project allows the
REST integration tests to be executed using the real S3 service when
all the required environment variables are provided. When no env var
is provided, then the tests are executed using the fixture added
in #29296.
The REST tests located at the `repository-s3`plugin project now only
verify that the plugin is correctly loaded.
The REST tests have been adapted to allow a bucket name and a base
path to be specified as env vars. This way it is possible to run the tests
with different base paths (could be anything, like a CI job name or a
branch name) without multiplicating buckets.
Related to #29349
This removes some monitoring tests that have been silenced for a long
time. These tests don't really provide any value for the upgrade suite and
they just create noise due to their occasional timing-related failures.
Add the oss tar distribution to the packaging test plugin. Test the oss
tar distribution in the core packaging tests, and the non-oss tar
distribution in the x-pack packaging tests.
In order to have less qa/ projects, this commit removes the
watcher-smoke-test-mustache and the watcher-smoke-test-painless projects
and moves the tests of both into the watcher-smoke-test projects.
This results in less projects to parse when calling gradle and a shorter test
time due to being able to run everything within one elasticsearch
instance.
Tests need to wait for changes to the job's established memory usage to
propagate and an over enthusiastic optimisation meant jobs were updated
from stale state causing recent change to be lost.
This commit makes x-pack a module and adds it to the default
distrubtion. It also creates distributions for zip, tar, deb and rpm
which contain only oss code.
Dimitris Athanasiou
Martijn van Groningen
Boaz Leskes
David Kyle
Costin Leau
Armin Braun
Andrei Stefan
Konrad Beiske
Ioannis Kakavas
Igor Motov
Alpar Torok
Nik Everett
Tanguy Leroux
Ryan Ernst
Luca Cavanna
Tom Veasey
Jason Tedor
Hendrik Muhs
David Roberts
Christoph Büscher
Andy Bristol
Adrien Grand
Tim Vernum
Alexander Reelsen
Yannick Welsch
Jay Modi
jaymode
David Turner
Zachary Tong
Chris Earle
Michael Basnight