Today throughout the codebase, catch throwable is used with reckless
abandon. This is dangerous because the throwable could be a fatal
virtual machine error resulting from an internal error in the JVM, or an
out of memory error or a stack overflow error that leaves the virtual
machine in an unstable and unpredictable state. This commit removes
catch throwable from the codebase and removes the temptation to use it
by modifying listener APIs to receive instances of Exception instead of
the top-level Throwable.
Relates #19231
Rename `fields` to `stored_fields` and add `docvalue_fields`
`stored_fields` parameter will no longer try to retrieve fields from the _source but will only return stored fields.
`fields` will throw an exception if the user uses it.
Add `docvalue_fields` as an adjunct to `fielddata_fields` which is deprecated. `docvalue_fields` will try to load the value from the docvalue and fallback to fielddata cache if docvalues are not enabled on that field.
Closes#18943
We have long worked to capture different partitioning scenarios in our testing infra. This PR adds a new variant, inspired by the Jepsen blogs, which was forgotten far - namely a partition where one node can still see and be seen by all other nodes. It also updates the resiliency page to better reflect all the work that was done in this area.
Update-By-Query and Delete-By-Query use internal versioning to update/delete documents. But documents can have a version number equal to zero using the external versioning... making the UBQ/DBQ request fail because zero is not a valid version number and they only support internal versioning for now. Sequence numbers might help to solve this issue in the future.
Previously all rest handlers would take Client in their injected ctor.
However, it was only to hold the client around for runtime. Instead,
this can be done just once in the HttpService which handles rest
requests, and passed along through the handleRequest method. It also
should always be a NodeClient, and other types of Clients (eg a
TransportClient) would not work anyways (and some handlers can be
simplified in follow ups like reindex by taking NodeClient).
`RestHandler`s are highly tied to actions so registering them in the
same place makes sense.
Removes the need to for plugins to check if they are in transport client
mode before registering a RestHandler - `getRestHandlers` isn't called
at all in transport client mode.
This caused guice to throw a massive fit about the circular dependency
between NodeClient and the allocation deciders. I broke the circular
dependency by registering the actions map with the node client after
instantiation.
Stored scripts are pulled from the cluster state, and the current api
requires passing the ClusterState on each call to compile. However, this
means every user of the ScriptService needs to depend on the
ClusterService. Instead, this change makes the ScriptService a
ClusterStateListener. It also simplifies tests a lot, as they no longer
need to create fake cluster states (except when testing stored scripts).
Instead of implementing onModule(ActionModule) to register actions,
this has plugins implement ActionPlugin to declare actions. This is
yet another step in cleaning up the plugin infrastructure.
While I was in there I switched AutoCreateIndex and DestructiveOperations
to be eagerly constructed which makes them easier to use when
de-guice-ing the code base.
`stored_fields` parameter will no longer try to retrieve fields from the _source but will only return stored fields.
`fields` will throw an exception if the user uses it.
Add `docvalue_fields` as an adjunct to `fielddata_fields` which is deprecated. `docvalue_fields` will try to load the value from the docvalue and fallback to fielddata cache if docvalues are not enabled on that field.
Closes#18943
This makes this sequence:
```
curl -XDELETE localhost:9200/source,dest?pretty
for i in $( seq 1 100 ); do
curl -XPOST localhost:9200/source/test -d'{"test": "test"}'; echo
done
curl localhost:9200/_refresh?pretty
curl -XPOST 'localhost:9200/_reindex?pretty&wait_for_completion=false' -d'{
"source": {
"index": "source"
},
"dest": {
"index": "dest"
}
}'
curl 'localhost:9200/_tasks/Jsyd6d9wSRW-O-NiiKbPcQ:237?wait_for_completion&pretty'
```
Return task *AND* the response to the user.
This also renames "result" to "response" in the persisted task info
to line it up with how we name the objects in Elasticsearch.
This change removes some unnecessary dependencies from ClusterService
and cleans up ClusterName creation. ClusterService is now not created
by guice anymore.
In 2.0 we added plugin descriptors which require defining a name and
description for the plugin. However, we still have name() and
description() which must be overriden from the Plugin class. This still
exists for classpath plugins. But classpath plugins are mainly for
tests, and even then, referring to classpath plugins with their class is
a better idea. This change removes name() and description(), replacing
the name for classpath plugins with the full class name.
This adds a get task API that supports GET /_tasks/${taskId} and
removes that responsibility from the list tasks API. The get task
API supports wait_for_complation just as the list tasks API does
but doesn't support any of the list task API's filters. In exchange,
it supports falling back to the .results index when the task isn't
running any more. Like any good GET API it 404s when it doesn't
find the task.
Then we change reindex, update-by-query, and delete-by-query to
persist the task result when wait_for_completion=false. The leads
to the neat behavior that, once you start a reindex with
wait_for_completion=false, you can fetch the result of the task by
using the get task API and see the result when it has finished.
Also rename the .results index to .tasks.
Writeable is better for immutable objects like TimeValue.
Switch to writeZLong which takes up less space than the original
writeLong in the majority of cases. Since we expect negative
TimeValues we shouldn't use
writeVLong.
Today we use a random source of UUIDs for assigning allocation IDs,
cluster IDs, etc. Yet, the source of randomness for this is not
reproducible in tests. Since allocation IDs end up as keys in hash maps,
this means allocation decisions and not reproducible in tests and this
leads to non-reproducible test failures. This commit modifies the
behavior of random UUIDs so that they are reproducible under tests. The
behavior for production code is not changed, we still use a true source
of secure randomness but under tests we just use a reproducible source
of non-secure randomness.
It is important to note that there is a test,
UUIDTests#testThreadedRandomUUID that relies on the UUIDs being truly
random. Thus, we have to modify the setup for this test to use a true
source of randomness. Thus, this is one test that will never be
reproducible but it is intentionally so.
Relates #18808
Folded grok processor into ingest-common module.
The rest tests have been moved to ingest-common module as well, because these tests don't run in the rest-api-spec module but in the distribution:integ-test-zip module
and adding a test plugin there felt just wrong to me. I think this is ok. I left a tiny ingest rest test behind in that tests with an empty pipeline.
Removed messy tests, these tests were already covered in the rest tests
Added ingest test plugin in test infra so that each module testing integration with ingest doesn't need write its own plugin
Moved reindex ingest tests to qa module
Closes#18490
This commit refactors the handling of thread pool settings so that the
individual settings can be registered rather than registering the top
level group. With this refactoring, individual plugins must now register
their own settings for custom thread pools that they need, but a
dedicated API is provided for this in the thread pool module. This
commit also renames the prefix on the thread pool settings from
"threadpool" to "thread_pool". This enables a hard break on the settings
so that:
- some of the settings can be given more sensible names (e.g., the max
number of threads in a scaling thread pool is now named "max" instead
of "size")
- change the soft limit on the number of threads in the bulk and
indexing thread pools to a hard limit
- the settings names for custom plugins for thread pools can be
prefixed (e.g., "xpack.watcher.thread_pool.size")
- remove dynamic thread pool settings
Relates #18674
The assertBusy method currently has both a Runnable and Callable
version. This has caused confusion with type inference and lambdas
sometimes, in particular with java 9. This change removes the callable
version as nothing was actually using it.
The retry test has failed a couple of times in CI because it wasn't able
to cause any retries. Putting it in a bash `while` loop shows that it
eventually does fail that way. The seed "4F6477A9C999CA20" seems especially
good at failing to get retries. It doesn't fail all the time, but more
than most.
This adds a retry to each test case, retrying a maximum of 10 times or
until it causes the retries. I've seen it fail to get retries 7 times
in a row but not go beyond that. Retrying doesn't seem to really hurt
the test runtime all that much. Most of the time is in the startup
cost.
Failing CI build that triggered this:
https://elasticsearch-ci.elastic.co/job/elastic+elasticsearch+master+periodic/852/console
This uses the same backoff policy we use for bulk and just retries until
the request isn't rejected.
Instead of `{"retries": 12}` in the response to count retries this now
looks like `{"retries": {"bulk": 12", "search": 1}`.
Closes#18059
QueryBuilder has generics, but those are never used: all call sites use
`QueryBuilder<?>`. Only `AbstractQueryBuilder` needs generics so that the base
class can contain a default implementation for setters that returns `this`.
All other values are errors.
Add java test for throttling. We had a REST test but it only ran against
one node so it didn't catch serialization errors.
Add Simple round trip test for rethrottle request
Do this by creating a Client subclass that automatically assigns the
parentTask to all requests that come through it. Code that doesn't want
to set the parentTask can call `unwrap` on the Client to get the inner
client instance that doesn't set the parentTask. Reindex uses this for
its ClearScrollRequest so that the request will run properly after the
reindex request has been canceled.
Now that the current uses of magical camelCase support have been
deprecated, we can remove these in master (sans remaining issues like
BulkRequest). This change removes camel case support from ParseField,
query types, analysis, and settings lookup.
see #8988
Callers should explicitly handle parents - either using EMPTY_TASK_ID when
a parent isn't possible or piping parents from the TransportRequest when
possible.
This change cleans up a few things in QueryParseContext and QueryShardContext
that make it hard to reason about the state of these context objects and are
thus error prone and should be simplified.
Currently the parser that used in QueryParseContext can be set and reset any
time from the outside, which makes reasoning about it hard. This change makes
the parser a mandatory constructor argument removes ability to later set a
different ParseFieldMatcher. If none is provided at construction time, the
one set inside the parser is used. If a ParseFieldMatcher is specified at
construction time, it is implicitely set on the parser that is beeing used.
The ParseFieldMatcher is only kept inside the parser.
Also currently the QueryShardContext historically holds an inner QueryParseContext
(in the super class QueryRewriteContext), that is mainly used to hold the parser
and parseFieldMatcher. For that reason, the parser can also be reset, which leads
to the same problems as above. This change removes the QueryParseContext from
QueryRewriteContext and removes the ability to reset or retrieve it from the
QueryShardContext. Instead, `QueryRewriteContext#newParseContext(parser)` can be
used to create new parse contexts with the given parser from a shard context
when needed.
Currently we are able to set a ParseFieldMatcher on XContentParsers,
mainly to conveniently carry it around to be available where the
actual parsing happens. This was just recently introduced together
with ObjectParser so that ObjectParser can make use of deprecation
logging and throwing errors while parsing.
This however is trappy because we create parsers in so many places in
the code and it is easy to forget setting the right ParseFieldMatcher.
Instead we should hold the ParseFieldMatcher only in the parse contexts
(e.g. QueryParseContext).
This PR removes the ParseFieldMatcher from XContentParser. ObjectParser
can still make use of it because we can make the otherwise unbounded
`context` type to extend an interface that makes sure contexts used in
ObjectParser can supply a ParseFieldMatcher. Contexts in ObjectParser
are now no longer optional, but it is sufficient to pass in a small
lambda expression in places where no other context is available.
Relates to #17417
When we pass down both parser and QueryParseContext to a method, we cannot
make sure that the parser contained in the context and the parser that is
parsed as an argument have the same state. This removes the parser argument
from methods where we currently have both the parser and the parse context
as arguments and instead retrieves the parse from the context inside the
method.
* Inner hits can now only be provided and prepared via setter in the nested, has_child and has_parent query.
* Also made `score_mode` a required constructor parameter.
* Moved has_child's min_child/max_children validation from doToQuery(...) to a setter.
BulkByScrollTaskTest#testDelayAndRethrottle was getting rejected exceptions
every once in a while. This was reproducible ~20% of the time for me. I
added a CyclicBarrier to prevent the test from shutting down the thread pool
before the threads get finished.
This allows the user to update the reindex throttle on the fly, with changes
that speed up the throttling being applied immediately and changes that
slow down the throttling being applied during the next batch. This means
that if a user throttles reindex in such a way that it tries to sleep for
16 years and then realizes that they've done something wrong then they
can change the throttle and reindex will wake up again. We don't apply
slow downs immediately so we never get in danger of losing the scan context.
Also, if reindex is canceled while it is sleeping (how it honor throttling)
then it'll immediately wake up and cancel itself.
`text` fields will have fielddata disabled by default. Fielddata can still be
enabled on an existing index by setting `fielddata=true` in the mappings.
readFrom is confusing because it requires an instance of the type that it
is reading but it doesn't modify it. But we also have (deprecated) methods
named readFrom that *do* modify the instance. The "right" way to implement
the non-modifying readFrom is to delegate to a constructor that takes a
StreamInput so that the read object can be immutable. Now that we have
`@FunctionalInterface`s it is fairly easy to register things by referring
directly to the constructor.
This change modifying NamedWriteableRegistry so that it does that. It keeps
supporting `registerPrototype` which registers objects to be read by
readFrom but deprecates it and delegates it to a new `register` method
that allows passing a simple functional interface. It also cuts Task.Status
subclasses over to using that method.
The start of #17085
The test was checking that we'd set the headers properly but in some cases
the request had yet to come in because it was running on another thread.
Now we wait for the headers to show up before failing the test.
Closes#17299
If the user asks for a refresh but their reindex or update-by-query
operation touched no indexes we should just skip the resfresh call
entirely. Without this commit we refresh *all* indexes which is totally
wrong.
Closes#17296
We current have a ClusterService interface, implemented by InternalClusterService and a couple of test classes. Since the decoupling of the transport service and the cluster service, one can construct a ClusterService fairly easily, so we don't need this extra indirection.
Closes#17183
Without this commit fetching the status of a reindex from a node that isn't
coordinating the reindex will fail. This commit properly registers reindex's
status so this doesn't happen. To do so it moves all task status registration
into NetworkModule and creates a method to register other statuses which the
reindex plugin calls.
Today index names are often resolved lazily, only when they are really
needed. This can be problematic especially when it gets to mapping updates
etc. when a node sends a mapping update to the master but while the request
is in-flight the index changes for whatever reason we would still apply the update
since we use the name of the index to identify the index in the clusterstate.
The problem is that index names can be reused which happens in practice and sometimes
even in a automated way rendering this problem as realistic.
In this change we resolve the index including it's UUID as early as possible in places
where changes to the clusterstate are possible. For instance mapping updates on a node use a
concrete index rather than it's name and the master will fail the mapping update iff
the index can't be found by it's <name, uuid> tuple.
Closes#17048
The refresh tests were failing rarely due to refreshes happening
automatically on indexes with -1 refresh intervals. This commit moves
the refresh test into a unit test where we can check if it was attempted
so we never get false failures from background refreshes.
It also stopped refresh from being run if the reindex request was canceled.
Indexing failures have caused the reindex http request to fail for a while
now. Both search and indexing failures cause it to abort. But search
failures didn't cause a non-200 response code from the http api. This
fixes that.
Also slips in a fix to some infrequently failing rest tests.
Closes#16037
Implements CompositeIndicesRequest on UpdateByQueryRequest and
ReindexRequest so that plugins can reason about the request. In both cases
this implementation is imperfect but useful because instead of listing
all requests that make up the request it instead attempts to make dummy
requests that represent the requests that it will later make.
_wait_for_completion defaults to false. If set to true then the API will
wait for all the tasks that it finds to stop running before returning. You
can use the timeout parameter to prevent it from waiting forever. If you
don't set a timeout parameter it'll default to 30 seconds.
Also adds a log message to rest tests if any tasks overrun the test. This
is just a log (instead of failing the test) because lots of tasks are run
by the cluster on its own and they shouldn't cause the test to fail. Things
like fetching disk usage from the other nodes, for example.
Switches the request to getter/setter style methods as we're going that
way in the Elasticsearch code base. Reindex is all getter/setter style.
Closes#16906
Jason Tedor
Jim Ferenczi
Ryan Ernst
Boaz Leskes
Tanguy Leroux
Nik Everett
Yannick Welsch
Alexander Reelsen
Simon Willnauer
Simon Willnauer
Martijn van Groningen
Adrien Grand
Isabel Drost-Fromm
Christoph Büscher
Clinton Gormley
Ali Beyad