* Revert "Fix the Text class package change in example plugins (#128316)"
This reverts commit cc486480e3.
* Revert "Update Text class to use native java ByteBuffer (#127666)"
This reverts commit db0c3c7a28.
In the case that the mark is at the very end of the stream,
the slice's stream could already be closed by the time we
call reset(), and so the stream must be reopened.
Today, with `FROM`, we can partition a shard into multiple slices,
allowing multiple drivers to execute against a single shard.
For time-series, specifically rate aggregation, the data must arrive in
order; therefore, we currently use a single driver. With this change, we
split the pipeline into two parts—the time-series source and the
rest—allowing two drivers to run concurrently.
It turns out in #128234 that Lucene pushdown on ST_DISTANCE with invalid points (latitude out of range) will cause all documents to be returned, since DISJOINT on an invalid circle is true for all documents. We could either add an extra check for spatial pushdown that the geometries are valid, or add validation at geometry creation. This second option is much easier to implement, and a more comprehensive approach, as it prevents invalid geometries in many more places, hopefully reducing the likelihood of subtle and obscure bugs like #128234 happening in future.
When using synthetic source, runtime fields data may come from doc values. Doc values iterators can only be read once, and in increasing docId order. But if a runtime field is referenced multiple times in a query, currently the same doc value iterator will be used. This causes an error, as the second field reference will attempt to read the same iterator from a lower docId than was previously used. The fix is to create a new source loader, and thus a new doc value iterator, if the requested docId is lower than the last seen docId.
* Use walkFileTree for extractModuleNameFromDirectory
* More tests in TestBuildInfoPluginFuncTest
* Remove stray line from debugging
* [CI] Auto commit changes from spotless
* Eliminate List.reversed() call
---------
Co-authored-by: elasticsearchmachine <infra-root+elasticsearchmachine@elastic.co>
* remove Enterprise Search service account + tests
* [CI] Auto commit changes from spotless
* remove all tests referencing ent-search
* [CI] Auto commit changes from spotless
* ensure yaml tests have correct length
* set proper length expected
* use index for slow log test that works
* skip account REST tests with invalid length
* skip Test service account tokens test too
* remove added space
* add changelog for PR
* update
---------
Co-authored-by: elasticsearchmachine <infra-root+elasticsearchmachine@elastic.co>
Classes like UnusedStatsRemover delete orphaned documents without an associated job. When the indices are made read-only it will start failing as read-only means no delete.
This PR ensures that the non-writable indices are not included in the delete-by-query requests.
Creates a `ROUND_TO` function that rounds it's input to one of the
provided values. Like so:
```
ROUND_TO(v, 0, 5000, 10000, 20000, 40000, 100000)
v | ROUND_TO
0 | 0
100 | 0
6000 | 5000
45001 | 40000
999999 | 100000
```
For some sequences of numbers you could do this with the `/` operator -
but for arbitrary sequences of numbers you needed `CASE` which is quite
slow. And hard to read!
Rewriting the example above would look like:
```
CASE (
v < 5000, 0,
v < 10000, 5000,
v < 20000, 10000,
v < 40000, 20000,
v < 100000, 40000,
100000
)
```
Even better, this is *fast*:
```
(operation) Mode Cnt Score Error Units
round_to_4_via_case avgt 7 138.124 ± 0.738 ns/op
round_to_4 avgt 7 0.805 ± 0.011 ns/op
round_to_3 avgt 7 0.739 ± 0.011 ns/op
round_to_2 avgt 7 0.651 ± 0.009 ns/op
date_trunc avgt 7 2.425 ± 0.018 ns/op
```
I've included a comparison to `DATE_TRUNC` above because we should be
able to rewrite `DATE_TRUNC` into `ROUND_TO` when we know the date range
of the index. This doesn't do it now, but it should be possible.
Currently, the DiskThresholdMonitor code considers shutdown metadata to identify nodes that are being replaced. If the node-to-be-replaced (source) leaves the cluster before the corresponding shutdown metadata is removed from the cluster state, we can have a NPE. This PR adds a test for that and improves a bit code to handle node ids for source and target from shutdown metadata.
Fixes#100201
Our APMTracer doesn't like nulls - this is a sensible thing, as APM in general does not allow nulls (it only allows a precise set of types).
This PR changes the attribute to a sentinel "" in place of null values. It also makes a small change to APMTracer to give a better error message in case of null values in attributes.
This test is expected to complete a deeply abusive lookup join without
circuit breaking. It was sometimes circuit breaking. This lowers the
data we feed to the test so its slightly less abusive. It's still plenty
abusive.
Closes#127365
Patrick Doyle