This test was failing in #34004 due to a race, and although #34296 made
the failures rarer they did not actually fix the race. Then in #99201 we
fixed the race but the resulting test over-synchronizes and no longer
meaningfully verifies the concurrent behaviour we were originally trying
to check. It also fails for other reasons. This commit reverts back to
the original test showing that we might run the action at most once
after cancellation without any further synchronization, but fixes the
assertion to use the value of the counter observed immediately after the
cancellation since we cannot be sure that no extra iterations execute
before the cancellation completes.
This commit adjusts the behavior of the node shutdown status
calculation to avoid reporting a `STALLED` status when ILM is
in the process of doing something that might result in the
migration being temporarily stalled (at time of writing, that
means shrinking an index).
This check is only made if ILM is not stopped.
Backport of #98367
In Java 21 List now extends SequencedCollection, instead of Collection
directly. When resolving methods Painless starts at the defined type,
and iterates up through super classes and interfaces. Unfortunately if a
superinterface was not known, as it is for SequencedCollection since it
is not in the allowed list of classes, method resolution would give up.
This commit adjusts the superinterface interation to continue traversing
until the method is found or no more superinterfaces are found.
fixes#97022
(cherry picked from commit 16b45575c3)
# Conflicts:
# gradle.properties
Co-authored-by: Ryan Ernst <ryan@iernst.net>
RestHandler has a number of methods that affect the behaviour of request
processing. If the handler is wrapped (e.g. SecurityRestFilter or
DeprecationRestHandler) then these methods must be delegated to the
underlying handler.
This commit introduces a new abstract base class `FilterRestHandler`
that correctly delegates these methods so that wrappers (subclasses) do
not need to implement the behaviour on a case-by-case basis
Backport of: #98861
Prior to this change NodeReplacementAllocationDecider was unconditionally skipping both replacement source and target nodes when calculation auto-expand replicas. This is fixed by autoexpanding to the replacement node if source node already had shards of the index
Backport of PR #96281 amended for 7.17.x
Closes#89527
Co-authored-by: Ievgen Degtiarenko <ievgen.degtiarenko@elastic.co>
This is a backport of multiple work items related to authentication enhancements for HTTP,
which were originally merged in the 8.8 - 8.9 releases.
Hence, the HTTP (only the netty4-based implementation (default), not the NIO one) authentication
implementation gets a throughput boost (especially for requests failing authn).
Relates to: ES-6188 #92220#95112
The docs for this API say the following:
> If the API fails, you can safely retry it. Only a successful response
> guarantees that the node has been removed from the voting
> configuration and will not be reinstated.
Unfortunately this isn't true today: if the request adds no exclusions
then we do not wait before responding. This commit makes the API wait
until all exclusions are really applied.
Backport of #98386, plus the test changes from #98146 and #98356.
This change avoids unnecessary substring allocations and recursion calls
when more than two consecutive wildcards (`*`) are detected. Instead
skipping and calling a method recursively, we now try to skip all
consecutive `*` chars at once.
* Update certs for PKI tests and re-enable tests (#97766)
The certs for the PKI tests expired and the test was muted.
This commit follows the instructions in the read to update the certs and unmutes the test.
The certs will now expire 20 years from now.
fixes: #97756
* precommit
in 2.17.2 (patch release) log4j has made a refactoring that requires a Configuration to be manually passed into the created PatternLayout
If the Configuration is not passed, the System Variable lookup will not work This results in cluster.name field not being populated in logs
This commit creates a PatternLayout with a DefaultConfiguration (the same was used previous to the refactoring)
backports #97679
* Migrate to data tiers routing configures correct default for mounted indices
(cherry picked from commit eaec9602a9c4a4b092f36fd35f58d1e4b8faca52)
Signed-off-by: Andrei Dan <andrei.dan@elastic.co>
* Add isExplicitDataTier method back as it's used in 7.17
---------
Co-authored-by: Elastic Machine <elasticmachine@users.noreply.github.com>
The current recursive nested field handling implementation in FieldFetcher
can be O(n^2) in the number of nested mappings, whether or not a nested
field has been requested or not. For indexes with a very large number of
nested fields, this can mean it takes multiple seconds to build a FieldFetcher,
making the fetch phase of queries extremely slow, even if no nested fields
are actually asked for.
This commit reworks the logic so that building nested fetchers is only
O(n log n) in the number of nested mappers; additionally, we only pay this
cost for nested fields that have been requested.
* Port lucene fix github-12352 to Elasticsearch 7.17
* Update docs/changelog/96721.yaml
* Support for Byte and Short as vector tiles value tag (#97619)
Byte and short data types are not supported by the vector tiles specification. If a user tries to add one of those, it
actually gets ignored. This add s support to those values by casting them as integers. In addition it will throw an
exception if the value added is a type it cannot understand so it is not silently ignored.
# Conflicts:
# x-pack/plugin/src/yamlRestTest/resources/rest-api-spec/test/vector-tile/10_basic.yml
Today we document that tasks may not react to cancellations immediately,
but in practice it's surprising to users and kind of a bug if they run
for too long after being cancelled. This commit adds a little extra
detail about the information to collect to troubleshoot such a
situation.
