Currently we read in 64KB blocks from the network. When TLS is not
enabled, these bytes are normally passed all the way to the application
layer (some exceptions: compression). For the HTTP layer this means that
these bytes can live throughout the entire lifecycle of an indexing
request.
The problem is that if the reads from the socket are small, this means
that 64KB buffers can be consumed by 1KB or smaller reads. If the socket
buffer or TCP buffer sizes are small, the leads to massive memory
waste. It has been identified as a major source of OOMs on coordinating
nodes as Elasticsearch easily exhausts the heap for these network bytes.
This commit resolves the problem by placing a handler after the TLS
handler to copy these bytes to a more appropriate buffer size as
necessary. This comes after TLS, because TLS is a framing layer which
often resolves this problem for us (the 64KB buffer will be decoded
into a more appropriate buffer size). However, this extra handler will
solve it for the non-TLS pipelines.
Currently Netty will batch compression an entire HTTP response
regardless of its content size. It allocates a byte array at least of
the same size as the uncompressed content. This causes issues with our
attempts to remove humungous G1GC allocations. This commit resolves the
issue by split responses into 128KB chunks.
This has the side-effect of making large outbound HTTP responses that
are compressed be send as chunked transfer-encoding.
Currently we duplicate our specialized cors logic in all transport
plugins. This is unnecessary as it could be implemented in a single
place. This commit moves the logic to server. Additionally it fixes a
but where we are incorrectly closing http channels on early Cors
responses.
Currently we log the NettyAllocator description when the netty plugin is
created. Unfortunately, this hits certain static fields in Netty which
triggers the settings of the number of CPU processors. This conflicts
with out Elasticsearch behavior to override this based on a setting.
This commit resolves the issue by logging after the processors have been
set.
Currently the netty pool chunk size defaults to 16MB. The number does
not play well with the G1GC which causes this to consume entire regions.
Additionally, we normally allocated arrays of size 64KB or less. This
means that Elasticsearch could handle a smaller pool chunk size to play
nicer with the G1GC.
* 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
In #60297 we added some tests related to logging from the transport
layer, but these tests failed occasionally since the cluster
was kept alive between test invocations but the logging framework
expected it only to be used for a single test. With this commit we
reduce the scope of the internal test cluster to `TEST` to solve this
problem.
Closes#60321.
Transport connections between nodes remain in place until one or other
node shuts down or the connection is disrupted by a flaky network.
Today it is very difficult to demonstrate that transient failures and
cluster instability are caused by the network even though this is often
the case. In particular, transport connections open and close without
logging anything, even at `DEBUG` level, making it very hard to quantify
the scale of the problem or to correlate the networking problems with
external events.
This commit adds the missing `DEBUG`-level logging when transport
connections open and close, and also tracks the total number of
transport connections a node has opened as a measure of the stability of
the underlying network.
Introduce a javaRestTest source set and task to compliment the yamlRestTest.
javaRestTest differs such that the code is sourced from Java and may have
different dependencies and setup requirements for the test clusters. This also
allows the tests to run in parallel in different cluster instances to prevent any
cross test contamination between the two types of tests.
Included in this PR is all :modules no longer use the integTest task. The tests
are now driven by test, yamlRestTest, javaRestTest, and internalClusterTest.
Since only :modules (and :rest-api-spec) have been converted to yamlRestTest
we can now disable the integTest task if either yamlRestTest or javaRestTest have
been applied. Once all projects are converted, we can delete the integTest task.
related: #56841
related: #59444
keepalives tell any intermediate devices that the connection remains alive, which helps with overzealous firewalls that are
killing idle connections. keepalives are enabled by default in Elasticsearch, but use system defaults for their
configuration, which often times do not have reasonable defaults (e.g. 7200s for TCP_KEEP_IDLE) in the context of
distributed systems such as Elasticsearch.
This PR sets the socket-level keep_alive options for network.tcp.{keep_idle,keep_interval} to 5 minutes on configurations
that support it (>= Java 11 & (MacOS || Linux)) and where the system defaults are set to something higher than 5
minutes. This helps keep the connections alive while not interfering with system defaults or user-specified settings
unless they are deemed to be set too high by providing better out-of-the-box defaults.
This commit moves the modules REST tests to the
newly introduced yamlRestTest source set. A few
tests have also been re-named to include the correct
IT suffix. Without changing the names, the testing
conventions task would fail since now that the YAML
tests are no longer present pacify the convention.
These tests have moved to the internalClusterTest
source set.
related: #56841
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)
Currently we are leaving the settings to default port range in the nio
and netty4 http server test. This has recently led to tests failing due
to what appears to be a port conflict with other processes. This commit
modifies these tests to use the test case helper method to generate port
ranges.
Fixes#58433 and #58296.
* 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
Netty4HttpServerTransportTests has started to fail intermittently. It
seems like unexpected successful responses are being received when the
test is simulating errors. This commit adds logging to the test to
provide additional information when there is an unexpected success. It
also adds the logging to the nio http test.
Adds assertions to Netty to make sure that its threads are not polluted by thread contexts (and
also that thread contexts are not leaked). Moves the ClusterApplierService to use the system
context (same as we do for MasterService), which allows to remove a hack from
TemplateUgradeService and makes it clearer that applying CS updates is fully executing under
system context.
Adds assertions to Netty to make sure that its threads are not polluted by thread contexts (and
also that thread contexts are not leaked). Moves the ClusterApplierService to use the system
context (same as we do for MasterService), which allows to remove a hack from
TemplateUgradeService and makes it clearer that applying CS updates is fully executing under
system context.
Follow up to #56961:
We can be a little more efficient than just serializing at the IO loop by serializing
only when we flush to a channel. This has the advantage that we don't serialize a long
queue of messages for a channel that isn't writable for a longer period of time (unstable network,
actually writing large volumes of data, etc.).
Also, this further reduces the time for which we hold on to the write buffer for a message,
making allocations because of an empty page cache recycler pool less likely.
Almost every outbound message is serialized to buffers of 16k pagesize.
We were serializing these messages off the IO loop (and retaining the concrete message
instance as well) and would then enqueue it on the IO loop to be dealt with as soon as the
channel is ready.
1. This would cause buffers to be held onto for longer than necessary, causing less reuse on average.
2. If a channel was slow for some reason, not only would concrete message instances queue up for it, but also 16k of buffers would be reserved for each message until it would be written+flushed physically.
With this change, the serialization happens on the event loop which effectively limits the number of buffers that `N` IO-threads will ever use so long as messages are small and channels writable.
Also, this change dereferences the reference to the concrete outbound message as soon as it has been serialized to save some more on GC.
This reduces the GC time for a default PMC run by about 50% in experiments (3 nodes, 2G heap each, loopback ... obvious caveat is that GC isn't that heavy in the first place with recent changes but still a measurable gain).
I also expect it to be helpful for master node stability by causing less of a spike if master is e.g. hit by a large number of requests that are processed batched (e.g. shard snapshot status updates) and responded to in a short time frame all at once.
Obviously, the downside to this change is that it introduces more latency on the IO loop for the serialization. But since we read all of these messages on the IO loop as well I don't see it as much of a qualitative change really and the more predictable buffer use seems much more valuable relatively.
We don't need to hold on to the request body past the beginning of sending
the response. There is no need to keep a reference to it until after the response
has been sent fully and we can eagerly release it here.
Note, this can be optimized further to release the contents even earlier but for now
this is an easy increment to saving some memory on the IO pool.
Elasticsearch requires that a HttpRequest abstraction be implemented
by http modules before server processing. This abstraction controls when
underlying resources are released. This commit moves this abstraction to
be created immediately after content aggregation. This change will
enable follow-up work including moving Cors logic into the server
package and tracking bytes as they are aggregated from the network
level.
In most cases we are seeing a `PooledHeapByteBuf` here now. No need to
redundantly create an new `ByteBuffer` and single element array for it
here when we can just directly unwrap its internal `byte[]`.
We never do any file IO or other blocking work on the transport threads
so no tangible benefit can be derived from using more threads than CPUs
for IO.
There are however significant downsides to using more threads than necessary
with Netty in particular. Since we use the default setting for
`io.netty.allocator.useCacheForAllThreads` which is `true` we end up
using up to `16MB` of thread local buffer cache for each transport thread.
Meaning we potentially waste CPUs * 16MB of heap for unnecessary IO threads in addition to obvious inefficiencies of artificially adding extra context switches.
Currently Elasticsearch creates independent event loop groups for each
transport (http and internal) transport type. This is unnecessary and
can lead to contention when different threads access shared resources
(ex: allocators). This commit moves to a model where, by default, the
event loops are shared between the transports. The previous behavior can
be attained by specifically setting the http worker count.
Currently there is a clear mechanism to stub sending a request through
the transport. However, this is limited to testing exceptions on the
sender side. This commit reworks our transport related testing
infrastructure to allow stubbing request handling on the receiving side.
This is a first cut at giving NodeInfo the ability to carry a flexible
list of heterogeneous info responses. The trick is to be able to
serialize and deserialize an arbitrary list of blocks of information. It
is convenient to be able to deserialize into usable Java objects so that
we can aggregate nodes stats for the cluster stats endpoint.
In order to provide a little bit of clarity about which objects can and
can't be used as info blocks, I've introduced a new interface called
"ReportingService."
I have removed the hard-coded getters (e.g., getOs()) in favor of a
flexible method that can return heterogeneous kinds of info blocks
(e.g., getInfo(OsInfo.class)). Taking a class as an argument removes the
need to cast in the client code.
The use of available processors, the terminology, and the settings
around it have evolved over time. This commit cleans up some places in
the codes and in the docs to adjust to the current terminology.
This commit moves the action name validation and circuit breaking into
the InboundAggregator. This work is valuable because it lays the
groundwork for incrementally circuit breaking as data is received.
This PR includes the follow behavioral change:
Handshakes contribute to circuit breaking, but cannot be broken. They
currently do not contribute nor are they broken.
* Refactor nodes stats request builders to match requests (#54363)
* Remove hard-coded setters from NodesInfoRequestBuilder
* Remove hard-coded setters from NodesStatsRequest
* Use static imports to reduce clutter
* Remove uses of old info APIs
This is a simple naming change PR, to fix the fact that "metadata" is a
single English word, and for too long we have not followed general
naming conventions for it. We are also not consistent about it, for
example, METADATA instead of META_DATA if we were trying to be
consistent with MetaData (although METADATA is correct when considered
in the context of "metadata"). This was a simple find and replace across
the code base, only taking a few minutes to fix this naming issue
forever.
Currently all of our transport protocol decoding and aggregation occurs
in the individual transport modules. This means that each implementation
(test, netty, nio) must implement this logic. Additionally, it means
that the entire message has been read from the network before the server
package receives it.
This commit creates a pipeline in server which can be passed arbitrary
bytes to handle. Internally, the pipeline will decode, decompress, and
aggregate the messages. Additionally, this allows us to run many
megabytes of bytes through the pipeline in tests to ensure that the
logic works.
This work will enable future work:
Circuit breaking or backoff logic based on message type and byte
in the content aggregator.
Sharing bytes with the application layer using the ref counted
releasable network bytes.
Improved network monitoring based specifically on channels.
Finally, this fixes the bug where we do not circuit break on the correct
message size when compression is enabled.
Elasticsearch has a number of different BytesReference implementations.
These implementations can all implement the interface in different ways
with subtly different behavior and performance characteristics. On the
other-hand, the JVM only represents bytes as an array or a direct byte
buffer. This commit deletes the specialized Netty implementations and
moves to using a generic ByteBuffer reference type. This will allow us
to focus on standardizing performance and behave around a smaller number
of implementations that can be used by all components in Elasticsearch.
Backport of #52542.
This commit is part of issue #40366 to remove disabled Xlint warnings
from gradle files. In particular, it removes the Xlint exclusions from
the following files:
- benchmarks/build.gradle
- client/client-benchmark-noop-api-plugin/build.gradle
- x-pack/qa/rolling-upgrade/build.gradle
- x-pack/qa/third-party/active-directory/build.gradle
- modules/transport-netty4/build.gradle
For the first three files no code adjustments were needed. For
x-pack/qa/third-party/active-directory move the suppression at the code
level. For transport-netty4 replace the variable arguments with
ArrayLists and remove any redundant casts.
Now that the FIPS 140 security provider is simply a test dependency
we don't need the thirdPartyAudit exceptions, but plugin-cli and
transport-netty4 do need jarHell disabled as they use the non fips
BouncyCastle security provider as a test dependency too.
rest-api-spec/test/10_basic.yml would check that transport_types is
`netty4` but we run FIPS 140 tests with default distribution and
transport_types is `security4`
This change changes the way to run our test suites in
JVMs configured in FIPS 140 approved mode. It does so by:
- Configuring any given runtime Java in FIPS mode with the bundled
policy and security properties files, setting the system
properties java.security.properties and java.security.policy
with the == operator that overrides the default JVM properties
and policy.
- When runtime java is 11 and higher, using BouncyCastle FIPS
Cryptographic provider and BCJSSE in FIPS mode. These are
used as testRuntime dependencies for unit
tests and internal clusters, and copied (relevant jars)
explicitly to the lib directory for testclusters used in REST tests
- When runtime java is 8, using BouncyCastle FIPS
Cryptographic provider and SunJSSE in FIPS mode.
Running the tests in FIPS 140 approved mode doesn't require an
additional configuration either in CI workers or locally and is
controlled by specifying -Dtests.fips.enabled=true
* Stop Allocating Buffers in CopyBytesSocketChannel (#49825)
The way things currently work, we read up to 1M from the channel
and then potentially force all of it into the `ByteBuf` passed
by Netty. Since that `ByteBuf` tends to by default be `64k` in size,
large reads will force the buffer to grow, completely circumventing
the logic of `allocHandle`.
This seems like it could break
`io.netty.channel.RecvByteBufAllocator.Handle#continueReading`
since that method for the fixed-size allocator does check
whether the last read was equal to the attempted read size.
So if we set `64k` because that's what the buffer size is,
then wirte `1M` to the buffer we will stop reading on the IO loop,
even though the channel may still have bytes that we can read right away.
More imporatantly though, this can lead to running OOM quite easily
under IO pressure as we are forcing the heap buffers passed to the read
to `reallocate`.
Closes#49699
Tim Brooks
Rene Groeschke
David Turner
Jake Landis
Yannick Welsch
Luca Cavanna
Armin Braun
William Brafford
Jason Tedor
Rory Hunter
Maria Ralli
Ioannis Kakavas