This commit adds the required infrastructure for instrumenting the
GraphQL engine and datafetchers in order to collect metrics.
With this infrastructure, we can collect metrics such as:
* "graphql.request", a timer for GraphQL query
* "graphql.datafetcher", a timer for GraphQL datafetcher calls
* "graphql.request.datafetch.count", a distribution summary of
datafetcher count per query
* "graphql.error", an error counter
See gh-29140
Previously the parameters were not ignored so */*;q=0.8 was not
identified as accepting */*. As a result, the most recent producible
content type was returned rather than the default.
This commit updates the argument resolver to ignore parameters.
The default content type is now produced in response to a request
that accepts */*, irrespective of its parameters.
Fixes gh-28446
Change the cache in `CachingOperationInvoker` to be a reference based
map and also clean stale entries when a specific threshold is met.
Fixes gh-28313
Update Prometheus `TextOutputFormat` so that OpenMetrics is used in
preference to text output when an appropriate accept header is found.
If the accept header contains `*/*` or is missing then the text format
will be used.
See gh-28130
Add an `isDefault()` method to `Producible` which can be used to
indicate which of the enum values should be used when the accept header
is `*/*` or `null`.
Prior to this commit, the last enum value was always used as the
default.
See gh-28130
Previously, the URI template handler installed by the client metrics
interceptor would always capture the URI template and push it onto the
deque, irrespective of whether auto timing was enabled. When
auto-timing is disabled the deque is never polled so this led to its
unrestricted growth.
This commit updates the URI template handler so that a URI template is
only pushed onto the deque when the auto timing configuration enables
the interceptor.
Fixes gh-26915
Change `CompositeHealth.getDetails()` to public so that it serializes
correctly when `MapperFeature.CAN_OVERRIDE_ACCESS_MODIFIERS` is
disabled.
Fixes gh-26797
Rework existing `@Timer` annotation support to remove duplicate code
and offer general purpose utilities that can be used in future metrics
support.
See gh-23112
See gh-22217
Extract common method logic to a `HandlerMethodTimedAnnotations`
utility class and also add some caching. Also move the test code
into a single class.
See gh-23112
Fix package tangle in the actuator endpoint package by relocating a
few classes.
The `Producible` and `ProducibleOperationArgumentResolver` classes have
been moved from `endpoint.annotation` to `endpoint` since they aren't
directly tied to annotations.
The `ApiVersion` class has been moved from `endpoint.http` to
`endpoint` since it needs to implement `Producible` and isn't really
tied to HTTP.
Closes gh-25914
This commit reworks the initial proposal so that jobs and triggers are
treated as first class concepts.
`/actuator/quartz` now returns the group names for jobs and triggers.
`actuator/quartz/jobs` returns the job names, keyed by the available
group names, while `/actuator/quartz/triggers` does the same for
triggers.
`/actuator/jobs/{groupName}` provides an overview of a job group. It
provides a map of job names with the class name of the job.
implementation
`/actuator/triggers/{groupName}` provides an overview of a trigger
group. There are five supported trigger implementations: cron, simple,
daily time interval, calendar interval, and custom for any other
implementation. Given that each implementation has specific settings,
triggers are split in five objects.
`/actuator/jobs/{groupName}/{jobName}` provides the full details of a
particular job. This includes a sanitized data map and a list of
triggers ordered by next fire time.
`/actuator/triggers/{groupName}/{triggerName}` provides the full details
of a particular trigger. This includes the state, its type, and a
dedicate object containing implementation-specific settings.
See gh-10364
Prior to this commit, some exceptions handled at the controller or
handler function level would:
* not bubble up to the Spring Boot error handling support
* not be tagged as part of the request metrics
This situation is inconsistent because in general, exceptions handled at
the controller level can be considered as expected behavior.
Also, depending on how the exception is handled, the request metrics
might not be tagged with the exception.
This will be reconsidered in gh-23795.
This commit prepares a transition to the new situation. Developers can
now opt-in and set the handled exception as a request attribute. This
well-known attribute will be later read by the metrics support and used
for tagging the request metrics with the exception provided.
This mechanism is automatically used by the error handling support in
Spring Boot.
Closes gh-24028
Previously, would log an error for any exception and also stop
publishing for an UnknownHostException. By constrast, Micrometer's
PushMeterRegistry treats all exceptions the same, logging a warning
and continuing with subsequent push attempts.
This commit updates the push gateway manager's behaviour to match
PushMeterRegistry. UknownHostExceptions no longer receive special
treatment and push (and delete) failures are now logged as warnings
rather than errors.
Fixes gh-25804
Prior to this commit, the Actuator instrumentation for WebFlux servers
would not record metrics in two cases:
* the client disconnects before the response has been sent
* a server timeout is triggered before the response is sent
This commit improves the existing instrumentation to record metrics in
these cases. Since the causes of timeouts/disconnections can vary a lot,
the chosen "outcome" tag for metrics is "UNKNOWN".
Closes gh-23606
Update `AbstractWebMvcEndpointHandlerMapping` to chain any caught
InvalidEndpointRequestExceptions so that a more complete stacktrace
is available. The exception has also been updated to a
`ResponseStatusException` so that the reason can be propagated.
Fixes gh-25642
Previously, the auto-configuration for DataSource initialization and
the properties used to configure it were part of the general
DataSource auto-configuration and properties.
This commit moves the auto-configuration of DataSource initialization
out into a separate top-level auto-configuration class. Similarly,
the properties for configuring DataSource initialization have been
moved from `spring.datasource.*` into `spring.sql.init.*`.
The old initialization-related `spring.datasource.*` properties have
been deprecated but can still be used. When they are used, they new,
separate initialization auto-configuration will back off. In other
words, the initialization related `spring.datasource.*` properties
and the `spring.sql.init.*` properties cannot be used in combination.
Closes gh-25323
Previously, a root URI configured via RestTemplateBuilder's rootUri
method and RootUriTemplateHandler was not taken into account when
generated the URI tag for RestTemplate request metrics.
This commit updates MetricsClientHttpRequestInterceptor to be aware
of RootUriTemplateHandler and capture the URI template once the
root URI has been applied.
Fixes gh-25744
Refine the new `Producible` support so that it can also be used with
`@ReadOperation`, `@WriteOperation` and `@DeleteOperation` annotations.
This update allows the same enum to be used both as an argument and as
an indicator of the media-types that an operation may produce.
Closes gh-25738
Update the actuator @Enpoint` infrastructure code so that operations
may inject enums that indicate the type of output to produce. A new
`Producible` interface can be implemented by any enum that indicates
the mime-type that an enum value produces.
The new `OperationArgumentResolver` provides a general strategy for
resolving operation arguments with `ProducibleOperationArgumentResolver`
providing support for `Producible` enums. Existing injection support has
been refactored to use the new resolver.
See gh-25738
This commit modifies the actuator `EnvironmentEndpoint` to allow
primitive wrapper types to be serialized in the response data
structure.
Fixes gh-24307
When `EnvironmentEndpoint` is building a response to return to the
web infrastructure, it creates a data structure containing all
property values from all property sources. Prior to this commit, it
was possible for the response data structure to contain property
values that were not serializable to JSON by Jackson, which would
cause an exception to be thrown by the web infrastructure. This
commit ensures the data structure is serializable to JSON by
ensuring property values are primitives or Strings, and returning
a placeholder value if a property value is of any other type.
Fixes gh-23805
This commit adds support for Redis cache metrics. Users can opt-in for
statistics using the "spring.cache.redis.enable-statistics" property.
Closes gh-22701
Constructor calls like new AtomicInteger(0) cause a volatile write that
can be saved in cases where the constructor parameter is the default
value.
See gh-23575
Prior to this commit, Actuator would sanitize properties values when
serializing them on the dedicated endpoint. Keys like "password" or
"secret" are entirely sanitized, but other keys like "uri" or "address"
are considered as URI types and only the password part of the user info
is sanitized.
This commit fixes the sanitization process where lists of such URI types
would not match the first entries of the list since they're starting
with `'['`. This commit improves the regexp matching process to sanitize
all URIs within a collection.
The documentation is also updated to better underline the processing
difference between complete sanitization and selective sanitization for
URIs.
Fixes gh-23037
Prior to this commit, the `WebClientExchangeTags`, when given a request
without a string template, would only get the request path to create the
"uri" tag for metrics. This is inconsistent with the
`RestTemplateExchangeTags`, which are taking the full request URI minus
the protocol+host+port.
This commit aligns the `WebClientExchangeTags` behavior in this case.
Closes gh-22832
With the introduction of health indicators that only require the
CqlSession, this commit deprecates the health indicators that require
Spring Data since the latter build on top of the former.
Closes gh-23226
Prior to this commit, Spring Boot would auto-configure both
Elasticsearch variants: `RestClient` ("Low Level" client) and
`RestHighLevelClient` ("High Level" client).
Since one can be derived from the other, this would create complex and
unclear situations depending on what developers provided with their
configuration.
`RestHighLevelClient` is mostly for actual use of the Elasticsearch API,
with support for specific methods and (de)serialization. On the other
hand, `RestClient` is merely wrapping the Apache HTTP client for
load-balancing support and low level HTTP features.
This commit completely removes the support for `RestClient` in Spring
Boot and now requires the presence of the
`org.elasticsearch.client:elasticsearch-rest-high-level-client`
dependency for REST client support with Elasticsearch.
Closes gh-22358
This commit builds on top of gh-22603 and exposes data collected by the
`BufferingApplicationStartup` on a dedicated `"/startup"` Actuator
endpoint.
Closes gh-23213
This commit replaces the Neo4j-OGM based health checks with one based on
the Neo4j Java driver. A Reactive variant is also added in this commit.
See gh-22302
This commit harmonizes dependency declarations for Jackson in the
actuator. Both Jackson and JSR 310 are back to optional in the core
actuator module and mandatory when using the auto-configuration.
Closes gh-22624
Prior to this commit, configuring a reactive Elasticsearch client would
auto-configure an Actuator Health check using a synchronous client, with
the default configuration properties (so tarting localhost:9200).
This would lead to false reports of unhealthy Elasticsearch clusters
when using reactive clients.
This commit reproduces the logic for MongoDB repositories: if a reactive
variant is available, it is selected for the health check
infrastructure.
See gh-21042
Previously, if TomcatMetricsBinder destroy() was called before it had
received an ApplicationStartedEvent an NPE would be thrown due to
TomcatMetrics being null. This NPE was then caught and logged at
warning level by the disposable bean adapter.
This prevents the NPE by checking that the TomcatMetrics instance is
null before calling close() on it.
See gh-22141
Previously, the thread dump endpoint's response could exceed
WebClient's in-memory buffer limt when there were a large number of
threads or the threads had large stacks.
This commit disables WebClient's in-memory buffer size limit so that
the test passing is not dependent on the number of active threads and
their stack sizes.
Closes gh-22101
This commit changes the information provided by
RedisReactiveHealthIndicator to include cluster details when Spring
Data Redis detects that Redis is running in a clustered configuration.
This brings the reactive and non-reactive Redis health indicators
into alignment.
Fixes gh-21514
Prior to Spring Data Redis version 2.2.8, the contents of the
Properties object returned from the
ReactiveRedisConnection.ServerCommands.info API were the same
for clustered and non-clustered Redis configurations, containing a set
of key/value pairs. This allowed ReactiveRedisHealthIndicator to get
a version property using a well-known key. Starting with Spring Data
Redis 2.2.8, the info property keys contain a host:port prefix in a
clustered Redis configuration. This prevented
ReactiveRedisHealthIndicator from getting the version property as
before and resulted in the health always being reported as DOWN.
This commit adjusts ReactiveRedisHealthIndicator to detect the
clustered configuration from Spring Data Redis and find the version
property for one of the reported cluster nodes.
Fixes gh-22061
Previously, Spring Boot's modules published Gradle Module Metadata
(GMM) the declared a platform dependency on spring-boot-dependencies.
This provided versions for each module's own dependencies but also had
they unwanted side-effect of pulling in spring-boot-dependencies
constraints which would influence the version of other dependencies
declared in the same configuration. This was undesirable as users
should be able to opt in to this level of dependency management, either
by using the dependency management plugin or by using Gradle's built-in
support via a platform dependency on spring-boot-dependencies.
This commit reworks how Spring Boot's build uses
spring-boot-dependencies and spring-boot-parent to provide its own
dependency management. Configurations that aren't seen by consumers are
configured to extend a dependencyManagement configuration that has an
enforced platform dependency on spring-boot-parent. This enforces
spring-boot-parent's version constraints on Spring Boot's build without
making them visible to consumers. To ensure that the versions that
Spring Boot has been built against are visible to consumers, the
Maven publication that produces pom files and GMM for the published
modules is configured to use the resolved versions from the module's
runtime classpath.
Fixes gh-21911
In JDK 15 the concept of hidden classes was introduced, which also
affects Lambdas in so far that Class.getCanonicalName() will return null
for those. This commit uses Class.getName() as a fallback when no
canonical name is available.
See gh-21713
This commit provides a CassandraDriverHealthIndicator and
CassandraDriverReactiveHealthIndicator that do not require Spring Data.
As a result, a health indicator for Cassandra is provided even if the
application does not use Spring Data.
See gh-20887
Update `EndpointDiscoverer` so that `@Endpoint` and `@EndpointExtension`
beans are created as late as possible.
Prior to this commit, endpoint beans and extension beans would be
created during the discovery phase which could cause early bean
initialization. The problem was especially nasty when using an embedded
servlet container since `ServletEndpointRegistrar` is loaded as the
container is initialized. This would trigger discovery and load all
endpoint beans, including the health endpoint, and all health indicator
beans.
Fixes gh-20714
Prior to this commit, there was a property server.error.include-details
that allowed configuration of the message and errors attributes in a
server error response.
This commit separates the control of the message and errors attributes
into two separate properties named server.error.include-message and
server.error.include-binding-errors. When the message attribute is
excluded from a servlet response, the value is changed from a
hard-coded text value to an empty value.
Fixes gh-20505
This commit updates HazelcastHealthIndicator and
HazelcastCacheMeterBinderProvider so that they work with
Hazelcast 4 while retaining compatibility with Hazelcast 3. Reflection
is used when necessary.
This commit also adds a smoke test that validates those features are
working when Hazelcast 4 is on the classpath.
Closes gh-21169
Prior to this commit, there was a cycle between `StatusAggregator` and
`SimpleStatusAggregator`, which caused a static initialization bug -
depending on which class (the implementation or its interface) was
loaded first.
This commit turns the static field of the `StatusAggregator` interface
into a static method to avoid this problem.
Fixes gh-21211
Prior to this commit, default error responses included the message
from a handled exception. When the exception was a BindException, the
error responses could also include an errors attribute containing the
details of the binding failure. These details could leak information
about the application.
This commit removes the exception message and binding errors detail
from error responses by default, and introduces a
`server.error.include-details` property that can be used to cause
these details to be included in the response.
Fixes gh-20505
Update the `HealthEndpointGroups` customization support to use a
post-processor rather than a mutable registry. Although this approach
is slightly less flexible, it removes a lot of complexity from the
`HealthEndpointGroups` code. Specifically, it allows us to drop the
`HealthEndpointGroupsRegistry` interface entirely.
The probe health groups are now added via the post-processor if they
aren't already defined. Unlike the previous implementation, users are
no longer able to customize status aggregation and http status code
mapping rules _unless_ they also re-define the health indicators that
are members of the group.
See gh-20962
Relocate probe auto-configuration from the `kubernetes` package to
`availability` since probes could also be used on other platforms.
The classes have also been renamed to named to `AvailabilityProbes...`
See gh-20962
Rename `LivenessProbeHealthIndicator` to `LivenessStateHealthIndicator`
and `ReadinessProbeHealthIndicator` to `ReadinessStateHealthIndicator`.
Also introduce a general purpose `AvailabilityStateHealthIndicator`
class.
See gh-20962
Create a general purpose `AvailabilityState` interface and refactor
the existing `LivenessState` and `ReadinessState` to use it. A single
`AvailabilityChangeEvent` is now used to carry all availability state
updates.
This commit also renames `ApplicationAvailabilityProvider` to
`ApplicationAvailabilityBean` and extracts an `ApplicationAvailability`
interface that other beans can inject. The helps to hide the event
listener method, which is really internal.
Finally the state enums have been renamed as follows:
- `LivenessState.LIVE` -> `LivenessState.CORRECT`
- `ReadinessState.READY` -> `ReadinessState.ACCEPTING_TRAFFIC`
- `ReadinessState.UNREADY` -> `ReadinessState.REFUSING_TRAFFIC`
See gh-20962
Stephane Nicoll
Jonatan Ivanov
izeye
Brian Clozel
Madhura Bhave
Phillip Webb
Andy Wilkinson
Stefano Cordio
Scott Frederick
charissathomas
Leo Li
dreis2211
Stephane Nicoll
bono007
Billy Tobon
Gerrit Meier
Hoan Vu Hong
weixsun
Matthias Stock
anshlykov
Vedran Pavic
Marten Deinum
Scott Frederick
xJoeWoo
Peter Paul Bakker
tomekl007
Madhura Bhave
Andy Wilkinson
David Good
Pradipta Sarma
Emily Tsanova
Gabriele Bianchet-David
ksmilek
biergit
Aleksander Lech
im47cn
Alexandre Dutra
lltx
Jon Schneider
Phillip Webb
Brian Clozel