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
With its initial fix in gh-18444, the `WebClient` instrumentation would
record all CANCEL signals, including:
* when a `timeout` expires and the response has not been received
* when the client partially consumes the response body
Since the second use case is arguable intentional, this commit restricts
the instrumentation and thus avoids recording two events for a single
request in that case.
Closes gh-18444
Prior to this commit, cancelled client requests (for example as a result
of a `timeout()` reactor operator would not be recorded by Micrometer.
This commit instruments the cancelled signal for outgoing client
requests and assigns a status `CLIENT_ERROR`.
The cancellation can be intentional (triggering a timeout and falling
back on a faster alternative) or considered as an error. The intent
cannot be derived from the signal itself so we're considering it as a
client error.
Closes gh-18444
The system keyspace has a replication factor of 1 and is local to each
node; it is therefore recommended to query system.local with a
consistency level of ONE or LOCAL_ONE.
Stronger consistency levels may result in an Unavailable error, but this
does not mean that the node is down.
See gh-20709
Prior to this commit, `LivenessState` and `ReadinessState` were
immutable classes. This was done in order to have additional behavior
and information in those classes.
Because the current implementation doesn't need this, this commit turns
those classes into simple enums.
Additional state and information can be added to the
`*StateChangedEvent` classes.
See gh-19593
This commit upgrades the algorithm when trailing slash are to be
ignored. Previously a root URI (i.e. "/") would result to to empty
string which is an issue for monitoring system that requires tag values
to be non empty. If the URI is a single character, the trailing is not
applied and "/" is left as is.
Closes gh-20536
This commit moves the core Liveness and Readiness support to its own
`availability` package. We've made this a core concept independent of
Kubernetes.
Spring Boot now produces `LivenessStateChanged` and
`ReadinessStateChanged` events as part of the typical application
lifecycle.
Liveness and Readiness Probes (`HealthIndicator` components and health
groups) are still configured only when deployed on Kubernetes.
This commit also improves the documentation around Probes best practices
and container lifecycle considerations.
See gh-19593
Prior to this commit and as of Spring Boot 2.2.0, we would advise
developers to use the Actuator health groups to define custom "liveness"
and "readiness" groups and configure them with subsets of existing
health indicators.
This commit addresses several limitations with that approach.
First, `LivenessState` and `ReadinessState` are promoted to first class
concepts in Spring Boot applications. These states should not only based
on periodic health checks. Applications should be able to track changes
(and adapt their behavior) or update states (when an error happens).
The `ApplicationStateProvider` can be injected and used by applications
components to get the current application state. Components can also
track specific `ApplicationEvent` to be notified of changes, like
`ReadinessStateChangedEvent` and `LivenessStateChangedEvent`.
Components can also publish such events with an
`ApplicationEventPublisher`. Spring Boot will track startup event and
application context state to update the liveness and readiness state of
the application. This infrastructure is available in the
main spring-boot module.
If Spring Boot Actuator is on the classpath, additional
`HealthIndicator` will be contributed to the application:
`"LivenessProveHealthIndicator"` and `"ReadinessProbeHealthIndicator"`.
Also, "liveness" and "readiness" Health groups will be defined if
they're not configured already.
Closes gh-19593
Prior to this commit, `HealthContributor` would be exposed under the
main `HealthEndpoint` and subgroups, `HealthEndpointGroups`. Groups are
driven by configuration properties and there was no way to contribute
programmatically new groups.
This commit introduces the `HealthEndpointGroupsRegistry` (a mutable
version of `HealthEndpointGroups`) and a
`HealthEndpointGroupsRegistryCustomizer`. This allows configurations to
add/remove groups during Actuator auto-configuration.
Closes gh-20554
This commit upgrades to the Couchbase SDK v3 which brings the following
breaking changes:
* Bootstrap hosts have been replaced by a connection string and the
authentication is now mandatory.
* A `Bucket` is no longer auto-configured. The
`spring.couchbase.bucket.*` properties have been removed
* `ClusterInfo` no longer exists and has been replaced by a dedicated
API on `Cluster`.
* `CouchbaseEnvironment` no longer exist in favour of
`ClusterEnvironment`, the customizer has been renamed accordingly.
* The bootstrap-related properties have been removed. Users requiring
custom ports should supply the seed nodes and initialize a Cluster
themselves.
* The endpoints-related configuration has been consolidated in a
single IO configuration.
The Spring Data Couchbase provides an integration with the new SDK. This
leads to the following changes:
* A convenient `CouchbaseClientFactory` is auto-configured.
* Repositories are configured against a bucket and a scope. Those can
be set via configuration in `spring.data.couchbase.*`.
* The default consistency property has been removed in favour of a more
flexible annotation on the repository query methods instead. You can now
specify different query consistency on a per method basis.
* The `CacheManager` implementation is provided, as do other stores for
consistency so a dependency on `couchbase-spring-cache` is no longer
required.
See gh-19893
Co-authored-by: Michael Nitschinger <michael@nitschinger.at>
Previously, any HTTP request to an endpoint that included a principal
would bypass the cache. This prevented authenticated requests from
making use of the cache and its configurable time-to-live.
This commit updates the caching operation invoker to include the
principal, if any, in its cache key. As a result, requests that
include a principal will make use of the cache, potentially returning
the result of a previous invocation of the same endpoint by the same
principal.
Closes gh-19538
Unfortunately, while redundant for new applications, removing the
leading slash adversely affected existing application upon upgrades as
it caused Liquibase to re-apply every change log.
Closes gh-20177
When a request to the /actuator/env/{toMatch} endpoint does not match a
property, a response status 404 was being returned along with a body
containing the existing property sources. This commit removes the body
from the response to be more consistent with a typical 404 response.
Fixes gh-20314
This commit adds metrics support for `ConnectionPool` beans.
See gh-19988
Co-authored-by: Mark Paluch <mpaluch@pivotal.io>
Co-authored-by: Tadaya Tsuyukubo <tadaya@ttddyy.net>
This commit adds an health indicator for R2DBC. If a validation query is
provided, it is used to validate the state of the database. If not, a
check of the connection is issued.
See gh-19988
Co-authored-by: Mark Paluch <mpaluch@pivotal.io>
Prior to this commit, requests made by `HttpRequestInterceptor`
instances configured on `RestTemplate` would not be recorded
properly.
This commit ensures that nested requests are recorded separately.
Closes gh-20231
This commit changes DataSourceHealthIndicator to validate the connection
rather than issuing a query to the database. If a custom validation
query is specified, it uses that as before.
Closes gh-17582
As of spring-projects/spring-framework#22644, Spring Framework caches
the "produces" condition when matching for endpoints in the
`HandlerMapping` infrastructure. This has been improved in
spring-projects/spring-framework#23091 to prevent side-effects in other
implementations.
Prior to this commit, the Spring Boot actuator infrastructure for
`EndpointHandlerMapping` would not clear the cached attribute,
presenting the same issue as Spring Framework's infrastructure. This
means that a custom arrangement with custom `HandlerMapping` or
`ContentTypeResolver` would not work properly and reuse the cached
produced conditions for other, unintented, parts of the handler mapping
process.
This commit clears the cached data and ensures that other handler
mapping implementations are free of that side-effect.
Fixes gh-20150
Prior to this commit, Actuator endpoints would use the application
ObjectMapper instance for serializing payloads as JSON. This was
problematic in several cases:
* application-specific configuration would change the actuator endpoint
output.
* choosing a different JSON mapper implementation in the application
would break completely some endpoints.
Spring Boot Actuator already has a hard dependency on Jackson, and this
commit uses that fact to configure a shared `ObjectMapper` instance that
will be used by the Actuator infrastructure consistently, without
polluting the application context.
This `ObjectMapper` is used in Actuator for:
* JMX endpoints
* Spring MVC endpoints with an HTTP message converter
* Spring WebFlux endpoints with an `Encoder`
* Jersey endpoints with a `ContextResolver<ObjectMapper>`
For all web endpoints, this configuration is limited to the
actuator-specific media types such as
`"application/vnd.spring-boot.actuator.v3+json"`.
Fixes gh-12951
This partially re-applies the deprecation removal from commit
df1837a16b,
without removing CompositeHealthIndicator, HealthAggregator, and related
configuration that is required by Spring Cloud.
Update all dependencies declarations to use the form `scope(reference)`
rather than `scope reference`.
Prior to this commit we declared dependencies without parentheses unless
we were forced to add them due to an `exclude`.
Replace Gradle single quote strings with the double quote form
whenever possible. The change helps to being consistency to the
dependencies section where mostly single quotes were used, but
occasionally double quotes were required due to `${}` references.
This commit removes references of cache infra following the move to
Micrometer. We no longer ships an infinispan specific binder so the
dependency has been removed as well.
Closes gh-19838
This commit deprecates the only public accessor to
`CacheOperationInvoker` so that we can make the entire class package
private in the next feature release.
Closes gh-19089
This paves the way for publishing Gradle module metadata once the
problem caused by snapshot versions and our two-step publication
process has been addressed.
See gh-19609
This reverts commit b34a311d02 as,
having disabled the publishing of Gradle's module metadata (4f75ab5),
the changes are no longer needed.
See gh-19609
Previously, enforcedPlatform dependencies were using to pull in the
constraints defined in spring-boot-dependencies and
spring-boot-parent and applied them strictly so that the constrained
version had to be used. This worked as intended in Spring Boot's own
build but incorrectly enforced those same strict version requirements
on external consumers of Spring Boot's modules.
This commit reworks how Spring Boot defines its internal dependency
management so that platform dependencies are exposed to external
consumers while enforced platform dependencies are using internally.
See gh-19609
Prior to this commit, requests made by `HttpRequestInterceptor`
instances configured on `RestTemplate` would not be recorded
properly.
This commit ensures that nested requests are recorded separately.
See gh-19381
Previously, the endpoint used the same change log history service for
for each SpringLiquibase bean that it processed. This resulted in
pollution of the reported changes as the history of each bean was not
isolated.
This commit updates the endpoint to use a new history service for each
SpringLiquibase bean that is processed.
See gh-19171
Prior to this commit, ApiVersion was treated as a mandatory parameter in
CachingOperationInvokerAdvisor and thus prevented the
CachingOperationInvoker to kick in. By skipping ApiVersion in the same
way we're skipping SecurityContext we can avoid this.
In order to not return the same cached response, this commit also
changes the cache handling in CachingOperationInvoker to account for
different ApiVersions being passed.
See gh-18961
This commit makes sure that the health endpoint returns a default health
status when no contributors are available. Previously, it was returning
`null` which leads to a 404 when exposed via HTTP.
Closes gh-18676
Previously, @EndpointFilter would only have an effect when used as
an annotation or meta-annotation on the endpoint class itself. It
would have no effect when used on a super-class of the endpoint
bean's class.
This commit updates EndpointDiscoverer so that an @EndpointFilter
annotation or meta-annotation on a super-class will be found and
applied to the discovery process. This is achieved by using find…
rather than get… when retrieving the attributes for the EndpointFilter
annotation.
Fixes gh-17866
Update `NamedContributorsMapAdapter` to check for `null` keys or values
during construction. Also update `HealthEndpointSupport` to allow
null component entries.
See gh-18687
Update `HealthEndpointSupport` so that aggregate elements that don't
ultimately provide a contribution are filtered out. Prior to this
commit an NPE was returned when calculating the aggregate status.
Fixes gh-18687
Previously, Maven's default behaviour was relied up which resulted
in the artifact ID being appended to each URL as it was inherited.
This behaviour can only be disabled in Maven 3.6 and later, a version
that we cannot use due to an incompatibility with the Flatten Plugin.
This commit works around Maven's default behaviour by defining
properties for the SCM URL, connection, and developer connection and
then explicitly defining the settings in each pom using these
properties. The explicit definition of the properties in each pom
prevents them being inherited from the parent, thereby disabling the
unwanted appending of the artifact ID to the URL.
Fixes gh-18328
Extract reactor specific code to an inner class to protect
against ClassNotFound exceptions if reactor is not in use.
Also add support for `Flux`.
See gh-18339
Update `CachingOperationInvoker` so that TTL caching is applied directly
to reactive types. Prior to this commit, a `Mono` would be cached, but
the values that it emitted would not.
See gh-18339
Add a `show-components` property under `management.endpoint.health` and
`management.endpoint.health.group.<name>` that can be used to change
when components are displayed.
Prior to this commit it was only possible to set `show-details` which
offered an "all or nothing" approach to the resulting JSON. The new
switch allows component information to be displayed whilst still hiding
potentially sensitive details returned from the actual `HealthIndicator`.
Closes gh-15076
Update the health endpoint so the nested components are now exposed
under `components` rather than `details` when v3 of the actuator
REST API is being used.
This distinction helps to clarify the difference between composite
health (health composed of other health components) and health
details (technology specific information gathered by the indicator).
Since this is a breaking change for the REST API, it is only returned
for v3 payloads. Requests made accepting only a v2 response will have
JSON provided in the original form.
Closes gh-17929
Add `ApiVersion` enum that can be injected into actuator endpoints if
they need to support more than one API revision.
Spring MVC, WebFlux and Jersey integrations now detect the API version
based on the HTTP accept header. If the request explicitly accepts a
`application/vnd.spring-boot.actuator.v` media type then the version
is set from the header. If no explicit Spring Boot media type is
accepted then the latest `ApiVersion` is assumed.
A new v3 API revision has also been introduced to allow upcoming health
endpoint format changes. By default all endpoints now consume and
can produce v3, v2 and `application/json` media types.
See gh-17929
Allow legacy actuator endpoint IDs that contain dots to be transparently
migrated to the new format. This update will allow Spring Cloud users
to proactively migrate from endpoints such as `hystrix.stream` to
`hystrixstream`.
Closes gh-18148
Refactor the `org.springframework.boot.actuate.context` package
with the following changes:
- Deprecate several classes which would ideally be internal
- Replace `ConfigurationBeanFactoryMetadata` with a new
`ConfigurationPropertiesBean` class to better reflect that we no
longer maintain meta-data directly.
- Use constructor injection and final fields whenever possible
- Rename `ConfiguraionPropertiesBeanDefinition` to
`ConfigurationPropertiesValueObjectBeanDefinition` to align
with the binder changes made in commit 0b3015e4ff
- Add additional tests
Closes gh-16903
Prior to this commit, Spring Boot would use `Schedulers.elastic()` when
required to process blocking tasks in a reactive environment.
reactor/reactor-core#1804 introduced a new scheduler,
`Schedulers.boundedElastic()` that behaves quite similarly but:
* will limit the number of workers thread
* will queue tasks if no worker thread is available and reject them is
the queue is exceeds a limit
This allows Spring Boot to schedule blocking tasks as before and allows
greater flexibility.
Fixes gh-18269
See gh-18276
Update `SolrHealthIndicator` to fallback to a basic ping operation if
the `baseUrl` references a particular core rather than the root context.
Prior to this commit, if the Solr `baseUrl` pointed to a particular
core then the health indicator would incorrectly report `DOWN`.
See gh-16477
This commit renames ApplicationHealthIndicator to PingHealthIndicator
and changes the auto-configuration so that it is now always configured
by default.
Closes gh-17926
Update the `HealthEndpoint` to support health groups. The
`HealthEndpointSettings` interface has been replaced with
`HealthEndpointGroups` which provides access to the primary group
as well as an optional set of additional groups.
Groups can be configured via properties and may have custom
`StatusAggregator` and `HttpCodeStatusMapper` settings.
Closes gh-14022
Co-authored-by: Stephane Nicoll <snicoll@pivotal.io>
Overhaul `HealthEndpoint` support to make it easier to support health
groups. Prior to this commit the `HealthIndicator` interface was used
for both regular indicators and composite indicators. In addition the
`Health` result was used to both represent individual, system and
composite health. This design unfortunately means that all health
contributors need to be aware of the `HealthAggregator` and could not
easily support heath groups if per-group aggregation is required.
This commit reworks many aspects of the health support in order to
provide a cleaner separation between a `HealthIndicator`and a
composite. The following changes have been made:
- A `HealthContributor` interface has been introduced to represent
the general concept of something that contributes health information.
A contributor can either be a `HealthIndicator` or a
`CompositeHealthContributor`.
- A `HealthComponent` class has been introduced to mirror the
contributor arrangement. The component can be either
`CompositeHealth` or `Health`.
- The `HealthAggregator` interface has been replaced with a more
focused `StatusAggregator` interface which only deals with `Status`
results.
- `CompositeHealthIndicator` has been replaced with
`CompositeHealthContributor` which only provides access to other
contributors. A composite can no longer directly return `Health`.
- `HealthIndicatorRegistry` has been replaced with
`HealthContributorRegistry` and the default implementation now
uses a copy-on-write strategy.
- `HealthEndpoint`, `HealthEndpointWebExtension` and
`ReactiveHealthEndpointWebExtension` now extend a common
`HealthEndpointSupport` class. They are now driven by a
health contributor registry and `HealthEndpointSettings`.
- The `HealthStatusHttpMapper` class has been replaced by a
`HttpCodeStatusMapper` interface.
- The `HealthWebEndpointResponseMapper` class has been replaced
by a `HealthEndpointSettings` strategy. This allows us to move
role related logic and `ShowDetails` to the auto-configure module.
- `SimpleHttpCodeStatusMapper` and `SimpleStatusAggregator`
implementations have been added which are configured via constructor
arguments rather than setters.
- Endpoint auto-configuration has been reworked and the
`CompositeHealthIndicatorConfiguration` class has been replaced
by `CompositeHealthContributorConfiguration`.
- The endpoint JSON has been changed make `details` distinct from
`components`.
See gh-17926
Update `@Selector` with a `match` attribute that can be used to select
all remaining path segments. An endpoint method like this:
select(@Selector(match = Match.ALL_REMAINING) String... selection)
Will now have all reaming path segments injected into the `selection`
parameter.
Closes gh-17743
Using a random value for the logfile name caused
the logfile endpoint to return a 404 as the name
was resolved from the environment on every request.
This commit registers a bean for LogFile which is then
used by the logfile endpoint.
Fixes gh-17434
When a request that accepts text/plain is received, the threaddump
endpoint will now return a thread dump in plain text. The format of
this text is modelled after the output produced by JVisualVM when
connecting to a remote process over JMX. Note that this output does
not include all of the information in, for example, JStack's output
as it is not available via Java 8's ThreadInfo API.
Rather than the custom formatting logic, using ThreadInfo's toString()
method was considered but its output is documented as being undefined
and implementation specific. The implementation used while developing
this feature produced output that did not match that of JStack or
JVisualVM and truncated stack traces quite considerably.
At the time of writing the format produced by the endpoint could be
consumed by both Thread Dump Analyzer [1] and https://fastthread.io.
Closes gh-2339
[1] https://github.com/irockel/tda
Apply checkstyle rule to ensure that private and package private
classes do not have unnecessary public methods. Test classes have
also been unified as much as possible to use default scoped
inner-classes.
Closes gh-7316
On error cases, the "outcome" tag would be missing from recorded metrics
for the `WebClient`.
This commit fixes this issue and improves the reference documentation by
mentioning the tag values used for error cases, when the client response
is not received (I/O errors, client error, etc).
Fixes gh-17219
Since the move to JUnit 5, a number of tests were failing on Windows.
The majority were failing due to open file handles preventing the
clean up of the tests' temporary directory. This commit addresses
these failures by updating the tests to close JarFiles, InputStreams,
OutputStreams etc.
A change has also been made to CachingOperationInvokerTests to make
a flakey test more robust. Due to System.currentTimeMillis() being
less precise on Windows than it is on *nix platforms, the test could
fail as it would not sleep for long enough for the TTL period to have
expired.
Split the JUnit 5 `OutputCapture` class into separate `OutputExtension`
and `CapturedOutput` classes. The JUnit 5 callback methods are now
contained only in the `OutputExtension` class so no longer pollute the
public API that users will interact with.
The `CapturedOutput` class has also been updated to capture System.err
and System.out separately to allow distinct assertions if required.
Closes gh-17029
Refactor `Autotime` from a properties object to an interface and
change the existing metric recording implementations. The `AutoTimer`
interface is a general purpose callback that can be applied to a
`Timer.Builder` to configure it. Autotime properties are now located
in `spring-boot-actuator-autoconfigure` and have become an
implementation of the interface.
Closes gh-17026
When `management.metrics.web.server.auto-time-requests` is enabled
(default=true), Spring Boot collects metrics on controller methods even
when they are not annotated with `@Timed`.
When this happens, created metrics are based on the default
`@Timed` configuration and there is no way to customize the
configuration of those auto-timed controller metrics.
This commit adds default configurations to auto-timed requests on both
client and server sides.
See gh-15988
Rework `AbstractApplicationContextRunner.withBean` methods to
align signatures as much as possible with those provided by
the `ApplicationContext`.
Also update the implementation to use a dedicate member
variable rather than adding initializers.
Closes gh-16011
After a hierarchy change in Spring Framework in gh-22543,
`AbstractWebFluxEndpointHandlerMapping` doesn't need to override the
`getMappingPathPatterns` method anymore.
This commit migrates `AnnotationConfigReactiveWebApplicationContext`
parent to the `GenericApplicationContext` abstraction. Any use of
`AnnotationConfigWebApplicationContext` is also removed as it also
inherits from the `AbstractRefreshableApplicationContext` outdated
hierarchy.
A new `AnnotationConfigServletWebApplicationContext` context is
introduced instead, extending from `GenericApplicationContext` and
providing the counter part of the reactive context for the Servlet-based
web app tests.
See gh-16096
This commit updates CORS handling according to Framework changes
introduced via [1]. It also fixes tests according to the new behavior.
See gh-16410
[1] d27b5d0ab6.
Prior to this commit, exceptions nested in
`NestedServletExceptions` would not be recorded by the
`WebMvcMetricsFilter`. This commit ensures that exceptions
happening downstream (e.g. happening while writing the response
body itself) are properly recorded.
See https://github.com/micrometer-metrics/micrometer/issues/1190
See gh-16014
Update the `BaseConfiguration` class in `ScheduledTasksEndpointTests`
to be package private so that it can be enhanced by cglib.
Prior to merge commit 361437f4 the class was a lite configuration so
it didn't matter that it was a private class.
loadOnStartup property was missing from EndpointServlet and cannot be set
inside ServletEndpointRegistrar. Now it can be set and register a Servlet
with that integer property ready to act upon registration.
See gh-16053
When Spring Security is misconfigured it's possible to switch from an anonymous user
to a normal user. When switching back again, the corresponding
AuthenticationSwitchUserEvent will have a null target user. Previously, Actuator's
AuthenticationAuditListener would throw a NullPointerException when it received such an
event.
This commit updates the audit listener to defensively handled events with a null target
user.
Closes gh-15767
Prior to this commit, the `HttpTraceWebFilter` would collect the
response information (status and headers) for tracing purposes, after
the handling chain is done with the exchange - inside a
`doAfterSuccessOrError`.
Once the handler has processed the exchange, there is no strong
guarantee about the HTTP resources being still present. Depending on the
web server implementation, HTTP resources (including HTTP header maps)
might be recycled, because pooled in the first place.
This commit moves the collection and processing of the HTTP trace right
before the response is committed. This removes the need to handle
special cases with exceptions, since by that time all exception handlers
have processed the response and the information that we extract is the
information that's about to be written to the network.
Fixes gh-15819
This commit aligns the Spring WebFlux instrumentation on Spring MVC
since gh-12447.
From now on, if the best matching path pattern is not found,
the recorded uri tag will be "UNKNOWN".
Note that for WebFlux.fn, the pattern information is properly
recorded as of SPR-17395.
Closes gh-15609
Similar to what's ben done in gh-15420 for Spring MVC and Spring
WebFlux, this commit adds an outcome tag for the client side on both
`RestTemplate` and `WebClient`.
See gh-15594
This commit adds more information to the ElasticSearch REST
health indicator.
When the ES instance responds with an error HTTP status,
the health details now include the actual status code and reason phrase.
When the ES instance returns a HTTP 200 response, the entire response
map is used as health details.
See gh-15366
This commit adds `ElasticsearchRestHealthIndicator`, a new
`HealthIndicator` for Elasticsearch, using the Elasticsearch "low level
rest client" provided by the
`"org.elasticsearch.client:elasticsearch-rest-client"` dependency.
Note that Spring Boot will auto-configure both low and high level REST
clients, but since the high level one is using the former, a single
health indicator will cover both cases.
See gh-15211
This commit changes the requested endpoint for the Jest
HealthIndicator. The `"/_all/_stats"` was previously used, but
the response size can be quite large and costly.
This is now using the `"/_cluster/health"` endpoint.
Previously, when using Tomcat, a call to mappings endpoint would force
the initialization of any DispatcherServlets in the context. This was
done by calling allocate on Tomcat's StandardWrapper. This left the
wrapper in a state that would cause it to block for two seconds during
shutdown as the wrapper has an outstanding allocation.
This commit immediately deallocates the servlet after it has been
allocated. This ensures that the DispatcherServlet has been initialized
while also leaving the wrapper in a state that it can shut down
immediately when asked to do so.
Closes gh-14898
SPR-17395 ensures that WebFlux.fn is adding a request attribute of type
`PathPattern` on the `HandlerMapping.BEST_MATCHING_PATTERN_ATTRIBUTE`.
A specific tag provider for WebFlux.fn is no longer necessary.
See gh-14876
Update `WebEndpointDiscoverer` and related classes to that multiple
`PathMapper` beans can be registered. Mappers are now tried in order
until one returns a non-null value.
Closes gh-14841
Previously, Couchbase's health was determined by retrieving the bucket info
from the cluster info. This retrieval could take over one minute in some
cases even when Couchbase is health. This latency is too large for a health
check.
The Couchbase team have recommended the of a Cluster#diagnostics instead.
This provides a much lower latency view of the cluster's health. This
commit updates CouchbaseHealthIndicator to use Cluster#diagnostics while
retaining support, in a deprecated form, for the old info-based mechanism
should anyone want to opt back into that in 2.0.x.
Closes gh-14685
Update `ExposeExcludePropertyEndpointFilter` so that mixed case
endpoint IDs are supported. Prior to this commit it was not easy for
an endpoint to be missed by the filter due to the formatting of the
property value.
See gh-14773
Update the endpoint time-to-live binding logic so that mixed case
endpoint IDs are supported. Prior to this commit an
`InvalidConfigurationPropertyNameException` would be thrown when using
a camel case endpoint ID.
See gh-14773
Add an `EndpointID` class to enforce the naming rules that we support
for actuator endpoints. We now ensure that all endpoint names contain
only letters and numbers and must begin with a lower-case letter.
Existing public classes and interfaces have been changes so that String
based `endpointId` methods are deprecated and strongly typed versions
are preferred instead. A few public classes that we're not expecting
to be used directly have been changed without deprecated methods being
introduced.
See gh-14773
Update `MetricsEndpoint` so that only the first matching meter is used
when calculating the sum of of statistics.
Prior this this commit the endpoint would consider all Meters. This
caused incorrect statistics when multiple back-end systems were being
used since the registries contained in the `CompositeMeterRegistry`
would be iterated, and the same effective metric would be counted more
than once.
Closes gh-14497
Previously, the API required to create HttpTrace instances was
package-private. This made it difficult to implement an
HttpTraceRepository that persists the HttpTrace instances
rather than holding them in memory as it inhibited recreation of the
instances when they read from the persistent store.
This commit adds public constructors to HttpTrace and related classes
to enable recreation of an HttpTrace. The package-private methods for
mutating properties have not been made public to ensure that the
public API remains immutable.
Closes gh-14726
Previously, if the call to connection.start() hung, JmsHealthIndicator
would also hang and then never respond.
This commit introduces the use of an additional thread that waits for
up to 5 seconds for the connection to start. If the call to start
does not complete within that time, the connection is closed. The
call to close causes the call to start to throw an exception, thereby
stopping the hang and allowing the indicator to report that the
broker is down.
Closes gh-10809
Rework Prometheus push gateway support so that the central class can
be used outside of auto-configuration. The shutdown flags have also
been replaced with a single "shutdown-operation" property since it's
unlikely that both "push" and "delete" will be required.
It's also possible now to supply a `TaskScheduler` to the manager.
See gh-14353
This commit aligns SessionsEndpoint with
FindByIndexNameSessionRepository API improvements that simplifies
retrieval of sessions by principal name.
Closes gh-14124
This commit changes AbstractWebMvcEndpointHandlerMapping to
be a MatchableHandlerMapping. Additionally, EndpointRequest,
now delegates to MvcRequestMatcher for Spring MVC applications.
For all other applications, AntPathRequestMatcher is used as
a delegate.
Closes gh-13962
This commit makes sure to use a configurable timeout to check if the
Couchbase cluster is up, rather than relying on the default that can be
quite long.
Closes gh-13879
Previously, the setter of a property whose second letter is upper-case (
such as `oAuth2Uri`) was not detected properly. The JavaBean spec states
that, in such a case, the first letter should not be capitalized (i.e.
the setter should be `setoAuth2Uri` rather than `setOAuth2Uri`).
This commit makes sure that Jackson uses standard bean names and fixes
the setter detection algorithm to take this case into account.
Closes gh-13878
Add an `DispatcherServletPath` interface which provides a much more
consistent way to discover the path of the main dispatcher servet.
Prior to this commit, auto-configurations would often make use of the
`ServerProperties` class to discover the dispatcher servlet path. This
mechanism isn't very explicit and also makes it hard for us to relocate
that property in Spring Boot 2.1.
This commit also reverts most of fddc9e9c7e since it is now clear that
the supporting multiple dispatcher servlet paths will be much more
involved that we originally anticipated.
Closes gh-13834
Previously, when a DispatcherServlet was registered via a
ServletRegistrationBean, the mappings endpoint did not expose any
information about it as it wasn't detected.
This commit fixes the detection of available DispatcherServlets in the
context so that the mappings endpoint include them all.
Closes gh-13186
This commit improves the for Actuator Metrics in WebClient.
Unlike the server couterpart of WebFlux, using a `retry` operator on a
`WebClient` pipeline does resubscribes to the whole chain.
The previous implementation recorded start time at the time of pipeline
build phase, but outside of it. This doesn't work since retrying the
same pipeline doesn't update the recorded start time and the duration of
sequential calls are cumulative.
This is now fixed using the Reactor `Context`, since we're now recording
the start time at subscription time and record metrics on `onNext` and
`onError` signals.
Closes gh-12228
Previously, a NullPointerException would occur when WebFluxTags
attempted to create a method Tag for a request with a non-standard
method.
This commit updates WebFluxTags to use getMethodValue(), which will
never return null, rather than getMethod(), which may return null,
when determining the tag's value for the given request.
Closes gh-13596
As of Spring Framework 5.1, we're depending on the Reactor Californium
release train.
Reactor Netty is now at version 0.8 and changed its artifact
coordinates, package names and broke several APIs. Spring Framework is
now up-to-date with those changes and this commit does the same for
Spring Boot.
Note that in that process, the `NettyServerCustomizer` has been changed
since the former `HttpServerOptions.Builder` API is now gone from
Reactor Netty, and we're now relying on immutable server instances
instead of a stateful builder pattern.
See gh-13321
Previously, WebMvcTags would fall back to using the request’s path
info if Spring MVC’s best matching path pattern attribute was not
available. For non-Spring MVC web frameworks that support path
variables, such as Jersey, this led to an explosion of URI tags that
could exhaust the heap.
Closes gh-12447
This commit fixes endpoint extension discovery when the related endpoint
is sub-classed. Previously, a strict by type check was applied against
the `endpoint` attribute of `EndpointExtension`.
Rather than using a `Class` check, this commit extracts the id of an
endpoint and uses it to match its extension, if any.
Closes gh-13082
This commit improves the `health` endpoint to run health check for a
particular component or, if that component is itself a composite, an
instance of that component.
Concretely, it is now possible to issue a `GET` on
`/actuator/health/{component}` and
`/actuator/health/{component}/instance` to retrieve the health of a
component or an instance of a composite component, respectively.
If details cannot be showed for the current user, any request leads to a
404 and does not invoke the health check at all.
Closes gh-8865
This commit updates the initial proposal to add support for reactive
use cases as well. A reactive application can use
ReactiveHealthIndicatorRegistry as an alternative to
HealthIndicatorRegistry.
Closes gh-4965
This commit introduces HealthIndicatorRegistry which handles
registration of HealthIndicator instances. Registering new
HealthIndicator instances is now possible in runtime.
See gh-4965
Andy Wilkinson
Scott Frederick
Stephane Nicoll
dreis2211
Alexandre Dutra
lltx
Phillip Webb
Johnny Lim
Jon Schneider
Phillip Webb
Brian Clozel
Andreas Born
Said BOUDJELDA
Dmytro Nosan
zhangt2333
Ruslan Stelmachenko
Madhura Bhave
cbono
babjo
Christoph Strobl
Matthias Stock
luoyuan
Leo Li
Lars Grefer
thelproad
mattisonchao
wycm
contextshuffling
wonwoo
Markus Schuch
HaiTao Zhang
jason
lijuny
Jacques-Etienne Beaudet
ayudovin
Tadaya Tsuyukubo
Vedran Pavic
Sebastien Deleuze
Spring Operator
Gómez Díaz, Julio José
Mehmed Baždar
Michael Simons
igor-suhorukov
Nishant Raut
Oleksii Bondar
Filip Hrisafov
holy12345
Erlend Haugsnes
TwinProduction
Andre Restolho
Rahul Ahuja
Julian Devia
Hanope
pmehra
Mikalai Lushchytski
Michael McFadyen
durigon
Michael Pratt
artsiom
Johannes Edmeier
Jon Schneider
David Herberth