16 KiB
Spring Bootstrap Actuator Feature Guide
Here are some (most, hopefully all) the features of Spring Bootstrap Actuator with some commentary to help you start using them. We recommend you first build a project with the Actuator (e.g. the getting started project from the main README), and then try each feature in turn there.
TODO: some of these are features of Spring Bootstrap (or
SpringApplication
) not the Actuator.
TODO: group things together and break them out into separate files.
Commandline Arguments
Commandline arguments are passed on to any CommandLineRunner
beans
found in the application. Option arguments (starting with --
,
e.g. --server.port=9000
) are converted to a PropertySource
and
added to the Spring Environment
with first priority (they always
take precedence and override values from other sources). Properties
in the Environment
(including System properties and OS environment
variables) can always be injected into Spring components using
@Value
with placeholders, e.g.
@Component
public class MyService {
@Value("${app.message:Hello World}")
private String message;
...
}
The default value comes after the first colon (":").
Externalized Configuration
In addition to command line option arguments, Spring Bootstrap will
pick up a file called application.properties
in the root of your
classpath (if there is one) and add those properties to the Spring
Environment
. The search path for application.properties
is
actually, 1) root or classpath, 2) current directory, 3) /config
package in classpath, 4) /config
subdir of current directory. The
list is ordered by decreasing precedence (so properties can be
overridden by others with the same name defined in later locations).
The values in application.properties
are filtered through the
existing Environment
when they are used so you can refer back to
previously defined values (e.g. from System properties), e.g.
app.name: MyApp
app.description: ${app.name} is a Cool New App
Spring Bootstrap also binds the properties to any bean in your
application context whose type is @ConfigurationProperties
. The
Actuator provides some of those beans out of the box, so you can
easily customize server and management properties (ports etc.),
endpoint locations and logging. See below for more detail, or inspect
the *Properties
types in the Actuator jar.
Setting the Default Spring Profile
Spring Profiles are a way to segregate parts of the application
configuration and make it only available in certain environments. Any
@Component
that is marked with @Profile
will only be loaded in the
profile specified by the latter annotation.
Spring Bootstrap takes it a stage further. If you include in your
application.properties
a value for a property named
spring.active.profiles
then those profiles will be active by
default. E.g.
spring.active.profiles: dev,hsqldb
Profile-dependent configuration
Spring Bootstrap loads additional properties files if there are active
profiles using a naming convention application-{profile}.properties
.
Property values from those files override trhe default ones.
Custom Typesafe Externalized Configuration
If you want a strongly typed bean (or beans) to govern and validate
the configuration of your application beyond the built in properties,
all you need to do is create a @ConfigurationProperties
class, e.g.
@ConfigurationProperties(name="my")
public class MyProperties {
}
and declare one either explicitly (with @Bean
) or implicitly by
adding
@EnableConfigurationProperties(MyProperties.class)
to one of your @Configuration
(or @Component
) classes. Then you can
@Autowired
private MyProperties configuration = new MyProperties();
in any of your component classes to grab that configuration and use it.
Spring Bootstrap uses some relaxed rules for binding Environment
properties to @ConfigurationProperties
beans, so there doesn't need
to be an exact match between the Environment
property name and the
bean property name. Common examples where this is useful include
underscore separated (e.g. context_path
binds to contextPath
), and
capitalized (e.g. PORT
binds to port
) environment properties.
Spring will attempt to coerce the external application properties to
the right type when it binds to the @ConfigurationProperties
beans.
If you need custom type conversion you can provide a
ConversionService
bean (with bean id conversionService
) or custom
property editors (via a CustomEditorConfigurer
bean).
Spring will also validate the external configuration, by default using JSR-303 if it is on the classpath. So you can add annotations from that specification (or its implementations) to your custom properties, e.g.
@ConfigurationProperties(name="my")
public class MyProperties {
@NotNull
private String name;
// .. getters and setters
}
You can also add a custom Spring Validator
by creating a bean
definition called configurationPropertiesValidator
.
Using Project Lombok
You can safely use Project Lombok to
generate getters and setters for your @ConfigurationProperties
.
Refer to the documentation on the Lombok for how to enable it in your
compiler or IDE.
Using YAML instead of Properties
YAML is a superset of JSON, and as such is a very convenient format
for specifying hierarchical configuration data, such as that supported
by Spring Bootstrap Actuator. If you prefer to use
YAML instead of Properties files you just need to
include a file called application.yml
in the root of your classpath
You can if you like add profile specific YAML files
(application-${profile}.yml
), but a nicer alternative is to use YAML
documents inside application.yml
, with profile-specific documents
containing a spring.profiles
key. For example
server:
port: 8080
management:
port: 8080
address: 0.0.0.0
---
spring:
profiles: prod
management:
port: 8081
address: 10.2.68.12
Customizing the location of the External Configuration
If you don't like application.properties
or application.yml
as the
configuration file location you can switch to another location by
specifying the spring.config.name
(default application
) or the
spring.config.location
as environment properties, e.g. if launching
a jar which wraps SpringApplication
:
$ java -jar myproject.jar --spring.config.name=myproject
Providing Defaults for Externalized Configuration
For @ConfigurationProperties
beans that are provided by the
framework itself you can always change the values that are bound to it
by changing application.properties
. But it is sometimes also useful
to change the default values imperatively in Java, so get more control
over the process. You can do this by declaring a bean of the same
type in your application context, e.g. for the server properties:
@AssertMissingBean(ServerProperties.class)
@Bean
public ServerProperties serverProperties() {
ServerProperties server = new ServerProperties();
server.setPort(8888);
return server;
}
Note the use of @AssertMissingBean
to guard against any mistakes
where the bean is already defined (and therefore might already have
been bound).
Server Configuration
The ServerProperties
are bound to application properties, and
can be used to specify
-
The port that the application listens on for the its endpoints (
server.port
defaults to 8080) -
The address that the application endpoints are available on (
server.address
defaults to all local addresses, making it available to connections from all clients). -
The context root of the application endpoints (
server.context_path
defaults to "/")
Tomcat Container Configuration
If you want to use Tomcat as an embedded container include at least
org.apache.tomcat.embed:tomcat-embed-core
and one of the
org.apache.tomcat.embed:tomcat-embed-logging-*
libraries (depending
on the logging system you are using). Then, in addition to the
generic ServerProperties
, you can also bind server.tomcat.*
properties in the application properties (see
ServerProperties.Tomcat
).
- To enable the Tomcat access log valve (very common in production environments)
More fine-grained control of the Tomcat container is available if you
need it. Instead of letting Spring Bootstrap create the container for
you, just create a bean of type
TomcatEmbeddedServletContainerFactory
and override one of its
methods, or inject some customizations, e.g.
@Configuration
public class MyContainerConfiguration {
@Bean
public TomcatEmbeddedServletContainerFactory tomcatEmbeddedContainerFactory() {
TomcatEmbeddedServletContainerFactory factory = new TomcatEmbeddedServletContainerFactory();
factory.setConnector(new Connector("AJP/1.3"));
}
}
(the default connector uses the Http11NioProtocol
so the example if
overriding that behaviour).
Customizing Management Endpoints
The ManagementProperties
are bound to application properties, and
can be used to specify
-
The port that the application listens on for the management endpoints (defaults to 8080)
-
The address that the management endpoints are available on (if the port is different to the main server port). Use this to listen only on an internal or ops-facing network, for instance, or to only listen for connections from localhost (by specifying "127.0.0.1")
-
The context root of the management endpoints (TODO: does this work?)
The EndpointsProperties
are also bound, and you can use those to
change the paths of the management endpoints, e.g.
endpoints.error.path: /errors/generic
Error Handling
The Actuator provides an /error
endpoint by default that handles all
errors in a sensible way. If you want more specific error pages for
some conditions, the embedded servlet containers support a uniform
Java DSL for customizing the error handling. To do this you have to
have picked a container implementation (by including either Tomcat or
Jetty on the classpath), but then the API is the same. TODO: finish
this.
Customizing Logging
Spring Bootstrap uses SLF4J for logging, but leaves the implementation open. The Starter projects and the Actuator use JDK native logging by default, purely because it is always available. A default configuration file is provided for JDK logging, and also for log4j and logback. In each case there is console output and file output (rotating, 10MB file size).
The various logging systems can be activated by including the right
libraries on the classpath, and further customized by providing a
native configuration file in the root of the classpath, or in a
location specified by the Spring Environment
property
logging.config
.
Logger | Activation | Customization |
---|---|---|
JDK | slf4j-jdk14 | logging.properties |
Logback | logback | logback.xml |
Log4j | slfj4-log4j12, log4j | log4j.properties or log4j.xml |
To help with the customization some other properties are transferred
from the Spring Environment
to System properties:
Environment | System Property | Comments |
---|---|---|
logging.file | LOG_FILE | Used in default log configuration if defined |
logging.path | LOG_PATH | Used in default log configuration if defined |
PID | PID | The current process ID is discovered if possible and not already provided |
All the logging systems supported can consult System properties when
parsing their configuration files. See the defailt configurations in
spring-bootstrap.jar
for examples.
Application Context Initializers
To add additional application context initializers to the Bootstrap
startup process, add a comma-delimited list of class names to the
Environment
property context.initializer.classes
(can be specified
via application.properties
).
Info Endpoint
By default the Actuator adds an /info
endpoint to the main server.
It contains the commit and timestamp information from git.properties
(if that file exists) and also any properties it finds in the
environment with prefix "info".
To populate git.properties
in a
Maven build you can use the excellent
git-commit-id-plugin.
To populate the "info" map all you need to do is add some stuff to
application.properties
, e.g.
info.app.name: MyService
info.app.description: My awesome service
info.app.version: 1.0.0
If you are using Maven you can automcatically populate info properties
from the project using resource filtering. In your pom.xml
you
have (inside the <build/>
element):
<resources>
<resource>
<directory>src/main/resources</directory>
<filtering>true</filtering>
</resource>
</resources>
and then in the application.properties
you can refer to project
properties via placeholders, e.g.
project.artifactId: myproject
project.name: Demo
project.version: X.X.X.X
project.description: Demo project for info endpoint
info.build.artifact: ${project.artifactId}
info.build.name: ${project.name}
info.build.description: ${project.description}
info.build.version: ${project.version}
(notice that in the example we used project.*
to set some values to
be used as fallbacks if the Maven resource filtering has for some
reason not been switched on).
Security - Basic Authentication
To secure your endpoints just add Spring Security Javaconfig to the classpath. By default HTTP Basic authentication will be applied to every request in the main server (and the management server if it is running on the same port). There is a single account by default, and you can test it like this:
$ mvn user:password@localhost:8080/metrics
... stuff comes out
If the management server is running on a different port it is unsecured by default. If you want to secure it you can add a security auto configuration explicitly
Security - HTTPS
Ensuring that all your main endpoints are only available over HTTPS is
an important chore for any application. If you are using Tomcat as a
servlet container, then the Actuator will add Tomcat's own
RemoteIpValve
automatically if it detects some environment settings,
and you should be able to rely on the HttpServletRequest
to report
whether or not it is secure (even downstream of the real SSL
termination endpoint). The standard behaviour is determined by the
presence or absence of certain request headers ("x-forwarded-for" and
"x-forwarded-proto"), whose names are conventional, so it should work
with most front end proxies. You switch on the valve by adding some
entries to application.properties
, e.g.
server.tomcat.remote_ip_header: x-forwarded-for
server.tomcat.protocol_header: x-forwarded-proto
(The presence of either of those properties will switch on the
valve. Or you can add the RemoteIpValve
yourself by adding a
TomcatEmbeddedServletContainerFactory
bean.)
Spring Security can also be configured to require a secure channel for
all (or some requests). To switch that on in an Actuator application
you just need to set security.require_https: true
in
application.properties
.
Audit Events
The Actuator has a flexible audit framework that will publish events once Spring Security is in play (authentication success and failure and access denied exceptions by default). This can be very useful for reporting, and also to implement a lock-out policy based on authentication failures.
You can also choose to use the audit services for your own business
events. To do that you can either inject the existing
AuditEventRepository
into your own components and use that directly,
or you can simply publish AuditApplicationEvent
via the Spring
ApplicationContext
(using ApplicationEventPublisherAware
).
Metrics Customization
Metrics come out on the /metrics
endpoint. You can add additional
metrics by injecting a MetricsRepository
into your application
components and adding metrics whenever you need to. To customize the
MetricsRepository
itself, just add a bean definition of that type to
the application context (only in memory is supported out of the box
for now).
Customizing the Health Indicator
The application always tells you if it's healthy via the /health
endpoint. By default it just responds to a GET witha 200 status and a
plain text body containing "ok". If you want to add more detailed
information (e.g. a description of the current state of the
application), just add a bean of type HealthIndicator
to your
application context, and it will take the place of the default one.