I have configured my application with config server and github supported external config files. It works fine when I am having single instance of my application in cloud foundry.
But for multiple instance it is said to implement spring cloud bus to apply external config changes to all the instances. For this I have bind my config server with rabbit MQ instance available on Pivotal Cloud foundry. have added spring.cloud.starter.bus.amqp jar in my build.gradle file.
Problem: But when I am hitting POST request to client app at:
http://server:port/bus/refresh the call goes to controller rather than refreshing all the instances and failing as no mapping for same.
Please let me know if I am missing any configuration to make spring-cloud-bus work.
Thanks in advance!
application.properties(Client application):
spring.profiles=cloud
spring.rabbitmq.host= 10.17.128.102
spring.rabbitmq.port= 5672
spring.rabbitmq.virtual-host= *****
spring.rabbitmq.username= ******
spring.rabbitmq.password= *****
rabbit.exchangeName= demoPartyServiceexc
rabbit.routingKey= demoPartyService
rabbit.queueName= demoPartyServicequeue
logging.level.ROOT= ERROR
bootstrap.properties(Client application):
spring.application.name=demo-api
spring.cloud.bus.enabled=true
spring.cloud.config.bus.enabled=true
spring.cloud.bus.amqp.enabled=true
spring.cloud.bus.refresh.enabled=true
spring.cloud.bus.env.enabled=true
spring.cloud.config.uri=https://config-a5e99419-8179-47f7-8c23-62ed5b38db0b.cf.com
spring.cloud.config.server.bootstrap=true
spring.cloud.config.server.git.uri= My Github repository URI
spring.cloud.config.server.git.username= ********
spring.cloud.config.server.git.password= ********
application.properties file in GIT repo:
logging.level.ROOT=WARN
What are server.servlet-path and management.context-path in client app set to?
I think you might need to send the request to http://host/<management.context-path>/bus/refresh.
Or better yet, to http://config-server-host/monitor so that it publishes a message to a topic in RabbitMQ and all client apps get notified.
Configuration snippets, source code and more details could be found at my blog post: Refreshable Configuration using Spring Cloud Config Server, Spring Cloud Bus, RabbitMQ and Git
Related
I am creating a spring boot microservice project with intelij IDEA.
Currently I have developed three seperate spring boot rest services as customer service, vehicle service and spring cloud config server. Spring cloud config server is pointing to a github repository.
The issue is sometimes above projects take more than 10 minutes to run and sometimes does't run and give an error message as "failed to check application readystate intellij attached provider for the vm is not found". I have no idea why this happens ?
There are two possible causes:
1. IntelliJ IDEA and the Spring application are running in different JVMs.
There is a bug for IntelliJ IDEA regarding that:
https://youtrack.jetbrains.com/issue/IDEA-210665
Here is short summary:
IntelliJ IDEA uses local JMX connector for retrieving Spring Boot actuator endpoint's data by default. However, it could be impossible to get local JMX connector address via attach api if Spring Boot application and IntelliJ IDEA are run by different JVMs. In this case, add the following lines to VM options of your Spring Boot run configuration:
-Dcom.sun.management.jmxremote.port={some_port}
-Dcom.sun.management.jmxremote.authenticate=false
-Dcom.sun.management.jmxremote.ssl=false
As mentioned in the official Oracle documentation, this configuration is insecure. Any remote user who knows (or guesses) your port number and host name will be able to monitor and control your Java applications and platform.
2. Prolonged time to retrieve local hostname
You can check that time using inetTester. Normally it should take only several milliseconds to complete. If it takes a long time then you can add the hostname returned by inetTester to /etc/hosts file like this:
127.0.0.1 localhost winsky
::1 localhost winsky
I am new to Spring Config Server/Client technologies.
I am using a spring config server to hold some config values.
Config clients will connect to the server and get the values.
If i change some of the config values at the config server, then currently I have to refresh the clients to load the config details from config server again by invoking "/refresh" on each client.
Is there anyway the clients will be notified by the config server and they will then reload the configuration again ?
Yes there is a way.
The solution is to use the Spring Cloud Bus. Using this module, you would link multiple clients to the server using a message broker. The only message broker implementation currently supported by this module is AMQP. Once the clients are connected to the server, invoking the endpoint on the server /bus/refresh will automatically broadcast the configuration changes to all the subscribed clients. This therefore means it is possible to reload configuration changes for any number of clients with one single refresh request which originates at the server.
I'm deploying Spring Cloud Dataflow 1.4.0.RELEASE on my org's Cloud Foundry. I took the CF deployable server and used a manifest to deploy it. We use it mostly for tasks developed using Spring Cloud Task. In the documentation (that uses PCF) suggests this configuration:
SPRING_CLOUD_DEPLOYER_CLOUDFOUNDRY_TASK_SERVICES my_mysql
Where my_mysql is the DB to keep track of task related information. Documentation suggests using this as CF's cups but that won't work since my org uses Vault to store keys and secrets so DB connections are not allowed in such way.
Since I did not provided any value for SPRING_CLOUD_DEPLOYER_CLOUDFOUNDRY_TASK_SERVICES in my manifest when I start SPDF it throws an error indicating that bean "taskService" could not be created.
Haven't found a workaround for this. Is this the only way to specify DB connection for tasks in version 1.4?
You can also pass a deployment property when launching the task:
task launch --name mytask --properties "deployer.mytask.cloudfoundry.services=-Duser.timezone=my_mysql"
This way, you don't have to use the server level environment property SPRING_CLOUD_DEPLOYER_CLOUDFOUNDRY_TASK_SERVICES
I'm going to use Spring Config Service (SCS) for our Microservices Architecture.
Currently our Cloud stack is on AWS.
Since SCS will run on a Docker, thanks to a Pipeline + Cloud Formation, and our config repository will be on a private GitHub repository with encrypted values:
Is there any best practice to refresh the repository that will be "pulled" inside the Docker?
How can I update it on any instances? (since my service will be load balanced with HA).
pleas refer following poc
https://github.com/pooja-varma/cloud-config-and-eureka-server
may be it helps you
Config clients don't poll for changes. It has to be triggered and there is a EnvironemntChanged listened by the application and any changes in the properties will be loaded again. If you need more control over when it has to be refreshed and if you want that to be atomic I would recommend you to use #RefreshScope which are lazy proxies and initialized only when they are used. The Environment of your application is pulled every time and it also actuator endpoints to the rescue as well.
Please refer the documentation here.
http://cloud.spring.io/spring-cloud-static/docs/1.0.x/spring-cloud.html#_refresh_scope
I am trying to register a spring cloud task jar which is residing in a ftp server , from a spring cloud data flow container which is running on PCF.
Can somebody help me in terms of how to do a "app register" from a ftp location from SCDF if SCDF supports that ? How do i register a ftp location as remote repository with user name & passwd ?
Thanks!
Nilanjan
As discussed in the email, currently, in SCDF, we support file://, maven://, http://, docker://, and hdfs:// as the file resolution strategies. Besides, for cloud platforms, it'd be recommended that you resolve and register artifacts from maven or docker to be more resilient.
If you're running SCDF on PCF, you could bind the SCDF-server to volume-services. By having all the custom apps in volume-services, SCDF-server could then resolve artifacts from the designated directory via file:// resolution - this is another good alternative for PCF deployments.