Invoking /management/refresh on service endpoints results in service getting de-registered.

[devlanp]

Overview of the issue

Invoking /refresh on service endpoints results in service getting de-registered.

Motivation for or Use Case

Was trying to reload the configurations using @RefreshScope

Reproduce the error

Bring up Registry
Bring up any service
Call /management/refresh on the service.

Related issues

#421

Version

spring-cloud-starter (1.1.3.Release)

[ryanjbaxter]

Did you say you had a sample app to reproduce the issue?

[devlanp]

@ryanjbaxter @dsyer Thanks for your help.

Following are the details.

I m using the Jhipster (https://jhipster.github.io/) stack to develop micro service.

Following are the steps to reproduce:

Registry:
https://github.com/jhipster/jhipster-registry

Service:
Attached.
test.zip

Steps to reproduce:

1. Start registry,
2. Start service.
3. Place a post request to the endpoint: http://localhost:8081/management/refresh
4. The service logs would read:
   Shutting down DiscoveryClient ...
   Unregistering ...
   DiscoveryClient_TEST/test:8e580d0c17f4b27396089018c4b32b98 - deregister status: 200

I'll keep the thread posted on my findings. Let me know if I missed anything.

[spencergibb]

Using a very old version Brixton.SR7

[devlanp]

@spencergibb I see this happening even with Camden.SR4. What am I missing then?

[ryanjbaxter]

I bumped the version to Camden.SR5 and made a POST to http://localhost:8081/management/refresh. Everything worked as expected.

2017-02-23 13:15:36.775  INFO 44332 --- [  XNIO-2 task-4] com.netflix.discovery.DiscoveryClient    : Shutting down DiscoveryClient ...
2017-02-23 13:15:36.776  INFO 44332 --- [  XNIO-2 task-4] com.netflix.discovery.DiscoveryClient    : Unregistering ...
2017-02-23 13:15:36.781  INFO 44332 --- [  XNIO-2 task-4] com.netflix.discovery.DiscoveryClient    : DiscoveryClient_TEST/test:217780f8b59a18e25946dbd6970a7f1c - deregister  status: 200
2017-02-23 13:15:36.782  INFO 44332 --- [  XNIO-2 task-4] com.netflix.discovery.DiscoveryClient    : Completed shut down of DiscoveryClient
2017-02-23 13:15:36.786  INFO 44332 --- [  XNIO-2 task-4] c.n.d.provider.DiscoveryJerseyProvider   : Using JSON encoding codec LegacyJacksonJson
2017-02-23 13:15:36.786  INFO 44332 --- [  XNIO-2 task-4] c.n.d.provider.DiscoveryJerseyProvider   : Using JSON decoding codec LegacyJacksonJson
2017-02-23 13:15:36.786  INFO 44332 --- [  XNIO-2 task-4] c.n.d.provider.DiscoveryJerseyProvider   : Using XML encoding codec XStreamXml
2017-02-23 13:15:36.786  INFO 44332 --- [  XNIO-2 task-4] c.n.d.provider.DiscoveryJerseyProvider   : Using XML decoding codec XStreamXml
2017-02-23 13:15:36.892  INFO 44332 --- [  XNIO-2 task-4] c.n.d.s.r.aws.ConfigClusterResolver      : Resolving eureka endpoints via configuration
2017-02-23 13:15:36.893  INFO 44332 --- [  XNIO-2 task-4] com.netflix.discovery.DiscoveryClient    : Disable delta property : false
2017-02-23 13:15:36.893  INFO 44332 --- [  XNIO-2 task-4] com.netflix.discovery.DiscoveryClient    : Single vip registry refresh property : null
2017-02-23 13:15:36.893  INFO 44332 --- [  XNIO-2 task-4] com.netflix.discovery.DiscoveryClient    : Force full registry fetch : false
2017-02-23 13:15:36.893  INFO 44332 --- [  XNIO-2 task-4] com.netflix.discovery.DiscoveryClient    : Application is null : false
2017-02-23 13:15:36.893  INFO 44332 --- [  XNIO-2 task-4] com.netflix.discovery.DiscoveryClient    : Registered Applications size is zero : true
2017-02-23 13:15:36.893  INFO 44332 --- [  XNIO-2 task-4] com.netflix.discovery.DiscoveryClient    : Application version is -1: true
2017-02-23 13:15:36.893  INFO 44332 --- [  XNIO-2 task-4] com.netflix.discovery.DiscoveryClient    : Getting all instance registry info from the eureka server
2017-02-23 13:15:36.897  INFO 44332 --- [  XNIO-2 task-4] com.netflix.discovery.DiscoveryClient    : The response status is 200
2017-02-23 13:15:36.898  INFO 44332 --- [  XNIO-2 task-4] com.netflix.discovery.DiscoveryClient    : Starting heartbeat executor: renew interval is: 30
2017-02-23 13:15:36.898  INFO 44332 --- [  XNIO-2 task-4] c.n.discovery.InstanceInfoReplicator     : InstanceInfoReplicator onDemand update allowed rate per min is 4
2017-02-23 13:15:36.899  INFO 44332 --- [  XNIO-2 task-4] com.netflix.discovery.DiscoveryClient    : Discovery Client initialized at timestamp 1487873736899 with initial instances count: 1
2017-02-23 13:15:36.900  WARN 44332 --- [  XNIO-2 task-4] com.netflix.discovery.DiscoveryClient    : Saw local status change event StatusChangeEvent [timestamp=1487873736900, current=UP, previous=DOWN]
2017-02-23 13:15:36.901  INFO 44332 --- [nfoReplicator-0] com.netflix.discovery.DiscoveryClient    : DiscoveryClient_TEST/test:147781fb281c4c00c262233092ee3fbc: registering service...
2017-02-23 13:15:36.912  INFO 44332 --- [nfoReplicator-0] com.netflix.discovery.DiscoveryClient    : DiscoveryClient_TEST/test:147781fb281c4c00c262233092ee3fbc - registration status: 204

[devlanp]

I see the following traces and doubt if it is correct! Unregistering, followed by re-register ?

[ XNIO-2 task-4] com.netflix.discovery.DiscoveryClient : Unregistering ...
[ XNIO-2 task-4] com.netflix.discovery.DiscoveryClient : DiscoveryClient_TEST/test:217780f8b59a18e25946dbd6970a7f1c - deregister status: 200
[ XNIO-2 task-4] com.netflix.discovery.DiscoveryClient : Saw local status change event StatusChangeEvent [timestamp=1487873736900, current=UP, previous=DOWN]

[ryanjbaxter]

Why wouldn't you expect that? You are restarting the app, so it makes sense that the app would unregister with Eureka when shutting down and then re-register once it starts back up.

[devlanp]

Ryan, feel free to correct my understanding. The issue is in /management/refresh call which is just the configuration refresh and I m not expecting the service to get registered and de-registered and the log says it has de-registered. Am I missing anything?

[ryanjbaxter]

I believe it has to do with the fact that calling /refresh will cause all the beans to be recreating. Since the DiscoveryClient class is registered as a bean it will be destroyed and then recreated. When the bean is being destroyed this will cause the shutdown method to be called.

[devlanp]

#421. Going by this, I thought its an issue and it is already fixed.

[ryanjbaxter]

#421 was about PERMANENTLY deregistering the application. In you case it deregisters the application and then reregisters it. So it seems to be working as expected.

[devlanp]

Ok. Thanks for the clarification.