These are chat archives for spring-cloud/spring-cloud

19th
Sep 2016
vbalaut
@vbalaut
Sep 19 2016 02:36
@joshlong Thanks
pradeepkusingh
@pradeepkusingh
Sep 19 2016 04:03
@dsyer : I ma seeing this error in APIGateway logs and seeing lot of 502
xOUT Mon Sep 19 03:59:57 UTC 2016 TRACE [APP:null TRACKING_ID:null ACCOUNT_NUM:null] org.springframework.web.servlet.mvc.method.annotation.ServletInvocableHandlerMethod - Method [error] returned [<500 Internal Server Error,{timestamp=Mon Sep 19 03:59:57 UTC 2016, status=500, error=Internal Server Error, exception=com.netflix.zuul.exception.ZuulException, message=GENERAL},{}>]
This one also
org.springframework.cloud.netflix.zuul.filters.post.SendErrorFilter - Error during filtering
this is a vanilla ZUUL
no filter
need your help to get some details on it
Dominik Gätjens
@dcg
Sep 19 2016 07:19
are there any known issues with ribbon +spring-dev-tools auto-restart? got an strange Exception after the automatic restart Caused by: java.util.concurrent.RejectedExecutionException: Task java.util.concurrent.ScheduledThreadPoolExecutor$ScheduledFutureTask@4172bee1 rejected from java.util.concurrent.ScheduledThreadPoolExecutor@79efa849[Terminated, pool size = 0, active threads = 0, queued tasks = 0, completed tasks = 1]
ccit-spence
@ccit-spence
Sep 19 2016 07:40
@dcg I do remember seeing that error. I am using Spring Boot 1.4.0 now and have not seen it
Dominik Gätjens
@dcg
Sep 19 2016 07:42
@ccit-spence thanks atm i'm trying to upgrade to 1.4 :)
ccit-spence
@ccit-spence
Sep 19 2016 07:44
I am using Brixton SR5 and 1.4. I am working on a project right now and can use devtools without the error
Dominik Gätjens
@dcg
Sep 19 2016 07:45
cool, im on 1.3.2 ans Birxton.SR4 i'll try to upgrade
ccit-spence
@ccit-spence
Sep 19 2016 07:45
The only thing I know of as a bug is the HAL browser will not work with 1.4.0. It is a known bug.
Dominik Gätjens
@dcg
Sep 19 2016 07:46
ah ok, that one i dont need, i'm using postmane+copy&paste ;D
Dave Syer
@dsyer
Sep 19 2016 07:47
@pradeepkusingh not enough information yet. Do you have a stack trace or something?
A sample app?
Dominik Gätjens
@dcg
Sep 19 2016 07:54
btw doesn anyone know the difference in eclipse between run as application and run as spring-boot application
Dave Syer
@dsyer
Sep 19 2016 07:56
You get a slightly different launcher. The Spring Boot one appears on the Boot dashboard, and has special selectors for active profiles and property overrides.
Plus the console has ANSI colours.
The code that runs in the JVM is the same
Dominik Gätjens
@dcg
Sep 19 2016 07:57
ok good, than it's not related to my problem... after upgrading from spring boot 1.3.2 to 1.4 my app can't fetch services from eureka ...
ccit-spence
@ccit-spence
Sep 19 2016 08:05
@dsyer Is it possible to add a link to a response from RestTemplate calling an SDR endpoint. The link needs to be from Eureka. so instead of Link link = new Link("http://localhost:8080/something"); it needs to be Link link = new Link("http://service-name/something");
Do I need to make a separate call with RestTemplate to get the current link location?
Or go as far as to create a service that gets that link, then add the link
Dave Syer
@dsyer
Sep 19 2016 08:08
@ccit-spence I don't know what the best way to do that is really. You're going to end up scrubbing JSON
And it's going to suck.
I guess if you know it's from a Spring HATEOAS app you know about the structure and you can use the Resource abstraction
Did you know about DiscoverableResource?
Seems like it's in the same area
ccit-spence
@ccit-spence
Sep 19 2016 08:10
No did not know about that
will take a look
the link I am adding is not HATEOAS, the problem is the discovery part
I am adding it to a HATEOAS response though
Bascially it is a client in front of an SDR endpoint that transforms the data with additional links from other services
ccit-spence
@ccit-spence
Sep 19 2016 08:15
@dsyer Did you mean DiscoveredResource can’t find DiscoverableResource
Dominik Gätjens
@dcg
Sep 19 2016 08:42
on auto-restart spring starts to build my singelton beans again, but there seems to be stuff registered from the first run org.springframework.beans.factory.BeanCreationException: Error creating bean with name 'hystrixMDCCopier' defined in class path resource [MyClass.class]: Bean instantiation via factory method failed; nested exception is org.springframework.beans.BeanInstantiationException: Failed to instantiate [SecurityHystrixMDCCopier]: Factory method 'hystrixMDCCopier' threw exception; nested exception is java.lang.IllegalStateException: Another strategy was already registered.
Dominik Gätjens
@dcg
Sep 19 2016 08:59
guess its just bad config on my side
ccit-spence
@ccit-spence
Sep 19 2016 09:23
@dsyer I have been reading a bit about DynamicServiceInstanceProvider DiscoveredResource etc… Is that all new stuuf to make the customer-stores sample better?
Seems to eliminate a lot of code
Dave Syer
@dsyer
Sep 19 2016 09:23
I haven't used it. Ollie contributed it.
So I guess it might have been folded into the sample?
ccit-spence
@ccit-spence
Sep 19 2016 09:24
Am I understanding the intended purpose correctly? Speaking specifically about the integration part of the customer stores example
Dave Syer
@dsyer
Sep 19 2016 09:24
@dcg it's quite likely that hystrix doesn't play nice with Spring Boot devtools.
@ccit-spence probably. I didn't really have anything to do with that. Maybe we can summon @olivergierke
ccit-spence
@ccit-spence
Sep 19 2016 09:25
FYI, The project I have up now is using hystrixcommands with no issue with devtools
my project is fairly minimal however
Dominik Gätjens
@dcg
Sep 19 2016 09:31
i seem to have fixed the hystrix problem but now i'm suffering with multiple creation of EhCacheManager
ccit-spence
@ccit-spence
Sep 19 2016 09:49
@dsyer Am I correct that if I need multiple beans of DiscoveredResource I would just add a Qualifier? I think I am starting to make sense of how it was intended to work.
It is just a Bean so Qualifier should apply like any other Bean is my assumption
ccit-spence
@ccit-spence
Sep 19 2016 09:55
@olivergierke Guessing that DiscoveredResource requires a HAL endpoint?
pradeepkusingh
@pradeepkusingh
Sep 19 2016 12:15
@dsyer I am seeing only this much info as debug logs. It doesn't show all stack trace. I emanled zuul.debug.request also
still not much info in logs
Dominik Gätjens
@dcg
Sep 19 2016 12:34
somebody have an idea why Eureka-Discovery doesnt work if i start the App via Eclipse 14:35:42 WARN DiscoveryEnabledNIWSServerList - EurekaClient has not been initialized yet, returning an empty list {} if i start with gradle bootrun everything works fine
pradeepkusingh
@pradeepkusingh
Sep 19 2016 12:40
This message was deleted
[d.comcast.net-1] c.n.e.cluster.ReplicationTaskProcessor : Network level connection to peer sparrow-eureka-server-prod.g1.net; retrying after delay
Mon Sep 19 2016 06:30:54 GMT-0600 (MDT) [App/1] OUT at com.netflix.eureka.cluster.DynamicGZIPContentEncodingFilter.handle(DynamicGZIPContentEncodingFilter.java:48) ~[eureka-core-1.4.10.jar!/:1.4.10]
Mon Sep 19 2016 06:30:54 GMT-0600 (MDT) [App/1] OUT at com.sun.jersey.client.apache4.ApacheHttpClient4Handler.handle(ApacheHttpClient4Handler.java:187) ~[jersey-apache-client4-1.19.1.jar!/:1.19.1]
Mon Sep 19 2016 06:30:54 GMT-0600 (MDT) [App/1] OUT at com.sun.jersey.api.client.WebResource$Builder.post(WebResource.java:570) ~[jersey-client-1.19.1.jar!/:1.19.1]
Mon Sep 19 2016 06:30:54 GMT-0600 (MDT) [App/1] OUT at com.netflix.eureka.util.batcher.TaskExecutors$BatchWorkerRunnable.run(TaskExecutors.java:187) [eureka-core-1.4.10.jar!/:1.4.10]
Mon Sep 19 2016 06:30:54 GMT-0600 (MDT) [App/1] OUT at com.netflix.eureka.cluster.ReplicationTaskProcessor.process(ReplicationTaskProcessor.java:71) ~[eureka-core-1.4.10.jar!/:1.4.10]
Mon Sep 19 2016 06:30:54 GMT-0600 (MDT) [App/1] OUT at com.netflix.discovery.EurekaIdentityHeaderFilter.handle(EurekaIdentityHeaderFilter.java:27) ~[eureka-client-1.4.10.jar!/:1.4.10]
Dave Syer
@dsyer
Sep 19 2016 13:21
@dcg if that's the only clue you have it's a bit thin
Maybe you can see from the /env endpoint that there is a difference in the two apps (the one you run from the command line and the one you run from the IDE)?
Dominik Gätjens
@dcg
Sep 19 2016 13:29
@dsyer i will have a look atm i can tell that in LegacyEurekaClientProvider DiscoveryManager.getInstance().getDiscoveryClient(); returns null if this rings any Bells... it seems like @EnableEurekaClient gets ignored
Dave Syer
@dsyer
Sep 19 2016 13:30
@EnableDiscoveryClient is preferred I think
but I don't know why that would change anything
Looks like a lifecycle issue to me. Your app wants to access the DiscoveryManager before it has been configured
But I really don't know why
You might have to strip down your app to a minimal sample to understand it
A break point in DiscoveryManager.getInstance().getDiscoveryClient()would tell you where it is being called. That might be useful to know.
Dominik Gätjens
@dcg
Sep 19 2016 13:34
its calles when i call my webservice
so the app is fully initalized
Dave Syer
@dsyer
Sep 19 2016 13:34
OK, that didn't help then.
Dominik Gätjens
@dcg
Sep 19 2016 13:34
the app doesnt register in eureka, either
Dave Syer
@dsyer
Sep 19 2016 13:35
For the same reason probably
Dominik Gätjens
@dcg
Sep 19 2016 13:35
yes i think so too
the eureka-url belongs in to the bootstrap.properties?
Dave Syer
@dsyer
Sep 19 2016 13:36
Not necessarily
Only if you need eureka at bootstrap
You don't normally
Spencer Gibb
@spencergibb
Sep 19 2016 13:37
DiscoveryManager.getInstance().getDiscoveryClient() isn't used anymore, it's gone away from singleton use
Dave Syer
@dsyer
Sep 19 2016 13:38
Good point. I guess we need to ask: what version of spring cloud netflix?
Dominik Gätjens
@dcg
Sep 19 2016 13:41
hmm... '| +--- org.springframework.cloud:spring-cloud-netflix-core:1.1.5.RELEASE'
i'm using mavenBom "org.springframework.cloud:spring-cloud-starter-parent:Brixton.SR5"
Dave Syer
@dsyer
Sep 19 2016 13:46
That's odd then. Maybe the difference between command line and IDE is the classpath.
Messed up in IDE I guess.
Dominik Gätjens
@dcg
Sep 19 2016 13:48
accidentally i added spring.cloud.config.discovery.enabled=true and it resolves the problem o.0
Spencer Gibb
@spencergibb
Sep 19 2016 13:57
that doesn’t make sense
Dominik Gätjens
@dcg
Sep 19 2016 13:57
yep it doesn'T but without this property EurekaClientAutoConfiguration doesnt trigger
but i can't see why
Spencer Gibb
@spencergibb
Sep 19 2016 13:59
we’d need a sample project that reproduces the problem to dig in more.
Dominik Gätjens
@dcg
Sep 19 2016 14:01
"eureka.client.enabled=true" solves the problem, too. this makes more sense, but i don't really get why it works with gradle bootrun without the prop
omg i'm just stupid
i'm sorry for wasting your time -.- application.propeties from /src/integrationTest got picked up by the eclipse-run and there is eureka.client.enabled=false ....
Spencer Gibb
@spencergibb
Sep 19 2016 14:04
:stuck_out_tongue_winking_eye:
Dominik Gätjens
@dcg
Sep 19 2016 14:04
what a waste of half a day
:worried:
pradeepkusingh
@pradeepkusingh
Sep 19 2016 15:16
@spencergibb - we are trying to use spring boot 1.3.7 and seeing this issue
Could not resolve all dependencies for configuration ':compile'.
15:32:16.145 > Could not find org.springframework.data:spring-data-keyvalue:1.0.4.RELEASE.
15:32:16.145 Required by:
15:32:16.145 com.comcast.activation.sparrow:glass-commit:0.0.1-SNAPSHOT > org.springframework.data:spring-data-redis:1.7.1.RELEASE
of s-d-keyvalue
pradeepkusingh
@pradeepkusingh
Sep 19 2016 15:19
i know that , but this is coming as dependencies
but I am not specifiying it in gradle
part of spring data dependencies
Spencer Gibb
@spencergibb
Sep 19 2016 15:20
s-d-redis says it depends on 1.1.1.RELEASE of s-d-keyvalue
running with mvn -X would tell you where the version is being overridden, not sure with gradle.
pradeepkusingh
@pradeepkusingh
Sep 19 2016 15:25
Just saw it is known issue
spring-projects/spring-boot#5894
Spencer Gibb
@spencergibb
Sep 19 2016 15:25
yeah, didn’t think it had anything to do with spring cloud
pradeepkusingh
@pradeepkusingh
Sep 19 2016 15:31
Thank you :)
Another question.. I am trying to use Brixton rc5 and seeing this error
ERR Caused by: org.springframework.beans.factory.BeanCreationException: Error creating bean with name 'hystrixStreamTask' defined in class path resource [org/springframework/cloud/netflix/hystrix/stream/HystrixStreamAutoConfiguration.class]: Initialization of bean failed; nested exception is java.lang.IllegalStateException: Encountered invalid @Scheduled method 'sendMetrics': Invalid fixedRateString value "120000 " - cannot parse into integer
which looks strange
with Angel sr6, same apps runs fine
Spencer Gibb
@spencergibb
Sep 19 2016 15:37
@pradeepkusingh no idea. it’s just a placeholder. Unless there is a spring version mismatch
pradeepkusingh
@pradeepkusingh
Sep 19 2016 16:14
@dsyer 502 Bad gateway . Registered endpoint failed to handle the request .. returnin response to misbehaving host
what does this mean..
we are seeing frequently using ZUUL+Eureka
and we are in cloud foundry and don't see apps are unhealthy
Spencer Gibb
@spencergibb
Sep 19 2016 16:16
that message doesn’t originate from zuul
pradeepkusingh
@pradeepkusingh
Sep 19 2016 16:16
Registered endpoint failed to handle the request
it seems to me ribbon is giving this error ?
Registered endpoint failed to handle the request ..
Spencer Gibb
@spencergibb
Sep 19 2016 16:17
I don’t think so
pradeepkusingh
@pradeepkusingh
Sep 19 2016 16:18
so from where this eror coming any idea ?
looks like CF
Please do 'cf logs <appname> --recent'. A 502 message often means that the application crashed and the logs can help reveal why.
pradeepkusingh
@pradeepkusingh
Sep 19 2016 16:32
Thanks Spencer..
@spencer and @dsyer : We are in comcast and having bit issues with Eureka and Zuul configuartion
and seeing 502 bad gateway error
I am ready to open a support ticket and need your help to debug further
what kind of information you are looking for let me know
2016-09-18T23:28:55.10-0400 [RTR/7] OUT activate-service.g.comcast.net - [19/09/2016:03:28:50 +0000] "POST /ServicesOrderRs/PrepareServicesOrder HTTP/1.1" 502 67 "-" "Apache-HttpClient/4.3.6 (java 1.5)" 69.252.194.146:60346 x_forwarded_for:"96.114.156.9, 69.252.194.146" vcap_request_id:36237f75-eeaa-4b6e-4361-2bd24981e546 response_time:5.041099364 app_id:e1f9dd60-ee9d-4e52-b4ae-7e75972f446c
lot of these kind of errors
Spencer Gibb
@spencergibb
Sep 19 2016 16:38
@pradeepkusingh where is that log line from?
pradeepkusingh
@pradeepkusingh
Sep 19 2016 16:39
this is Cloud foundry access logs
and ZUUL side we are seeing these errors
Spencer Gibb
@spencergibb
Sep 19 2016 16:40
for what?
for zuul?
pradeepkusingh
@pradeepkusingh
Sep 19 2016 16:41
xOUT Mon Sep 19 03:59:57 UTC 2016 TRACE [APP:null TRACKING_ID:null ACCOUNT_NUM:null] org.springframework.web.servlet.mvc.method.annotation.ServletInvocableHandlerMethod - Method [error] returned [<500 Internal Server Error,{timestamp=Mon Sep 19 03:59:57 UTC 2016, status=500, error=Internal Server Error, exception=com.netflix.zuul.exception.ZuulException, message=GENERAL},{}>]
in zuul we are seeing above errors
I am uanble to get more details .. though I haev emanled zuul.debug.request = true
logging:
level:
ROOT: TRACE
org.springframework.web: TRACE
com.comcast.sparrow.apigateway: TRACE
com.netflix.ribbon: TRACE
enabled everything without any luck ..
we want to know if down stream services or bad then why can't we haev more details
Please help . .we are in big mess right now
and having prod issues daily..
pradeepkusingh
@pradeepkusingh
Sep 19 2016 16:48
Let us know if we need to enable anything else
Dave Syer
@dsyer
Sep 19 2016 16:59
What can you see in the /routes endpoint?
Does it match the eureka service catalog?
pradeepkusingh
@pradeepkusingh
Sep 19 2016 17:00
1 minute
Dave Syer
@dsyer
Sep 19 2016 17:00
(It looks like the backend services are down, or registered with the wrong route)
pradeepkusingh
@pradeepkusingh
Sep 19 2016 17:01
we have only 1 route define for that app
but it has 8 instance running behind
Dave Syer
@dsyer
Sep 19 2016 17:01
And is it registered in eureka with the correct hostname?
pradeepkusingh
@pradeepkusingh
Sep 19 2016 17:02
yes.. let me show registry

<application>

<name>SERVICES-ORDER-PROD-A</name>

<instance>

<instanceId>
services-order-rs-main.g2.app.:c0d0b4402b1e42c9a83a95ad317dbbb3
</instanceId>

<hostName>services-order-rs-main.g2.t</hostName>

<app>SERVICES-ORDER-PROD-A</app>

<ipAddr>10.254.2.2</ipAddr>

<status>UP</status>

<overriddenstatus>UNKNOWN</overriddenstatus>

<port enabled="true">80</port>

<securePort enabled="false">443</securePort>

<countryId>1</countryId>

/eureka/apps
I remvoed comcast specific domain from above while posting in public network
i can see same for 8 ..
Dave Syer
@dsyer
Sep 19 2016 17:04
Not sure what you mean by that. C
Can you curl that route?
pradeepkusingh
@pradeepkusingh
Sep 19 2016 17:05
sure
Dave Syer
@dsyer
Sep 19 2016 17:06
Then I don't think there's anything wrong with your spring apps
Maybe there's an infrastructure problem?
Ask you network admins
s/you/your/
pradeepkusingh
@pradeepkusingh
Sep 19 2016 17:09
yes, let me curl that
pradeepkusingh
@pradeepkusingh
Sep 19 2016 17:18
curl is fine.. no issue
but problem with this is I don't know if one node out of 8 is bad
Dave Syer
@dsyer
Sep 19 2016 17:20
Could be. But that would show up in a curl as well.
pradeepkusingh
@pradeepkusingh
Sep 19 2016 17:21
yes. That is a good idea
Dave Syer
@dsyer
Sep 19 2016 17:21
If the 502 is intermittent I'd say that's exactly what it is: the instances are being recycled
Maybe they are crashing
Their logs will show you
pradeepkusingh
@pradeepkusingh
Sep 19 2016 17:22
we saw few crash.. but not all time and see same issue
Dave Syer
@dsyer
Sep 19 2016 17:22
It's really hard to understand what the issue is still
Al we know is there are sometimes some 502 responses from the router
Sounds like the app instances are crashing
pradeepkusingh
@pradeepkusingh
Sep 19 2016 17:24
so how to configure ribbon to retry on these situations ?
Dave Syer
@dsyer
Sep 19 2016 17:24
What version of spring cloud?
pradeepkusingh
@pradeepkusingh
Sep 19 2016 17:24
Angel sr6..
Dave Syer
@dsyer
Sep 19 2016 17:25
I think that has the retrying http client by default
pradeepkusingh
@pradeepkusingh
Sep 19 2016 17:25
and we can upgrade to Brixton if needed
Dave Syer
@dsyer
Sep 19 2016 17:25
We need @ryanjbaxter to fill you in on that
pradeepkusingh
@pradeepkusingh
Sep 19 2016 17:25
ok
Spencer Gibb
@spencergibb
Sep 19 2016 17:25
zuul uses the ribbon client
Dave Syer
@dsyer
Sep 19 2016 17:25
But you also should try and understand why the apps are crashing
pradeepkusingh
@pradeepkusingh
Sep 19 2016 17:26
yes, we are working on it..
but still in prod it is taking some time to reached to root cause and fix it..
ribbon:
ReadTimeout: 60000
ConnectTimeout: 60000
MaxAutoRetries: 2
MaxAutoRetriesNextServer: 1
I haev configured these
don't know if it can help
Spencer Gibb
@spencergibb
Sep 19 2016 17:27
it should, it will only retry GET.
pradeepkusingh
@pradeepkusingh
Sep 19 2016 17:27
aah..what about post ?
Dave Syer
@dsyer
Sep 19 2016 17:27
I think this one is a POST
Ryan has some ideas about how to make it retry everything
He's on PTO today I think
pradeepkusingh
@pradeepkusingh
Sep 19 2016 17:28
hmm.. do we have any article or any where .. I can use it ?
Spencer Gibb
@spencergibb
Sep 19 2016 17:29
ribbon.OkToRetryOnAllOperations=true
pradeepkusingh
@pradeepkusingh
Sep 19 2016 17:29
Tha is also true.. my bad last time ..it copy paste error
ReadTimeout: 300000
ConnectTimeout: 60000
MaxAutoRetries: 2
MaxAutoRetriesNextServer: 1
ServerListRefreshInterval: 10000
EnableZoneAffinity: true
OkToRetryOnAllOperations: true
pradeepkusingh
@pradeepkusingh
Sep 19 2016 17:44
One qq
What this tell us ?
Mon Sep 19 15:28:08 UTC 2016 TRACE [APP:null TRACKING_ID:null ACCOUNT_NUM:null] org.springframework.web.servlet.mvc.method.annotation.ServletInvocableHandlerMethod - Method [error] returned [<500 Internal Server Error,{timestamp=Mon Sep 19 15:28:08 UTC 2016, status=500, error=Internal Server Error, exception=com.netflix.zuul.exception.ZuulException, message=Forwarding error},{}>]
I don't see much help..
Dave Syer
@dsyer
Sep 19 2016 17:56
That just means the backend was down/misbehaving
Maybe if you only had 1 backend instance you'd find the problem quicker?
pradeepkusingh
@pradeepkusingh
Sep 19 2016 18:00
we have 8 instance running in prod
any way to generate better logs ?
means if it can give info.. for which service we are getting error..
then question comes why Eureka is not removing the bad or misbehaving nodes ?
Spencer Gibb
@spencergibb
Sep 19 2016 18:02
if the nodes continue to send heartbeats eureka doesn’t know any better
you don’t know which nodes are down, how would eureka know?
pradeepkusingh
@pradeepkusingh
Sep 19 2016 18:04
but I thought client send hearbeats.. is there any option I cna apply ot make things better ?
means can we ask Eureka to do health check..
or any better healthy mechanism ?
Spencer Gibb
@spencergibb
Sep 19 2016 18:07
eureka doesn’t do external health checks. You can configure the eureka client to translate the value of /health and send that to eureka, but if /health isn’t helping you now, it won’t help eureka either.
that may not be in Angel either
pradeepkusingh
@pradeepkusingh
Sep 19 2016 18:10
ok.. I think we can implement our own HealthCheckIndicator API , would that help ?
Spencer Gibb
@spencergibb
Sep 19 2016 18:11
is it helpful now? Can you go to /health of your 8 instances and see any problems?
pradeepkusingh
@pradeepkusingh
Sep 19 2016 18:11
problem with Cloud foundry is I can't hit individual instance directly
Spencer Gibb
@spencergibb
Sep 19 2016 18:11
then it won’t help
pradeepkusingh
@pradeepkusingh
Sep 19 2016 18:12
i know...I asked CF team from pivotal we they can allow us to configure individual URls for health probe.. no luck
Dave Syer
@dsyer
Sep 19 2016 18:14
I think if you have PCF 1.7 you can
But that isn't really your main problem
pradeepkusingh
@pradeepkusingh
Sep 19 2016 18:14
unfortunately.. we are still in 1.5..
Dave Syer
@dsyer
Sep 19 2016 18:14
You have a really unstable backend it seems
So you either need to get ribbon retrying in the proxy, or stop it crashing in the backend
pradeepkusingh
@pradeepkusingh
Sep 19 2016 18:15
Ayes, we are looking for soem guide lines to debug these issues..
Dave Syer
@dsyer
Sep 19 2016 18:15
Look at the logs
There's usually a AGM of why it crashed
AGM->sign
If the platform kills a container it normally logs something
pradeepkusingh
@pradeepkusingh
Sep 19 2016 18:17
it has a very limited info... we are monitoring using appd and it doesn't say if OOM or any run out of threads
that is causing more problem..
One more qq... when I am doing /health on APIgateway it shows down
but I can see traffic is going through
not sure why
Spencer Gibb
@spencergibb
Sep 19 2016 18:20
neither are we, you’ve given no details of what /health says.
pradeepkusingh
@pradeepkusingh
Sep 19 2016 18:21
{
status: "DOWN"
}
There is nothing much..except this
pradeepkusingh
@pradeepkusingh
Sep 19 2016 18:35
hmm.. There is no security applied. reading..
pradeepkusingh
@pradeepkusingh
Sep 19 2016 18:49
@spencergibb : in 2 minutes -- X-Zuul-ServiceId:POST-ACTIVATION-PROD-A, X-Cf-Routererror:endpoint_failure, Content-Length:67, Date:Mon, 19 Sep 2016 18:35:50 GMT, Content-Type:text/plain; charset=utf-8, X-Cf-Requestid:6000a755-799b-4fa0-4293-51823c9e90e8, : Body: 502 Bad Gateway: Registered endpoint failed to handle the request.
pradeepkusingh
@pradeepkusingh
Sep 19 2016 19:38
@spencergibb , @dsyer : Thanks for helping us. Atleast we have some guidence to move forward.
David Welch
@dwelch2344
Sep 19 2016 22:23
QQ: @RefreshScope on a javax.servlet.Filter – looks like I'm creating new instances whenever the scope changes / refreshes, but the old aren't deleted / removed. We basically have a filter that may or may not do processing based on a boolean we want to change at runtime. Do we need to track state if it's been run multiple times or something?
Spencer Gibb
@spencergibb
Sep 19 2016 22:25
If all you are doing is changing a value I wouldn’t use @RefreshScope, I’d put the value in @ConfigurationProperties.
that way your filter isn’t recreated
@dwelch2344 :arrow_up:
ccit-spence
@ccit-spence
Sep 19 2016 22:39
@spencergibb Do you know much about DiscoveredResource?
David Welch
@dwelch2344
Sep 19 2016 22:41
oh snap, thanks
Ooo so configuration props will update with refresh scope?
Spencer Gibb
@spencergibb
Sep 19 2016 22:43
@dwelch2344 yes, @ccit-spence a little bit
ccit-spence
@ccit-spence
Sep 19 2016 22:44
does it only work with HAL endpoints?
I noticed the dynamic provider uses Traverson.
I need to add a link to the root of a HAL response, the link is a service vbia eureka. It is a static endpoint however not HAL
Spencer Gibb
@spencergibb
Sep 19 2016 22:46
DiscoveredResource uses Traverson new Traverson(uri, MediaTypes.HAL_JSON);
ccit-spence
@ccit-spence
Sep 19 2016 22:47
Does that mean I would need to convert the static endpoint into a HATEOAS response then I could use it.
just making sure I understand the intended purpose/use case
Spencer Gibb
@spencergibb
Sep 19 2016 22:47
you’ve now gone past 'a bit'