I am carrying out a PoC on Ocelot Gateway to use for our microservices architecture. I have successfully used the Gateway to perform all the routing tasks uptil now.
It provides a QoS Service that helds configuration in ocelot.json for circuit breaker and breaks the circuit for a duration.
The configurations set in my ocelot.json file are as follows for a specific route:
{
"DownstreamPathTemplate": "/api/classes/exception/{id}",
"DownstreamScheme": "http",
"UpstreamPathTemplate": "/classes/exception/{id}",
"ServiceName": "class-service",
"UpstreamHttpMethod": [ "Get" ],
"LoadBalancerOptions": {
"Type": "LeastConnection"
},
"QoSOptions": {
"ExceptionsAllowedBeforeBreaking": 2,
"DurationOfBreak": 10000,
"TimeoutValue": 3000
}
}
Now, this works fine for the timeout issue, it breaks the circuit for 10s when the route does not respond in 3s.
But for the exception case, it does not break the circuit when there is an exception arising for 2 times when the route is being called.
Are there any more configurations required?
I have added Nuget Package for Ocelot and Ocelot.Provider.Polly and also added the following in my Startup.cs file:
s.AddOcelot().AddPolly();
Please help me through this.
I have followed the [Ocelot Documentation] https://ocelot.readthedocs.io/en/latest/features/qualityofservice.html to implement this
Related
In an NX monorepo I'm building 3 nestjs application, an auth-service + user-service and a gateway to start off with. They're all powered by apollo graphql and following the official nestjs documentation.
The issue that I'm having is that with both the user-service and auth-service up and processing requests successfully as individual servers, at the same time, the gateway throws
Couldn't load service definitions for "auth" at http://localhost:3100/apis/auth-service/graphql: 400: Bad Request
The services themselves are standard graphql applications, nothing basic.
The definitions for the gateway are as such:
{
server: {
debug: true,
playground: true,
autoSchemaFile: './apps/gateway/schema.gql',
sortSchema: true,
introspection: true,
cors: ['*'],
path: '/apis/gateway/graphql';
},
gateway: {
supergraphSdl: new IntrospectAndCompose({
subgraphHealthCheck: true,
subgraphs: [
{
name: 'user',
url: resolveSubgraphUrl('user'),
},
{
name: 'auth',
url: resolveSubgraphUrl('auth'),
},
],
}),
}
}
I have created an nx based monorepository that allows you to reproduce this issue easily, by spinning up all 3 servers in watch mode at the same time, on localhost, on different ports.
Everything is mapped as it should. Link: https://github.com/sebastiangug/nestjs-federation.git
The README contains the two commands necessary to run it as well as the same set of instructions plus the health-checks queries available from each service.
Versions used:
"#apollo/subgraph": "2.1.3",
"#apollo/federation": "0.37.1",
"#apollo/gateway": "2.1.3",
"apollo-server-express": "3.6.7",
"graphql": "16.5.0",
"#nestjs/graphql": "10.1.3",
"#nestjs/platform-express": "9.0.8",
Any ideas what I'm doing wrong here or what further configuration is necessary to achieve this?
Thank you
When using Quarkus microprofile as a rest client, how can I configure underlying HttpClient?
Like number of retries, connection pool size per host and so on?
Also is it possible to force client restart somehow (so connections pool will be restarted)?
https://download.eclipse.org/microprofile/microprofile-rest-client-2.0-RC2/microprofile-rest-client-2.0-RC2.html#_configuration_keys outlines the full set of configuration keys that can be used.
The ones you're looking for are:
{packageName}.{interfaceName}/mp-rest/connectTimeout
{packageName}.{interfaceName}/mp-rest/readTimeout
The RestClientBuilder also has methods for setting those properties if you're using the programmatic API instead of the CDI approach.
I'm not aware of any means of restarting the underlying HTTP client connection pool. What would be the use case for such a situation that doesn't require the whole application to be restarted?
So... After a lot of digging, here is a solution I've found so far. It is not obvious apparently:
To make it work in pure Java (no native)
Under resources/META-INF/services directory add file named org.eclipse.microprofile.rest.client.spi.RestClientBuilderListener containing class name of your implementation of RestClientBuilderListener interface. For example my.test.MyBuilderListener. This will allow ServiceLocator to execute your listener
Refer a property you want to modify from ResteasyClientBuilder, for example to set your custom value to connectionTTL code will looks like this:
public void onNewBuilder(RestClientBuilder builder) {
log.info("Changing TTL for connections");
builder.property("resteasy.connectionTTL", List.of(2L, TimeUnit.SECONDS));
}
Ie. add a resteasy. prefix to a property name
Profit
Now native support:
After steps above:
Set both MyBuildListener and ResteasyClientBuilder available for reflection by creating a file reflection-config.json:
[
{
"name": "org.jboss.resteasy.client.jaxrs.ResteasyClientBuilder",
"allDeclaredConstructors": true,
"allPublicConstructors": true,
"allDeclaredMethods": true,
"allPublicMethods": true,
"allDeclaredFields": true,
"allPublicFields": true
}, {
"name": "my.test.MyBuilderListener",
"allDeclaredConstructors": true,
"allPublicConstructors": true,
"allDeclaredMethods": true,
"allPublicMethods": true,
"allDeclaredFields": true,
"allPublicFields": true
}
]
Add service registration file to resources. Create a file named resources-config.json with content
{
"resources": [
{
"pattern": "META-INF/services/org\\.eclipse\\.microprofile\\.rest\\.client\\.spi\\.RestClientBuilderListener$"
}
]
}
register both files in application.yaml:
quarkus:
native:
additional-build-args: -H:ResourceConfigurationFiles=resources-config.json, -H:ReflectionConfigurationFiles=reflection-config.json
Native profit
Have fun
I am setting up Segment routing via Pathman-SR with ODL Nitrogen Controller and vMX Juniper routers. To allow this, I have to change IANA subojbects code points, but I am unable to do it...
Followed this documenntations, but still no result:
https://docs.opendaylight.org/en/stable-carbon/user-guide/pcep-user-guide.html#segment-routing
https://test-odl-docs.readthedocs.io/en/latest/user-guide/pcep-user-guide.html
I tried to update configuration via REST API, but when I send PUT request:
/restconf/config/pcep-segment-routing-app-config:pcep-segment-routing-app-config
with the body:
<pcep-segment-routing-config xmlns="urn:opendaylight:params:xml:ns:yang:controller:pcep:segment-routing-app-config">
<iana-sr-subobjects-type>true</iana-sr-subobjects-type>
</pcep-segment-routing-config>
I get the following error:
{
"errors": {
"error": [
{
"error-type": "protocol",
"error-tag": "invalid-value",
"error-message": "URI has bad format. Possible reasons:\n 1. \"pcep-segment-routing-app-config:pcep-segment-routing-app-config\" was not found in parent data node.\n 2. \"pcep-segment-routing-app-config:pcep-segment-routing-app-config\" is behind mount point. Then it should be in format \"/yang-ext:mount/pcep-segment-routing-app-config:pcep-segment-routing-app-config\"."
}
]
}
}
I think there is a typo in the URL in the doc, you have to use /restconf/config/pcep-segment-routing-app-config:pcep-segment-routing-config
You can check this guide for reference:
https://docs.opendaylight.org/projects/bgpcep/en/stable-neon/pcep/pcep-user-guide-active-stateful-pce.html#iana-code-points
I'm using Spring Boot health indicator from an actuator. So far example response looks like:
{
"status":"DOWN",
"details": {
"diskSpace": {
"status":"UP",
"details": {
"total":499963170816,
"free":250067189760,
"threshold":10485760
}
}
}
}
Because I need to make /actuator/health endpoint public, I need to hide details for health indicators, so I expect to get something like this:
{
"status":"DOWN",
"details": {
"diskSpace": {
"status":"UP"
}
}
}
For disk space it's not a big problem but e.g. for database I don't want to share exception message and details in case of it's outage. Also (as I mentioned at the beginning) it must be public so I don't want to make this endpoint 'when-authorized'. And at the end - it would be great if it's possible to do that without writing my own custom endpoint.
Is it possible at all?
This isn't possible at the time of writing (in Spring Boot 2.1 and earlier) without writing your own custom endpoint. I've opened an issue to consider it as an enhancement for a future version of Spring Boot.
There is a way to achieve this in Spring Boot 2.X
management:
health:
db:
enabled: false
diskspace:
enabled: false
mongo:
enabled: false
refresh:
enabled: false
More information can be found here https://docs.spring.io/spring-boot/docs/current/reference/html/production-ready-endpoints.html#_auto_configured_healthindicators
I have a django celery set up, and I'm switching over from rabbitmq to use SQS as the broker. I'm finding that my tasks decorated with a routing_key value are not producing messages in the broker?
Set up is:
CELERY_QUEUES = {
"default": {
"exchange": "default",
"binding_key": "default"},
"sentry": {
"exchange": "default",
"binding_key": "sentry"},
}
CELERY_DEFAULT_QUEUE = "default"
CELERY_DEFAULT_EXCHANGE = "default"
CELERY_DEFAULT_EXCHANGE_TYPE = "direct"
CELERY_DEFAULT_ROUTING_KEY = "default"
task is defined as
#task(routing_key = 'sentry', ignore_result = True)
def doSomething():
print "Hello"
doSomething.delay() # No message is produced
Everything routes fine into the default queue. Unusually everything works fine with rabbit mq?
The Amazon SQS console is showing the queue 'sentry', but no messages are sent to it (I'm not sure what created the queue)
Bonus: unusually, when I first tried this out (about 4 hours ago), some messages did appear to make it to the sentry queue? What could have possibly caused this?
Thanks
I tried your example with one change. Instead of defining the routing key in the task decorator, I defined the queue in the call to the task. And I changed the call from "delay" to "apply_async"
So my code looks like
#task
def doSomething():
print "Hello"
doSomething.apply_async(queue='sentry')