how to write block handler ?

[begemot63]

1. When blockHandler() method will be called?
2. I want to write block handler . How I can do it?
   The handler needs to return null object to method caller if exception was caught.

[sczyh30]

Hi, if you're referring block handler and fallback in @SentinelResource annotation, you may refer to https://github.com/alibaba/Sentinel/tree/master/sentinel-extension/sentinel-annotation-aspectj

[begemot63]

Hi Eric I read that document. However my code annotation did not work. I used this code: public class MemcachedCache extends AbstractValueAdaptingCache { @OverRide @SentinelResource(value = MEMCACHED_RESILIENCE_RESOURCE, fallback = "fallbackForLookup") public Object lookup(Object key) { ... // memcached calls } Note, the class which shown here MemcachedCache extends AbstractValueAdaptingCache which implements Cache interface. The class MemcachedCache used as implementation of Cacha interface to support Cachable spring boot annotation. So the loookup(), get(), put() method invoked by spring boot framework. It looks like your proxy pattern with @SentinelResource does not work. So I decided to use the Sentinel SphU API, (Not sure what SphU stands for , but) I have the following code now: public Object lookup(Object key) { ... Entry entry = null; try { entry = SphU.entry(MEMCACHED_RESILIENCE_RESOURCE, EntryType.IN, 1, key); ..... // access to memcached // to simulate error: throw new RuntimeException("error"); } catch (BlockException ex) { LOGGER.info("lookup, Cuitcut Breaker is open", name); handleMetrics("lookup", name, key.toString(), memCachedKey, ex); return null; } catch (Throwable th) { Tracer.trace(th); LOGGER.info("lookup, Remote cache '{}' is not available", name); ClusterNode createRoleNode = ClusterBuilderSlot.getClusterNode("memcached"); LOGGER.info("lookup, Exception Total Count: {}", createRoleNode.totalException()); LOGGER.info("lookup, Block Request Count: {}", createRoleNode.blockRequest()); LOGGER.info("lookup, DegradeRuleManager.getRules(): {} ",DegradeRuleManager.getRules()); handleMetrics("lookup", name, key.toString(), memCachedKey, th); return null; } finally { if(entry!=null) { entry.exit(); } } Log indicates I have defined 2 DegradeRule DegradeRule{resource=memcached, grade=1, count=0.6, limitApp=default, timeWindow=20, minRequestAmount=5, slowRatioThreshold=1.0, statIntervalMs=1000}, DegradeRule{resource=memcached, grade=2, count=10.0, limitApp=default, timeWindow=20, minRequestAmount=5, slowRatioThreshold=1.0, statIntervalMs=10}] Also log shows Exception Total Count: 11 however SphU.entry never throws BlockException exception. It means socket breaker was never opened . Did I missed something ? Regards Vlad

…

On Wed, Apr 21, 2021 at 6:38 PM Eric Zhao ***@***.***> wrote: Hi, if you're referring block handler and fallback in @SentinelResource annotation, you may refer to https://github.com/alibaba/Sentinel/tree/master/sentinel-extension/sentinel-annotation-aspectj — You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub <#2148 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AEW3G6I2W37O35RFQSPMVHDTJ54YNANCNFSM43JWRTMQ> .

[SomebodyHuang2017]

There are several steps in the troubleshooting method：

1. Make sure that the Spring AOP aspect can cover this class
2. If the lookup method is called in abstract class, then you should ensure that it calls the lookup method of the proxy class to take effect.

such as:
`@Configuration
@EnableAspectJAutoProxy(proxyTargetClass = true, exposeProxy = true)
public class AopConfiguration {

@Bean
public OceanusAspect oceanusAspect() {
    return new OceanusAspect();
}

}`

[begemot63]

Hi Eric The second approach also does not work The rules are not charged The exception counter is increased but BlockException was never thrown How to properly apply DegradeRule ?

[sczyh30]

It looks like your proxy pattern with @SentinelResource does not work.

Please check whether your class is a Spring bean (Spring AOP proxy requires the instance managed by Spring).

The exception counter is increased but BlockException was never thrown

Take rule1 as an example: DegradeRule{resource=memcached, grade=1, count=0.6, limitApp=default, timeWindow=20, minRequestAmount=5, slowRatioThreshold=1.0, statIntervalMs=1000}

This indicates: If your QPS > 5, error ratio > 60% in this second, the circuit breaker will be open. Please check whether your requests satisfy the condition.

[begemot63]

Hi Eric,
As you may see my code throw news RuntimeException("error"); every time we entry in the try block.
I remove DegradeRule{resource=memcached, grade=1, count=0.6, limitApp=default, timeWindow=20, minRequestAmount=5, slowRatioThreshold=1.0, statIntervalMs=1000}
and kept only second rule: DegradeRule{resource=memcached, grade=2, count=4.0, limitApp=default, timeWindow=10, minRequestAmount=5, slowRatioThreshold=1.0, statIntervalMs=1000}]
The method is called hundred times in the loop but BlockException is never thrown.
If I add FlowRule : FlowRule{resource=memcached, limitApp=default, grade=1, count=1.0, strategy=0, refResource=null, controlBehavior=0, warmUpPeriodSec=10, maxQueueingTimeMs=500, clusterMode=false, clusterConfig=null, controller=com.alibaba.csp.sentinel.slots.block.flow.controller.DefaultController@4a110bc1}
then the BlockExceptopn is thrown each time when method is called. The flow rule works only if count set to 1 , bigger value does not throw BlockException

[begemot63]

I solved the problem by increasing statIntervalMs