Revisit the design and usage of URL in Dubbo 3.0

[pinxiong]

The concept of URL-based design runs through Dubbo. There are two main threads, the provider and consumer.

• Provider

  1. registry-url: It provides all information related the registry center, which is built by RegistryConfig and the protocol begins with registry, such as :

     registry://127.0.0.1:2181/org.apache.dubbo.registry.RegistryService?application=dubbo-demo-api-provider&dubbo=2.0.2&pid=20855&registry=zookeeper&timestamp=1619492124796

     And then, registry-url will be enhanced if the proxy has been customized by users. You can see ServicesConfig#doExportUrlsFor1Protocol

     // For providers, this is used to enable custom proxy to generate invoker
     String proxy = url.getParameter(PROXY_KEY);
     if (StringUtils.isNotEmpty(proxy)) {
     registryURL = registryURL.addParameter(PROXY_KEY, proxy);
     }

     Finally, there are some changes in registry-url as follows:

     • registry-url's protocol will be updated from registry to zookeeper (if you use zookeeper as the registry center)

     • registry-url will add the parameter export and its value is provider-url(see below)

     zookeeper://127.0.0.1:2181/org.apache.dubbo.registry.RegistryService?application=dubbo-demo-api-provider&dubbo=2.0.2&export=dubbo%3A%2F%2F10.38.233.154%3A20880%2Forg.apache.dubbo.demo.DemoService%3Fanyhost%3Dtrue%26application%3Ddubbo-demo-api-provider%26bind.ip%3D10.38.233.154%26bind.port%3D20880%26default%3Dtrue%26deprecated%3Dfalse%26dubbo%3D2.0.2%26dynamic%3Dtrue%26generic%3Dfalse%26interface%3Dorg.apache.dubbo.demo.DemoService%26methods%3DsayHello%2CsayHelloAsync%26monitor%3Ddubbo%253A%252F%252Fmonitor-address%253A9090%253Fapplication%253Ddubbo-demo-api-provider%2526dubbo%253D2.0.2%2526interface%253Dorg.apache.dubbo.monitor.MonitorService%2526pid%253D20855%2526register.ip%253D10.38.233.154%2526timestamp%253D1619493040920%26pid%3D20855%26release%3D%26side%3Dprovider%26timestamp%3D1619492161107&pid=20855&timestamp=1619492124796

     • Call RegistoryProtocol#export to release the provider information in zookeeper's providers node

     dubbo://10.38.233.154:20880/org.apache.dubbo.demo.DemoService?anyhost=true&application=dubbo-demo-api-provider&default=true&deprecated=false&dubbo=2.0.2&dynamic=true&generic=false&interface=org.apache.dubbo.demo.DemoService&methods=sayHello,sayHelloAsync&pid=20855&release=&side=provider&timestamp=1619492161107

  2. monitor-url: It‘s designed for monitor and built by MonitorConfig and registry-url, which begins with dubbo, such as

     dubbo://monitor-address:9090?application=dubbo-demo-api-provider&dubbo=2.0.2&interface=org.apache.dubbo.monitor.MonitorService&pid=97749&register.ip=10.38.233.154&timestamp=1619514194521

  3. local-url: It provides local-to-local call for debugging, which begins with injvm, such as

     injvm://127.0.0.1/org.apache.dubbo.demo.DemoService?anyhost=true&application=dubbo-demo-api-provider&bind.ip=10.38.233.154&bind.port=20880&default=true&deprecated=false&dubbo=2.0.2&dynamic=true&generic=false&interface=org.apache.dubbo.demo.DemoService&methods=sayHello,sayHelloAsync&pid=97749&release=&side=provider&timestamp=1619514176258

  4. provider-url: It's the core url, which is to describe the provider's information and begins with dubbo. Also it's will be changed at runtime.

     • built by ServiceConfig first

     dubbo://10.38.233.154:20880/org.apache.dubbo.demo.DemoService?anyhost=true&application=dubbo-demo-api-provider&bind.ip=10.38.233.154&bind.port=20880&default=true&deprecated=false&dubbo=2.0.2&dynamic=true&generic=false&interface=org.apache.dubbo.demo.DemoService&methods=sayHello,sayHelloAsync&pid=20855&release=&side=provider&timestamp=1619492161107

     • enhance customized parameters by configurator

     // You can customize Configurator to append extra parameters
     if (ExtensionLoader.getExtensionLoader(ConfiguratorFactory.class)
                    .hasExtension(url.getProtocol())) {
     url = ExtensionLoader.getExtensionLoader(ConfiguratorFactory.class)
                                 .getExtension(url.getProtocol()).getConfigurator(url).configure(url);
     }

     • add dynamic parameter if it's absent

     url = url.addParameterIfAbsent(DYNAMIC_KEY, registryURL.getParameter(DYNAMIC_KEY));

     • Append parameter as monitor and value as monitor-url

     if (monitorUrl != null) {
     url = url.addParameterAndEncoded(MONITOR_KEY, monitorUrl.toFullString());
     }

     the url will be

     dubbo://10.38.233.154:20880/org.apache.dubbo.demo.DemoService?anyhost=true&application=dubbo-demo-api-provider&bind.ip=10.38.233.154&bind.port=20880&default=true&deprecated=false&dubbo=2.0.2&dynamic=true&generic=false&interface=org.apache.dubbo.demo.DemoService&methods=sayHello,sayHelloAsync&monitor=dubbo%3A%2F%2Fmonitor-address%3A9090%3Fapplication%3Ddubbo-demo-api-provider%26dubbo%3D2.0.2%26interface%3Dorg.apache.dubbo.monitor.MonitorService%26pid%3D20855%26register.ip%3D10.38.233.154%26timestamp%3D1619493040920&pid=20855&release=&side=provider&timestamp=1619492161107

  5. override-substribe-url: it used to avoid to cover subscription information when the provider subscribed

     // Subscribe the override data
     // FIXME When the provider subscribes, it will affect the scene : a certain JVM exposes the service and call
     //  the same service. Because the subscribed is cached key with the name of the service, it causes the
     //  subscription information to cover.
     final URL overrideSubscribeUrl = getSubscribedOverrideUrl(providerUrl);
     final OverrideListener overrideSubscribeListener = new OverrideListener(overrideSubscribeUrl, originInvoker);
     overrideListeners.put(overrideSubscribeUrl, overrideSubscribeListener);

     And the url begins with provider, such as

     provider://10.38.233.154:20880/org.apache.dubbo.demo.DemoService?anyhost=true&application=dubbo-demo-api-provider&bind.ip=10.38.233.154&bind.port=20880&category=configurators&check=false&default=true&deprecated=false&dubbo=2.0.2&dynamic=true&generic=false&interface=org.apache.dubbo.demo.DemoService&methods=sayHello,sayHelloAsync&monitor=dubbo%3A%2F%2Fmonitor-address%3A9090%3Fapplication%3Ddubbo-demo-api-provider%26dubbo%3D2.0.2%26interface%3Dorg.apache.dubbo.monitor.MonitorService%26pid%3D20855%26register.ip%3D10.38.233.154%26timestamp%3D1619493040920&pid=20855&release=&side=provider&timestamp=1619492161107

  6. cache-url: It saves in the disk and also will be updated if receive an notification of changes from registry center, which begins with empty, such as

     • AbstractRegistry#notify(URL, NotifyListener, List)

     protected void notify(URL url, NotifyListener listener, List<URL> urls) {
     //omit some codes
     // keep every provider's category.
      Map<String, List<URL>> result = new HashMap<>();
     for (URL u : urls) {
         if (UrlUtils.isMatch(url, u)) {
             String category = u.getParameter(CATEGORY_KEY, DEFAULT_CATEGORY);
             List<URL> categoryList = result.computeIfAbsent(category, k -> new ArrayList<>());
             categoryList.add(u);
         }
     }
     //omit some codes
     for (Map.Entry<String, List<URL>> entry : result.entrySet()) {
         String category = entry.getKey();
         List<URL> categoryList = entry.getValue();
         categoryNotified.put(category, categoryList);
         listener.notify(categoryList);
         // We will update our cache file after each notification.
         // When our Registry has a subscribe failure due to network jitter, we can return at least the existing cache URL.
         saveProperties(url);
     }
     }

     • The cache-url was in the dubbo-registry-dubbo-demo-api-provider-127.0.0.1-2181.cache file

     org.apache.dubbo.demo.DemoService=empty\://10.38.233.154\:20880/org.apache.dubbo.demo.DemoService?anyhost\=true&application\=dubbo-demo-api-provider&bind.ip\=10.38.233.154&bind.port\=20880&category\=configurators&check\=false&default\=true&deprecated\=false&dubbo\=2.0.2&dynamic\=true&generic\=false&interface\=org.apache.dubbo.demo.DemoService&methods\=sayHello,sayHelloAsync&monitor\=dubbo%3A%2F%2Fmonitor-address%3A9090%3Fapplication%3Ddubbo-demo-api-provider%26dubbo%3D2.0.2%26interface%3Dorg.apache.dubbo.monitor.MonitorService%26pid%3D70560%26register.ip%3D10.38.233.154%26timestamp%3D1619506379919&pid\=70560&release\=&side\=provider&timestamp\=1619506347010

• Consumer

  1. local-url: It's for checking if it's in JVM, which begins with injvm

     injvm://127.0.0.1/org.apache.dubbo.rpc.service.GenericService?application=dubbo-demo-api-consumer&dubbo=2.0.2&generic=true&interface=org.apache.dubbo.demo.DemoService&pid=20527&register.ip=10.38.233.154&release=&side=consumer&sticky=false&timestamp=1619520920163

  2. registry-url: It provides registry center information, which is built by ReferenceConfig and begins with registry

     registry://127.0.0.1:2181/org.apache.dubbo.registry.RegistryService?application=dubbo-demo-api-consumer&dubbo=2.0.2&pid=20527&registry=zookeeper&timestamp=1619521528200

     Also, it will be changed at runtime as follows:

     • Solve compatible problem, see ConfigValidationUtils#genCompatibleRegistries(List, boolean)

     private static List<URL> genCompatibleRegistries(List<URL> registryList, boolean provider) {
     List<URL> result = new ArrayList<>(registryList.size());
     registryList.forEach(registryURL -> {
         result.add(registryURL);
             if (provider) {
                 // for registries enabled service discovery, automatically register interface compatible addresses.
                 if (SERVICE_REGISTRY_PROTOCOL.equals(registryURL.getProtocol())
                            && registryURL.getParameter(REGISTRY_PUBLISH_INTERFACE_KEY, ConfigurationUtils.getDynamicGlobalConfiguration().getBoolean(DUBBO_PUBLISH_INTERFACE_DEFAULT_KEY, false))
                            && registryNotExists(registryURL, registryList, REGISTRY_PROTOCOL)) {
                     URL interfaceCompatibleRegistryURL = URLBuilder.from(registryURL)
                                .setProtocol(REGISTRY_PROTOCOL)
                                .removeParameter(REGISTRY_TYPE_KEY)
                                .build();
                     result.add(interfaceCompatibleRegistryURL);
                 }
             }
         });
     return result;
     }

     • Add monitor-url if it exists (you can see above to how to generate monitor-url), also add parameter as refer and value as all configuration of ReferenceConfig

     // if protocols not injvm checkRegistry
     if (!LOCAL_PROTOCOL.equalsIgnoreCase(getProtocol())) {
     checkRegistry();
     List<URL> us = ConfigValidationUtils.loadRegistries(this, false);
     if (CollectionUtils.isNotEmpty(us)) {
         for (URL u : us) {
             URL monitorUrl = ConfigValidationUtils.loadMonitor(this, u);
             if (monitorUrl != null) {
                 map.put(MONITOR_KEY, URL.encode(monitorUrl.toFullString()));
             }
             urls.add(u.addParameterAndEncoded(REFER_KEY, StringUtils.toQueryString(map)));
         }
     }
     // omit some codes
     }

     such as

     registry://127.0.0.1:2181/org.apache.dubbo.registry.RegistryService?application=dubbo-demo-api-consumer&dubbo=2.0.2&pid=20527&refer=application%3Ddubbo-demo-api-consumer%26dubbo%3D2.0.2%26generic%3Dtrue%26interface%3Dorg.apache.dubbo.demo.DemoService%26pid%3D20527%26register.ip%3D10.38.233.154%26side%3Dconsumer%26sticky%3Dfalse%26timestamp%3D1619520920163&registry=zookeeper&timestamp=1619521528200

     • the protocol changes from registry to zookeeper(if you use zookeeper as the registry center)

     zookeeper://127.0.0.1:2181/org.apache.dubbo.registry.RegistryService?application=dubbo-demo-api-consumer&dubbo=2.0.2&pid=20527&refer=application%3Ddubbo-demo-api-consumer%26dubbo%3D2.0.2%26generic%3Dtrue%26interface%3Dorg.apache.dubbo.demo.DemoService%26pid%3D20527%26register.ip%3D10.38.233.154%26side%3Dconsumer%26sticky%3Dfalse%26timestamp%3D1619520920163&timestamp=1619521528200

  3. customer-url: it's generated from registry-url, which begins with consumer. There are two kinds of it as follows

     generic is set true

     consumer://10.38.233.154/org.apache.dubbo.rpc.service.GenericService?application=dubbo-demo-api-consumer&dubbo=2.0.2&generic=true&interface=org.apache.dubbo.demo.DemoService&pid=20527&side=consumer&sticky=false&timestamp=1619520920163

     or false

     consumer://10.38.233.154/org.apache.dubbo.demo.DemoService?application=dubbo-demo-api-consumer&category=providers,configurators,routers&dubbo=2.0.2&generic=false&interface=org.apache.dubbo.demo.DemoService&methods=sayHello,sayHelloAsync&pid=63583&side=consumer&sticky=false&timestamp=1619520920163

     finally, its protocol will be changed, such as

     dubbo://10.38.233.154/org.apache.dubbo.demo.DemoService?application=dubbo-demo-api-consumer&dubbo=2.0.2&generic=true&interface=org.apache.dubbo.demo.DemoService&pid=20527&register.ip=10.38.233.154&side=consumer&sticky=false&timestamp=1619520920163

  4. directory-url: it's generated from registry-url, which begins with dubbo.

     private URL turnRegistryUrlToConsumerUrl(URL url) {
          return URLBuilder.from(url)
                  .setHost(queryMap.get(REGISTER_IP_KEY) == null ? url.getHost() : queryMap.get(REGISTER_IP_KEY))
                  .setPort(0)
                  .setProtocol(queryMap.get(PROTOCOL_KEY) == null ? DUBBO : queryMap.get(PROTOCOL_KEY))
                  .setPath(queryMap.get(INTERFACE_KEY))
                  .clearParameters()
                  .addParameters(queryMap)
                  .removeParameter(MONITOR_KEY)
                  .build();
     }

     such as

     dubbo://10.38.233.154/org.apache.dubbo.demo.DemoService?application=dubbo-demo-api-consumer&dubbo=2.0.2&generic=true&interface=org.apache.dubbo.demo.DemoService&pid=20527&register.ip=10.38.233.154&side=consumer&sticky=false&timestamp=1619520920163

  5. override-directory-url: it's generated from registry-url, which begins with dubbo. The override-directory-url is the same as directory-url first.

     dubbo://10.38.233.154/org.apache.dubbo.demo.DemoService?application=dubbo-demo-api-consumer&dubbo=2.0.2&generic=true&interface=org.apache.dubbo.demo.DemoService&pid=20527&register.ip=10.38.233.154&side=consumer&sticky=false&timestamp=1619520920163

  6. url-to-registry: It generated by all parameters in registry-url

     URL urlToRegistry = new URL(CONSUMER_PROTOCOL, parameters.remove(REGISTER_IP_KEY), 0, type.getName(), parameters);

     such as

     consumer://10.38.233.154/org.apache.dubbo.rpc.service.GenericService?application=dubbo-demo-api-consumer&dubbo=2.0.2&generic=true&interface=org.apache.dubbo.demo.DemoService&pid=20527&side=consumer&sticky=false&timestamp=1619520920163

  7. subscribed-url: It generated by url-to-registry and add parameter category

     public static URL toSubscribeUrl(URL url) {
     return url.addParameter(CATEGORY_KEY, PROVIDERS_CATEGORY + "," + CONFIGURATORS_CATEGORY + "," + ROUTERS_CATEGORY);
     }

     such as

     consumer://10.38.233.154/org.apache.dubbo.rpc.service.GenericService?application=dubbo-demo-api-consumer&category=providers,configurators,routers&dubbo=2.0.2&generic=true&interface=org.apache.dubbo.demo.DemoService&pid=20527&side=consumer&sticky=false&timestamp=1619520920163

[pinxiong]

Problem

• The responsibility for different URLs are unclear
  • The same URL means different things in different places
  • The names of URLs are irregularity
  • It's very difficult to find the differences between two similar URLs
• There are some invalid parameters in URL
  • We can find the same parameter timestamp in URL (such as provider-url)
  • Are all the parameters necessary in URLs ?
• High maintenance costs, easy to cause problems

Suggestion

• Define the boundaries or types of different URLs by protocols
• Define the necessary parameters of URLs for each of the different protocols
• Define the state changes relationship between different URLs with the same protocol
• Simplify the logic, use the way of pipeline or plugin to make every changes (URLs with the same protocol) clear
• Don't expose addParameter method to add parameter-value pair into the URL
• The sooner you refactor, the better