[fix] fix the workflow result transfer fail bug

oh-my-scheduler-server/src/main/java/com/github/kfcfans/oms/server/service/workflow/WorkflowInstanceManager.java

@@ -125,7 +125,7 @@ public class WorkflowInstanceManager {
             log.info("[Workflow-{}] start workflow successfully, wfInstanceId={}", wfInfo.getId(), wfInstanceId);
 
             // 真正开始执行根任务
-            runInstance(root.getJobId(), instanceId, wfInstanceId);
+            runInstance(root.getJobId(), instanceId, wfInstanceId, null);
         }catch (Exception e) {
 
             wfInstanceInfo.setStatus(WorkflowInstanceStatus.FAILED.getV());
@@ -193,6 +193,8 @@ public class WorkflowInstanceManager {
 
             // 重新计算需要派发的任务
             Map<Long, Long> jobId2InstanceId = Maps.newHashMap();
+            Map<Long, String> jobId2InstanceParams = Maps.newHashMap();
+
             AtomicBoolean allFinished = new AtomicBoolean(true);
             relyMap.keySet().forEach(jobId -> {
 
@@ -210,14 +212,15 @@ public class WorkflowInstanceManager {
                 }
 
                 // 所有依赖已经执行完毕，可以执行该任务
-                Map<Long, String> jobId2Result = Maps.newHashMap();
+                Map<Long, String> preJobId2Result = Maps.newHashMap();
-                // 构建下一个任务的入参 （jobId -> result）
+                // 构建下一个任务的入参 （前置任务 jobId -> result）
-                relyMap.get(jobId).forEach(jid -> jobId2Result.put(jid, jobId2Node.get(jid).getResult()));
+                relyMap.get(jobId).forEach(jid -> preJobId2Result.put(jid, jobId2Node.get(jid).getResult()));
 
-                Long newInstanceId = instanceService.create(jobId, wfInstance.getAppId(), JsonUtils.toJSONString(jobId2Result), wfInstanceId, System.currentTimeMillis());
+                Long newInstanceId = instanceService.create(jobId, wfInstance.getAppId(), JsonUtils.toJSONString(preJobId2Result), wfInstanceId, System.currentTimeMillis());
                 jobId2Node.get(jobId).setInstanceId(newInstanceId);
 
                 jobId2InstanceId.put(jobId, newInstanceId);
+                jobId2InstanceParams.put(jobId, JsonUtils.toJSONString(preJobId2Result));
 
                 log.debug("[Workflow-{}] workflowInstance(wfInstanceId={}) start to process new node(jobId={},instanceId={})", wfId, wfInstanceId, jobId, newInstanceId);
             });
@@ -233,7 +236,7 @@ public class WorkflowInstanceManager {
             workflowInstanceInfoRepository.saveAndFlush(wfInstance);
 
             // 持久化结束后，开始调度执行所有的任务
-            jobId2InstanceId.forEach((jobId, newInstanceId) -> runInstance(jobId, newInstanceId, wfInstanceId));
+            jobId2InstanceId.forEach((jobId, newInstanceId) -> runInstance(jobId, newInstanceId, wfInstanceId, jobId2InstanceParams.get(jobId)));
 
         }catch (Exception e) {
             wfInstance.setStatus(WorkflowInstanceStatus.FAILED.getV());
@@ -244,11 +247,19 @@ public class WorkflowInstanceManager {
         }
     }
 
-    private void runInstance(Long jobId, Long instanceId, Long wfInstanceId) {
+    /**
+     * 允许任务实例
+     * 需要将创建和运行任务实例分离，否则在秒失败情况下，会发生DAG覆盖更新的问题
+     * @param jobId 任务ID
+     * @param instanceId 任务实例ID
+     * @param wfInstanceId 工作流任务实例ID
+     * @param instanceParams 任务实例参数，值为上游任务的执行结果： preJobId to preJobInstanceResult
+     */
+    private void runInstance(Long jobId, Long instanceId, Long wfInstanceId, String instanceParams) {
         JobInfoDO jobInfo = jobInfoRepository.findById(jobId).orElseThrow(() -> new IllegalArgumentException("can't find job by id:" + jobId));
         // 洗去时间表达式类型
         jobInfo.setTimeExpressionType(TimeExpressionType.API.getV());
-        dispatchService.dispatch(jobInfo, instanceId, 0, null, wfInstanceId);
+        dispatchService.dispatch(jobInfo, instanceId, 0, instanceParams, wfInstanceId);
     }
 
 }

oh-my-scheduler-server/src/main/resources/application.properties

@@ -6,7 +6,7 @@ spring.jpa.open-in-view=false
 spring.data.mongodb.repositories.type=none
 
 # 文件上传配置
-spring.servlet.multipart.enabled =true
+spring.servlet.multipart.enabled=true
 spring.servlet.multipart.file-size-threshold=0
 spring.servlet.multipart.max-file-size=209715200
 spring.servlet.multipart.max-request-size=209715200

oh-my-scheduler-server/src/main/resources/logback-dev.xml

@@ -18,13 +18,12 @@
             <pattern>${CONSOLE_LOG_PATTERN}</pattern>
             <charset>utf8</charset>
         </encoder>
-        <filter class="ch.qos.logback.classic.filter.LevelFilter">
-            <level>debug</level>
-            <onMatch>DENY</onMatch>
-            <onMismatch>NEUTRAL</onMismatch>
-        </filter>
     </appender>
 
+    <logger name="com.github.kfcfans.oms" level="DEBUG" additivity="false">
+        <appender-ref ref="CONSOLE" />
+    </logger>
+
     <root level="INFO">
         <appender-ref ref="CONSOLE"/>
     </root>

oh-my-scheduler-worker-samples/src/main/java/com/github/kfcfans/oms/samples/processors/StandaloneProcessorDemo.java

@@ -36,6 +36,7 @@ public class StandaloneProcessorDemo implements BasicProcessor {
         }
 
         System.out.println("================ StandaloneProcessorDemo#process ================");
+        System.out.println(context.getJobParams());
         // 根据控制台参数判断是否成功
         boolean success = !"failed".equals(context.getJobParams());
         omsLogger.info("StandaloneProcessorDemo finished process,success: .", success);