gtsoft/gtsoft-snail-job-server
1
0
Fork 0
Code Issues Pull Requests 1 Actions Packages Projects 1 Releases Wiki Activity

fix(sj_1.1.0-beta2): 修复工作流节点未执行问题

Browse Source
This commit is contained in:
opensnail 2024-06-30 00:07:51 +08:00
parent dc320b1332
commit 9c34d2c660
7 changed files with 143 additions and 60 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

2
snail-job-server/snail-job-server-job-task/src/main/java/com/aizuda/snailjob/server/job/task/support/dispatch/JobExecutorResultActor.java
View File

@ -46,7 +46,7 @@ public class JobExecutorResultActor extends AbstractActor {
@Override
public Receive createReceive() {
return receiveBuilder().match(JobExecutorResultDTO.class, result -> {
SnailJobLog.LOCAL.info("更新任务状态. 参数:[{}]", JsonUtil.toJsonString(result));
SnailJobLog.LOCAL.debug("更新任务状态. 参数:[{}]", JsonUtil.toJsonString(result));
try {
Assert.notNull(result.getTaskId(), ()-> new SnailJobServerException("taskId can not be null"));
Assert.notNull(result.getJobId(), ()-> new SnailJobServerException("jobId can not be null"));

122
snail-job-server/snail-job-server-job-task/src/main/java/com/aizuda/snailjob/server/job/task/support/dispatch/WorkflowExecutorActor.java
View File

@ -64,7 +64,7 @@ public class WorkflowExecutorActor extends AbstractActor {
@Override
public Receive createReceive() {
return receiveBuilder().match(WorkflowNodeTaskExecuteDTO.class, taskExecute -> {
log.debug("工作流开始执行. [{}]", JsonUtil.toJsonString(taskExecute));
log.info("工作流开始执行. [{}]", JsonUtil.toJsonString(taskExecute));
try {
doExecutor(taskExecute);
@ -73,7 +73,8 @@ public class WorkflowExecutorActor extends AbstractActor {
SnailJobLog.LOCAL.error("workflow executor exception. [{}]", taskExecute, e);
handlerTaskBatch(taskExecute, JobTaskBatchStatusEnum.FAIL.getStatus(),
JobOperationReasonEnum.TASK_EXECUTION_ERROR.getReason());
SpringContext.getContext().publishEvent(new WorkflowTaskFailAlarmEvent(taskExecute.getWorkflowTaskBatchId()));
SpringContext.getContext()
.publishEvent(new WorkflowTaskFailAlarmEvent(taskExecute.getWorkflowTaskBatchId()));
} finally {
getContext().stop(getSelf());
}
@ -84,11 +85,14 @@ public class WorkflowExecutorActor extends AbstractActor {
WorkflowTaskBatch workflowTaskBatch = workflowTaskBatchMapper.selectById(taskExecute.getWorkflowTaskBatchId());
Assert.notNull(workflowTaskBatch, () -> new SnailJobServerException("任务不存在"));
if (SystemConstants.ROOT.equals(taskExecute.getParentId()) && JobTaskBatchStatusEnum.WAITING.getStatus() == workflowTaskBatch.getTaskBatchStatus()) {
handlerTaskBatch(taskExecute, JobTaskBatchStatusEnum.RUNNING.getStatus(), JobOperationReasonEnum.NONE.getReason());
if (SystemConstants.ROOT.equals(taskExecute.getParentId())
&& JobTaskBatchStatusEnum.WAITING.getStatus() == workflowTaskBatch.getTaskBatchStatus()) {
handlerTaskBatch(taskExecute, JobTaskBatchStatusEnum.RUNNING.getStatus(),
JobOperationReasonEnum.NONE.getReason());
Workflow workflow = workflowMapper.selectById(workflowTaskBatch.getWorkflowId());
JobTimerWheel.clearCache(MessageFormat.format(WorkflowTimerTask.IDEMPOTENT_KEY_PREFIX, taskExecute.getWorkflowTaskBatchId()));
JobTimerWheel.clearCache(
MessageFormat.format(WorkflowTimerTask.IDEMPOTENT_KEY_PREFIX, taskExecute.getWorkflowTaskBatchId()));
JobTimerWheel.registerWithWorkflow(() -> new WorkflowTimeoutCheckTask(taskExecute.getWorkflowTaskBatchId()),
Duration.ofSeconds(workflow.getExecutorTimeout()));
@ -103,8 +107,17 @@ public class WorkflowExecutorActor extends AbstractActor {
// 查到当前节点【ParentId】的所有兄弟节点是否有后继节点，若有则不能直接完成任务
Set<Long> allSuccessors = Sets.newHashSet();
for (Long nodeId : setView.immutableCopy()) {
allSuccessors.addAll(graph.successors(nodeId));
Set<Long> successors = graph.successors(nodeId);
if (CollUtil.isNotEmpty(successors)) {
for (final Long successor : successors) {
// 寻找当前的节点的所有前序节点
allSuccessors.addAll(graph.predecessors(successor));
}
allSuccessors.addAll(successors);
}
}
log.warn("父节点:[{}] 所有的节点:[{}]", taskExecute.getParentId(), allSuccessors);
// 若所有的兄弟节点的子节点都没有后继节点可以完成次任务
if (CollUtil.isEmpty(allSuccessors)) {
@ -118,26 +131,27 @@ public class WorkflowExecutorActor extends AbstractActor {
JobTaskBatch::getTaskBatchStatus, JobTaskBatch::getOperationReason, JobTaskBatch::getId)
.eq(JobTaskBatch::getWorkflowTaskBatchId, workflowTaskBatch.getId())
.in(JobTaskBatch::getWorkflowNodeId,
Sets.union(brotherNode, Sets.newHashSet(taskExecute.getParentId())))
Sets.union(allSuccessors, Sets.newHashSet(taskExecute.getParentId())))
);
List<WorkflowNode> workflowNodes = workflowNodeMapper.selectList(new LambdaQueryWrapper<WorkflowNode>()
.in(WorkflowNode::getId, Sets.union(allSuccessors, Sets.newHashSet(taskExecute.getParentId())))
.orderByAsc(WorkflowNode::getPriorityLevel));
Map<Long, List<JobTaskBatch>> jobTaskBatchMap = StreamUtils.groupByKey(allJobTaskBatchList, JobTaskBatch::getWorkflowNodeId);
Map<Long, List<JobTaskBatch>> jobTaskBatchMap = StreamUtils.groupByKey(allJobTaskBatchList,
JobTaskBatch::getWorkflowNodeId);
Map<Long, WorkflowNode> workflowNodeMap = StreamUtils.toIdentityMap(workflowNodes, WorkflowNode::getId);
List<JobTaskBatch> parentJobTaskBatchList = jobTaskBatchMap.get(taskExecute.getParentId());
// 如果父节点是无需处理则不再继续执行
if (CollUtil.isNotEmpty(parentJobTaskBatchList) &&
parentJobTaskBatchList.stream()
.map(JobTaskBatch::getOperationReason)
.filter(Objects::nonNull)
.anyMatch(JobOperationReasonEnum.WORKFLOW_SUCCESSOR_SKIP_EXECUTION::contains)) {
workflowBatchHandler.complete(taskExecute.getWorkflowTaskBatchId(), workflowTaskBatch);
return;
}
// if (CollUtil.isNotEmpty(parentJobTaskBatchList) &&
// parentJobTaskBatchList.stream()
// .map(JobTaskBatch::getOperationReason)
// .filter(Objects::nonNull)
// .anyMatch(JobOperationReasonEnum.WORKFLOW_SUCCESSOR_SKIP_EXECUTION::contains)) {
// workflowBatchHandler.complete(taskExecute.getWorkflowTaskBatchId(), workflowTaskBatch);
// return;
// }
WorkflowNode parentWorkflowNode = workflowNodeMap.get(taskExecute.getParentId());
// 失败策略处理
@ -153,7 +167,8 @@ public class WorkflowExecutorActor extends AbstractActor {
}
// 决策节点
if (Objects.nonNull(parentWorkflowNode) && WorkflowNodeTypeEnum.DECISION.getType() == parentWorkflowNode.getNodeType()) {
if (Objects.nonNull(parentWorkflowNode)
&& WorkflowNodeTypeEnum.DECISION.getType() == parentWorkflowNode.getNodeType()) {
// 获取决策节点子节点
Set<Long> successors = graph.successors(parentWorkflowNode.getId());
@ -163,9 +178,10 @@ public class WorkflowExecutorActor extends AbstractActor {
// 过滤掉非当前决策节点【ParentId】的子节点
&& successors.contains(workflowNode.getId())).collect(Collectors.toList());
} else {
if (!brotherNodeIsComplete(taskExecute, brotherNode, jobTaskBatchMap, parentWorkflowNode)) {
return;
}
// TODO 不通过兄弟节点去控制是否执行后续节点
// if (!brotherNodeIsComplete(taskExecute, brotherNode, jobTaskBatchMap, parentWorkflowNode)) {
// return;
// }
workflowNodes = workflowNodes.stream()
// 去掉父节点
@ -183,6 +199,7 @@ public class WorkflowExecutorActor extends AbstractActor {
// 只会条件节点会使用
Object evaluationResult = null;
log.info("待执行的节点为. workflowNodes:[{}]", StreamUtils.toList(workflowNodes, WorkflowNode::getId));
for (WorkflowNode workflowNode : workflowNodes) {
// 批次已经存在就不在重复生成
@ -191,6 +208,41 @@ public class WorkflowExecutorActor extends AbstractActor {
continue;
}
// 决策当前节点要不要执行
Set<Long> predecessors = graph.predecessors(workflowNode.getId());
boolean predecessorsComplete = arePredecessorsComplete(predecessors, jobTaskBatchMap, workflowNode);
if (!SystemConstants.ROOT.equals(taskExecute.getParentId()) && !predecessorsComplete) {
continue;
}
// if (!SystemConstants.ROOT.equals(taskExecute.getParentId())) {
// boolean isConinue = true;
// for (final Long predecessor : predecessors) {
// if (SystemConstants.ROOT.equals(predecessor)) {
// continue;
// }
// List<JobTaskBatch> jobTaskBatches = jobTaskBatchMap.get(predecessor);
// // 说明此节点未执行, 继续等待执行完成
// if (CollUtil.isEmpty(jobTaskBatches)) {
// SnailJobLog.LOCAL.info("批次为空存在未完成的兄弟节点. [{}] 待执行节点:[{}]", predecessor, workflowNode.getId());
// isConinue = Boolean.FALSE;
// continue;
// }
//
// boolean isCompleted = jobTaskBatches.stream().anyMatch(
// jobTaskBatch -> JobTaskBatchStatusEnum.NOT_COMPLETE.contains(jobTaskBatch.getTaskBatchStatus()));
// if (isCompleted) {
// SnailJobLog.LOCAL.info("存在未完成的兄弟节点. [{}] 待执行节点:[{}]", predecessor, workflowNode.getId());
// isConinue = Boolean.FALSE;
// }
// }
//
// // TODO 理论上都不会执行应该 return
// if (!isConinue) {
// continue;
// }
//
// }
// 执行DAG中的节点
WorkflowExecutor workflowExecutor = WorkflowExecutorFactory.getWorkflowExecutor(workflowNode.getNodeType());
@ -202,6 +254,11 @@ public class WorkflowExecutorActor extends AbstractActor {
context.setTaskBatchId(taskExecute.getTaskBatchId());
context.setTaskExecutorScene(taskExecute.getTaskExecutorScene());
context.setWfContext(workflowTaskBatch.getWfContext());
// 这里父节点取最新的批次判断状态
if (CollUtil.isNotEmpty(parentJobTaskBatchList)) {
context.setParentOperationReason(parentJobTaskBatchList.get(0).getOperationReason());
}
workflowExecutor.execute(context);
evaluationResult = context.getEvaluationResult();
@ -209,31 +266,32 @@ public class WorkflowExecutorActor extends AbstractActor {
}
private boolean brotherNodeIsComplete(WorkflowNodeTaskExecuteDTO taskExecute, Set<Long> brotherNode,
Map<Long, List<JobTaskBatch>> jobTaskBatchMap, WorkflowNode parentWorkflowNode) {
if (SystemConstants.ROOT.equals(taskExecute.getParentId())) {
return Boolean.TRUE;
}
private boolean arePredecessorsComplete(Set<Long> predecessors,
Map<Long, List<JobTaskBatch>> jobTaskBatchMap, WorkflowNode waitExecWorkflowNode) {
// 决策节点不需要等待其他的兄弟节点是否完成，一个完成直接流转到后继节点
if (WorkflowNodeTypeEnum.DECISION.getType() == parentWorkflowNode.getNodeType()) {
return Boolean.TRUE;
}
// if (WorkflowNodeTypeEnum.DECISION.getType() == parentWorkflowNode.getNodeType()) {
// return Boolean.TRUE;
// }
// 判断所有节点是否都完成
for (final Long nodeId : brotherNode) {
for (final Long nodeId : predecessors) {
if (SystemConstants.ROOT.equals(nodeId)) {
continue;
}
List<JobTaskBatch> jobTaskBatches = jobTaskBatchMap.get(nodeId);
// 说明此节点未执行, 继续等待执行完成
if (CollUtil.isEmpty(jobTaskBatches)) {
SnailJobLog.LOCAL.debug("存在未完成的兄弟节点. [{}]", nodeId);
SnailJobLog.LOCAL.info("批次为空存在未完成的兄弟节点. [{}] 待执行节点:[{}]", nodeId,
waitExecWorkflowNode.getId());
return Boolean.FALSE;
}
boolean isCompleted = jobTaskBatches.stream().anyMatch(
jobTaskBatch -> JobTaskBatchStatusEnum.NOT_COMPLETE.contains(jobTaskBatch.getTaskBatchStatus()));
if (isCompleted) {
SnailJobLog.LOCAL.debug("存在未完成的兄弟节点. [{}]", nodeId);
SnailJobLog.LOCAL.info("存在未完成的兄弟节点. [{}] 待执行节点:[{}]", nodeId, waitExecWorkflowNode.getId());
return Boolean.FALSE;
}
}

6
snail-job-server/snail-job-server-job-task/src/main/java/com/aizuda/snailjob/server/job/task/support/executor/workflow/DecisionWorkflowExecutor.java
View File

@ -21,6 +21,7 @@ import com.aizuda.snailjob.template.datasource.persistence.po.JobTask;
import com.aizuda.snailjob.template.datasource.persistence.po.JobTaskBatch;
import com.aizuda.snailjob.template.datasource.persistence.po.WorkflowTaskBatch;
import com.baomidou.mybatisplus.core.conditions.query.LambdaQueryWrapper;
import com.google.common.collect.Sets;
import lombok.RequiredArgsConstructor;
import lombok.extern.slf4j.Slf4j;
import org.springframework.stereotype.Component;
@ -28,6 +29,9 @@ import org.springframework.stereotype.Component;
import java.util.Objects;
import java.util.Optional;
import static com.aizuda.snailjob.common.core.enums.JobOperationReasonEnum.WORKFLOW_DECISION_FAILED;
import static com.aizuda.snailjob.common.core.enums.JobOperationReasonEnum.WORKFLOW_NODE_NO_REQUIRED;
/**
* @author xiaowoniu
* @date 2023-12-24 08:17:11
@ -59,7 +63,7 @@ public class DecisionWorkflowExecutor extends AbstractWorkflowExecutor {
Boolean result = (Boolean) Optional.ofNullable(context.getEvaluationResult()).orElse(Boolean.FALSE);
if (result) {
if (result || (Sets.newHashSet(WORKFLOW_NODE_NO_REQUIRED.getReason(), WORKFLOW_DECISION_FAILED.getReason()).contains( context.getParentOperationReason()))) {
// 多个条件节点直接是或的关系，只要一个成功其他节点就取消且是无需处理状态
taskBatchStatus = JobTaskBatchStatusEnum.CANCEL.getStatus();
jobTaskStatus = JobTaskStatusEnum.CANCEL.getStatus();

18
snail-job-server/snail-job-server-job-task/src/main/java/com/aizuda/snailjob/server/job/task/support/executor/workflow/JobTaskWorkflowExecutor.java
View File

@ -10,11 +10,17 @@ import com.aizuda.snailjob.server.job.task.dto.JobTaskPrepareDTO;
import com.aizuda.snailjob.server.job.task.support.JobTaskConverter;
import com.aizuda.snailjob.template.datasource.persistence.po.JobTask;
import com.aizuda.snailjob.template.datasource.persistence.po.JobTaskBatch;
import com.google.common.collect.Sets;
import lombok.RequiredArgsConstructor;
import org.springframework.stereotype.Component;
import java.util.Objects;
import static com.aizuda.snailjob.common.core.enums.JobOperationReasonEnum.WORKFLOW_DECISION_FAILED;
import static com.aizuda.snailjob.common.core.enums.JobOperationReasonEnum.WORKFLOW_NODE_CLOSED_SKIP_EXECUTION;
import static com.aizuda.snailjob.common.core.enums.JobOperationReasonEnum.WORKFLOW_NODE_NO_REQUIRED;
import static com.aizuda.snailjob.common.core.enums.JobOperationReasonEnum.WORKFLOW_SUCCESSOR_SKIP_EXECUTION;
/**
* @author xiaowoniu
* @date 2023-12-24 08:09:14
@ -36,7 +42,8 @@ public class JobTaskWorkflowExecutor extends AbstractWorkflowExecutor {
@Override
protected void afterExecute(WorkflowExecutorContext context) {
if (Objects.equals(context.getWorkflowNodeStatus(), StatusEnum.YES.getStatus())) {
if (!Sets.newHashSet(WORKFLOW_NODE_CLOSED_SKIP_EXECUTION.getReason(), WORKFLOW_NODE_NO_REQUIRED.getReason(), WORKFLOW_DECISION_FAILED.getReason())
.contains(context.getOperationReason())) {
return;
}
@ -64,11 +71,14 @@ public class JobTaskWorkflowExecutor extends AbstractWorkflowExecutor {
if (Objects.equals(context.getWorkflowNodeStatus(), StatusEnum.NO.getStatus())) {
context.setTaskBatchStatus(JobTaskBatchStatusEnum.CANCEL.getStatus());
context.setOperationReason(JobOperationReasonEnum.WORKFLOW_NODE_CLOSED_SKIP_EXECUTION.getReason());
context.setOperationReason(WORKFLOW_NODE_CLOSED_SKIP_EXECUTION.getReason());
context.setJobTaskStatus(JobTaskStatusEnum.CANCEL.getStatus());
// 执行下一个节点
workflowTaskExecutor(context);
} else if (Sets.newHashSet(WORKFLOW_NODE_NO_REQUIRED.getReason(), WORKFLOW_DECISION_FAILED.getReason()).contains( context.getParentOperationReason())) {
// 针对无需处理的批次直接新增一个记录
context.setTaskBatchStatus(JobTaskBatchStatusEnum.CANCEL.getStatus());
context.setOperationReason(JobOperationReasonEnum.WORKFLOW_NODE_NO_REQUIRED.getReason());
context.setJobTaskStatus(JobTaskStatusEnum.CANCEL.getStatus());
} else {
invokeJobTask(context);
}

11
snail-job-server/snail-job-server-job-task/src/main/java/com/aizuda/snailjob/server/job/task/support/executor/workflow/WorkflowExecutorContext.java
View File

@ -38,6 +38,16 @@ public class WorkflowExecutorContext {
*/
private Long parentWorkflowNodeId;
/**
* TODO 父节点批次状态
*/
private Integer parentJobTaskStatus;
/**
* 父节点批次操作原因状态
*/
private Integer parentOperationReason;
/**
* 任务属性
*/
@ -107,4 +117,5 @@ public class WorkflowExecutorContext {
* 工作流全局上下文
*/
private String wfContext;
}

2
snail-job-server/snail-job-server-starter/src/main/java/com/aizuda/snailjob/server/starter/schedule/OfflineNodeSchedule.java
View File

@ -48,7 +48,7 @@ public class OfflineNodeSchedule extends AbstractSchedule implements Lifecycle {
.le(ServerNode::getExpireAt, endTime));
if (CollUtil.isNotEmpty(serverNodes)) {
// 先删除DB中需要下线的机器
serverNodeMapper.deleteBatchIds(StreamUtils.toSet(serverNodes, ServerNode::getId));
serverNodeMapper.deleteByIds(StreamUtils.toSet(serverNodes, ServerNode::getId));
}
Set<RegisterNodeInfo> allPods = CacheRegisterTable.getAllPods();

2
snail-job-server/snail-job-server-web/src/main/java/com/aizuda/snailjob/server/web/service/handler/WorkflowHandler.java
View File

@ -190,7 +190,7 @@ public class WorkflowHandler {
graph, version);
} else {
if (WorkflowNodeTypeEnum.DECISION.getType() == nodeConfig.getNodeType()) {
throw new SnailJobServerException("决策节点不能作为叶子节点");
throw new SnailJobServerException("决策节点或者决策节点的后继节点不能作为叶子节点");
}
// 叶子节点记录一下