apache/dolphinscheduler · error · ServiceException

{joined failure messages, e.g. Failed do action <executeType

Error message

{joined failure messages, e.g. Failed do action <executeType> on workflowInstance: <id>reason: <error>}

What it means

ExecutorController.batchControlWorkflowInstance applies an action (executeType) to each workflow instance in the batch, collecting per-instance errors. If any instance failed, it joins all 'Failed do action <executeType> on workflowInstance: <id>reason: <error>' lines with newlines and throws a single ServiceException. This is an aggregate failure message, not one specific cause.

Source

Thrown at dolphinscheduler-api/src/main/java/org/apache/dolphinscheduler/api/controller/ExecutorController.java:356

    public Result<Void> batchControlWorkflowInstance(@RequestAttribute(value = Constants.SESSION_USER) User loginUser,
                                                     @RequestParam("workflowInstanceIds") String workflowInstanceIds,
                                                     @RequestParam("executeType") ExecuteType executeType) {

        String[] workflowInstanceIdArray = workflowInstanceIds.split(Constants.COMMA);
        List<String> errorMessage = new ArrayList<>();
        for (String strWorkflowInstanceId : workflowInstanceIdArray) {
            int workflowInstanceId = Integer.parseInt(strWorkflowInstanceId);
            try {
                execService.controlWorkflowInstance(loginUser, workflowInstanceId, executeType);
                log.info("Success do action {} on workflowInstance: {}", executeType, workflowInstanceId);
            } catch (Exception e) {
                errorMessage.add("Failed do action " + executeType + " on workflowInstance: " + workflowInstanceId
                        + "reason: " + e.getMessage());
                log.error("Failed do action {} on workflowInstance: {}, error: {}", executeType, workflowInstanceId, e);
            }
        }
        if (org.apache.commons.collections4.CollectionUtils.isNotEmpty(errorMessage)) {
            throw new ServiceException(String.join("\n", errorMessage));
        }
        return Result.success();
    }

    /**
     * execute task instance
     *
     * @param loginUser      login user
     * @param projectCode    project code
     * @param code           taskDefinitionCode
     * @param version        taskDefinitionVersion
     * @param warningGroupId warning group id
     * @param workerGroup    worker group
     * @return start task result code
     */
    @Operation(summary = "startTaskInstance", description = "RUN_TASK_INSTANCE_NOTES")
    @Parameters({
            @Parameter(name = "version", description = "VERSION", schema = @Schema(implementation = int.class, example = "1")),

View on GitHub (pinned to 02eac45a1b)

Solutions

  1. Read the per-line reasons in the thrown message to see which instance IDs failed and why, then retry only the failed ones.
  2. Before batching, verify each instance is in a state that allows the action (e.g. only RUNNING instances can be stopped).
  3. Reduce batch size and inspect/refresh the instance list so IDs are current and visible to the calling user.

Example fix

// before: blind batch
execClient.batchStop(allInstanceIds);
// after: filter to controllable instances first
List<Integer> stoppable = ids.stream().filter(id -> stateOf(id) == RUNNING).collect(toList());
execClient.batchStop(stoppable);
Defensive patterns

Strategy: try-catch

Validate before calling

List<Integer> eligible = instanceIds.stream()
    .filter(id -> {
        WorkflowInstance wi = find(id);
        return wi != null && wi.getState().isRunning() && canWrite(wi.getProjectCode());
    }).collect(toList());

Try / catch

try {
    executorController.batchControlWorkflowInstance(...);
} catch (ServiceException e) {
    Set<Integer> failed = Arrays.stream(e.getMessage().split("\n"))
        .map(l -> l.replaceAll(".*workflowInstance: (\\d+).*", "$1"))
        .map(Integer::valueOf)
        .collect(toSet());
    log.warn("Retrying batch without failed ids: {}", failed);
}

Prevention

When it happens

Trigger: Batch endpoints (e.g. /executor/batch-instance-execute) where at least one target workflow instance cannot be controlled — instance already finished, instance not found, state does not permit the action, or permission denied — producing at least one entry in errorMessage.

Common situations: Batch-stopping instances that already completed between selection and execution; operating on instances belonging to another project/user; stale UI lists referencing deleted instances.

Related errors

AI-assisted analysis of apache/dolphinscheduler@02eac45a1b (2026-09-06). Data as JSON: /api/errors/dd957e016b985f71. Report an issue: GitHub.