如何为Spring Batch内置指标添加基于作业参数的自定义标签

方案一：通过ThreadLocal+MeterFilter扩展内置指标 #

MeterFilter本身是全局组件，无法直接获取作业参数，但可以借助ThreadLocal传递作业上下文的方式实现需求，步骤如下：

1. 编写ThreadLocal工具类临时存储作业参数

public class JobParameterHolder {
    private static final ThreadLocal<Map<String, Object>> PARAMS = new ThreadLocal<>();

    public static void setParams(Map<String, Object> params) {
        PARAMS.set(params);
    }

    public static Map<String, Object> getParams() {
        return PARAMS.get();
    }

    public static void clear() {
        PARAMS.remove();
    }
}

2. 实现JobExecutionListener，在作业生命周期中传递/清理参数

@Component
public class JobParameterPassingListener implements JobExecutionListener {
    @Override
    public void beforeJob(JobExecution jobExecution) {
        JobParameters jobParameters = jobExecution.getJobParameters();
        Map<String, Object> paramMap = new HashMap<>();
        // 提取需要作为标签的作业参数，比如"businessType"
        jobParameters.getParameters().forEach((key, value) -> paramMap.put(key, value.getValue()));
        JobParameterHolder.setParams(paramMap);
    }

    @Override
    public void afterJob(JobExecution jobExecution) {
        JobParameterHolder.clear();
    }
}

3. 自定义MeterFilter，为spring.batch.*指标添加自定义标签

@Bean
public MeterFilter jobParameterMeterFilter() {
    return new MeterFilter() {
        @Override
        public Meter.Id map(Meter.Id id) {
            // 仅处理Spring Batch内置指标
            if (id.getName().startsWith("spring.batch.")) {
                Map<String, Object> params = JobParameterHolder.getParams();
                if (params != null && !params.isEmpty()) {
                    Tags tags = id.getTags();
                    // 将作业参数转为指标标签
                    for (Map.Entry<String, Object> entry : params.entrySet()) {
                        tags = tags.and(entry.getKey(), entry.getValue().toString());
                    }
                    return id.withTags(tags);
                }
            }
            return id;
        }
    };
}

4. 将Listener注册到目标Job

@Bean
public Job sampleJob(JobBuilderFactory jobBuilderFactory, Step sampleStep, JobParameterPassingListener listener) {
    return jobBuilderFactory.get("sampleJob")
            .listener(listener)
            .start(sampleStep)
            .build();
}

注意：该方案仅适用于单线程作业，多线程步骤场景下ThreadLocal可能出现参数串值问题，需结合步骤级Listener做线程内参数传递。

方案二：通过Job/Step监听器实现自定义指标（更稳妥） #

如果觉得ThreadLocal的方式过于繁琐，直接在监听器中实现自定义指标是更可控的选择，尤其适合复杂多线程场景：

1. 注入MeterRegistry，在监听器中自定义带标签的指标

@Component
public class CustomBatchMetricsListener implements JobExecutionListener, StepExecutionListener {
    private final MeterRegistry meterRegistry;

    public CustomBatchMetricsListener(MeterRegistry meterRegistry) {
        this.meterRegistry = meterRegistry;
    }

    @Override
    public void beforeJob(JobExecution jobExecution) {
        String jobName = jobExecution.getJobInstance().getJobName();
        String businessType = jobExecution.getJobParameters().getString("businessType");
        // 自定义作业启动指标，带上业务标签
        meterRegistry.counter("spring.batch.job.started", "job_name", jobName, "business_type", businessType).increment();
    }

    @Override
    public void afterJob(JobExecution jobExecution) {
        String jobName = jobExecution.getJobInstance().getJobName();
        String businessType = jobExecution.getJobParameters().getString("businessType");
        String status = jobExecution.getStatus().name();
        // 自定义作业完成指标，带上状态标签
        meterRegistry.counter("spring.batch.job.completed", "job_name", jobName, "business_type", businessType, "status", status).increment();
    }

    @Override
    public void beforeStep(StepExecution stepExecution) {
        String stepName = stepExecution.getStepName();
        String jobName = stepExecution.getJobExecution().getJobInstance().getJobName();
        String businessType = stepExecution.getJobExecution().getJobParameters().getString("businessType");
        // 自定义步骤启动指标
        meterRegistry.counter("spring.batch.step.started", "job_name", jobName, "step_name", stepName, "business_type", businessType).increment();
    }

    @Override
    public ExitStatus afterStep(StepExecution stepExecution) {
        String stepName = stepExecution.getStepName();
        String jobName = stepExecution.getJobExecution().getJobInstance().getJobName();
        String businessType = stepExecution.getJobExecution().getJobParameters().getString("businessType");
        String status = stepExecution.getStatus().name();
        // 自定义步骤完成指标
        meterRegistry.counter("spring.batch.step.completed", "job_name", jobName, "step_name", stepName, "business_type", businessType, "status", status).increment();
        return stepExecution.getExitStatus();
    }
}

2. 将监听器注册到Job/Step即可，注册方式与方案一一致。

方案选择建议 #

• 若希望复用Spring Batch原生的spring.batch.*指标，优先考虑方案一，但需注意线程安全问题；
• 若追求指标的可控性和扩展性，方案二更推荐，尤其是多线程批处理场景下能精准绑定作业参数。

内容的提问来源于stack exchange，提问作者ghostypants