Spring重试for循环

如何解决Spring重试for循环

我的RetryTemplate配置：

@Configuration
@EnableRetry
public class RetryTemplateConfig {

    @Value("${spring.retry.attempts}")
    private int maxAttempts;

    @Value("${spring.retry.period}")
    private long backOffPeriod;

    @Bean
    public RetryTemplate retryTemplate() {
        SimpleRetryPolicy retryPolicy = new SimpleRetryPolicy();
        retryPolicy.setMaxAttempts(maxAttempts);
        FixedBackOffPolicy backOffPolicy = new FixedBackOffPolicy();
        backOffPolicy.setBackOffPeriod(backOffPeriod);
        RetryTemplate retryTemplate = new RetryTemplate();
        retryTemplate.setRetryPolicy(retryPolicy);
        retryTemplate.setBackOffPolicy(backOffPolicy);
        return retryTemplate;
    }
}

计划的方法，该方法调用使用Retry的方法：

@Scheduled(cron = "${schedule.cron.update.users}")
    public void sendToUsers() throws Exception {
        log.info("Scheduled sending for users started");
        try {
            mailSender.sendToUsers();
        } catch (MessagingException | IOException | TemplateException e) {
            log.error("Error occurred while sending email message to users: {}",e.toString());
        }
        log.info("Scheduled sending for users finished");
    }

我要使用RetryTemplate的方法：

public void sendToUsers() throws Exception {
        String subject = reportGenerationService.getEmailSubjectForUser();
        Map<String,List<BadUtmMark>> utmMarksGroupedByEmail = userService.getUtmMarksGroupedByEmail(LocalDate.now());
        if (utmMarksGroupedByEmail.isEmpty()) {
            log.info("All's fine - no broken utm marks in database. Emails to users will not be send.");
        }
        for (Map.Entry<String,List<BadUtmMark>> pair : utmMarksGroupedByEmail.entrySet()) {
            retryTemplate.execute(retryContext -> {
                String emailTo = pair.getKey();
                List<BadUtmMark> badUtmMarks = pair.getValue();
                String report = reportGenerationService.getReportForUser(emailTo,badUtmMarks,template);
                MimeMessage mimeMessage = getMimeMessage(subject,report,Collections.singletonList(emailTo));
                log.info("Message will be sent to: {}; from: {}; with subject: {}",pair.getKey(),from,subject);
                mailSender.send(mimeMessage);
                return true;
            });
        }
    }

预期的行为：我想发送5个人的电子邮件。如果发生错误，请尝试再次向该用户发送电子邮件5次，然后如果重试已用尽，请继续操作，并在for循环中为下一个用户发送电子邮件。

真正的行为：如果发生错误，服务将捕获异常并停止循环。

如果我将重试逻辑移至此方法：

@Scheduled(cron = "${schedule.cron.update.users}")
public void sendToUsers() throws Exception {
    log.info("Scheduled sending for users started");
    try {
        retryTemplate.execute(retryContext - > {
            log.warn("Sending email to users. Attempt: {}",retryContext.getRetryCount());
            mailSender.sendToUsers();
            return true;
        });
    } catch (MessagingException | IOException | TemplateException e) {
        log.error("Error occurred while sending email message to users: {}",e.toString());
    }
    log.info("Scheduled sending for users finished");
}

它的效果更好，但仍达不到我的期望。在这种情况下，如果发生错误，服务将尝试再次发送电子邮件5次，但是如果重试已用尽，服务将停止循环。因此，如果其中一个用户发生错误，该服务将尝试为此用户再发送5次，然后停止运行，并忽略地图中的其他用户。

但是我想对地图中的每封电子邮件进行5次重试。我该怎么做？

解决方法

在您的第一个版本中，请改用此execute ...

    /**
     * Keep executing the callback until it either succeeds or the policy dictates that we
     * stop,in which case the recovery callback will be executed.
     *
     * @see RetryOperations#execute(RetryCallback,RecoveryCallback)
     * @param retryCallback the {@link RetryCallback}
     * @param recoveryCallback the {@link RecoveryCallback}
     * @throws TerminatedRetryException if the retry has been manually terminated by a
     * listener.
     */
    @Override
    public final <T,E extends Throwable> T execute(RetryCallback<T,E> retryCallback,RecoveryCallback<T> recoveryCallback) throws E {
        return doExecute(retryCallback,recoveryCallback,null);
    }

template.execute(context -> {
    ...
},context -> {
    logger.error("Failed to send to ...");
});

如果回调正常退出，则故障将“恢复”，并且异常不会重新引发。