生成的GUID是重复的，我的Java代码错误吗？

Question

According to the information I found on the Internet, the probability of GUID repetition is very small, so is there a problem in my code?

public class AccountTaskExecutorTask extends TimerTask {
    private static final Logger logger = Logger.getLogger(AccountTaskExecutorTask.class);
    private TellerDbCore.AccountTask.Builder aTask = null;

    public AccountTaskExecutorTask(TellerDbCore.AccountTask.Builder aTask) {
        this.aTask = aTask;
    }

    public static void schedule(TellerDbCore.AccountTask.Builder aTask) {
        Timer timer = new Timer();
        timer.schedule(new AccountTaskExecutorTask(aTask), 100L);
    }

    @Override
    public void run() {
        try {
            DataBaseStore dataBaseStore = null;
            try {
                dataBaseStore = DbHelper.getTransactableDbStore();
                invest(aTask, dataBaseStore);
                dataBaseStore.commitAndClose();
                dataBaseStore = null;
            } catch (Exception e) {
                logger.error("", e);
                if (dataBaseStore != null) {
                    dataBaseStore.rollbackAndClose();
                    dataBaseStore = null;
                }
            }
        } catch (Exception e) {
            logger.error("", e);
        }
    }


    private static void invest(TellerDbCore.AccountTask.Builder theTask, DataBaseStore dataBaseStore) throws Exception {
            switch (theTask.getTaskStatus()) {
                case TS_READY:
                    List<TellerDbCore.AccountSubTask.Builder> subList = AccountTaskHelper.querySubAtByMtId(theTask.getTaskId(), dataBaseStore);
                    if (subList.size() == 0) {
                        NewMethods.newAccountSubTask(theTask.getTaskId(), GUID.generateGUID().toLowerCase(),
                                theTask.getAccountId(), theTask.getProductType(), theTask.getTaskType(), theTask.getAmount(), dataBaseStore);
                    }
                    UpdateMethods.updateAccountTask(theTask.getTaskId(), null, BtsDbBase.TaskStatus.TS_PROCESSING, dataBaseStore);
                    break;
                case TS_PROCESSING:
                    break;
                case TS_SUCCESSED:
                case TS_FAILED:
                    break;
                default:
                    logger.info( theTask.getTaskStatus());
                    break;
            }
    }
}

In this code, the GUID is generated by GUID.generateGUID().toLowerCase() , the calss of package is oscore-2.2.4.jar . NewMethods.newAccountSubTask is to add data to the database.

this is the code of GUID

package com.opensymphony.util;

import com.opensymphony.module.random.Yarrow;
import java.math.BigInteger;

public final class GUID
{
  private static Yarrow rnd = new Yarrow();

  public static String generateFormattedGUID()
  {
    String guid = generateGUID();

    return guid.substring(0, 8) + '-' + guid.substring(8, 12) + '-' + guid.substring(12, 16) + '-' + guid.substring(16, 20) + '-' + guid.substring(20);
  }

  public static String generateGUID()
  {
    return new BigInteger(165, rnd).toString(36).toUpperCase();
  }
}

Answer 1

Let's say that the number of potential values of GUID is n, a very large natural number. Let's also say that the number of already chosen values is k. Now, the probability (p) of having a duplicate in values depends on n and k. n is a very large natural number, but let's see the possible values of k (I am using the assumption that all the possible values have largely the same probability):

k == 0

In this case you are getting the very first GUID and it is impossible to get a duplicate. p == 0

k === n

In this case you already have all the possible values and it is impossible to not get a duplicate. p == 1

0 <= k <= n

In this case, the probabiliti of getting a new value is:

p == (n - k) / n

because we need to exclude the already chosen k elements. On the other hand, the probability of getting a duplicate is

p' == k / n

so with the increase of k, p decreases and p' increases. Note, that the two first cases are particular cases of this general case. Now, what could be the solution, so that you will not get a duplicate?

Well, you are using a lower case, which is basically reduces the possible values to half. If you avoid using lower cases, then n practically doubles. Also, you can use two concatenated guid values for your new values, which will never be duplicated for your values used previously, since they differ in String length. Or, you could use an id before the gui String.