分佈式紅鎖的加鎖失敗的設計原理
1.先把3臺 redis key全部清空(爲了不受debug干擾,必須先刪除鎖)
127.0.0.1:6379> flushdb
OK
都設置爲30分鐘超時 過期
2.isLock = redLock.tryLock(10006030, 10006030, TimeUnit.MILLISECONDS);
試驗步驟:
- 先啓動3臺redis實例,然後啓動springboot
- springboot啓動成功後,停2臺redis實例。(不能先停2臺redis,不然springboot起不來)
步驟1:springboot啓動成功後,停2臺redis實例。
[root@node2 ~]# docker ps
CONTAINER ID IMAGE COMMAND CREATED STATUS PORTS NAMES
1e6e3900e68c redis:5.0.7 "docker-entrypoint.s…" 6 days ago Up 4 minutes 0.0.0.0:6383->6379/tcp redis-master-3
1b9030d50927 redis:5.0.7 "docker-entrypoint.s…" 6 days ago Up 4 minutes 0.0.0.0:6382->6379/tcp redis-master-2
c86403dcb3d8 redis:5.0.7 "docker-entrypoint.s…" 6 days ago Up 8 hours 0.0.0.0:6381->6379/tcp redis-master-1
[root@node2 ~]#
[root@node2 ~]#
[root@node2 ~]# docker stop redis-master-3 redis-master-2
redis-master-3
redis-master-2
@Override
public boolean tryLock(long waitTime, long leaseTime, TimeUnit unit) throws InterruptedException {
// try {
// return tryLockAsync(waitTime, leaseTime, unit).get();
// } catch (ExecutionException e) {
// throw new IllegalStateException(e);
// }
long newLeaseTime = -1;
if (leaseTime != -1) {
if (waitTime == -1) {
newLeaseTime = unit.toMillis(leaseTime);
} else {
newLeaseTime = unit.toMillis(waitTime)*2;
}
}
long time = System.currentTimeMillis();
long remainTime = -1;
if (waitTime != -1) {
remainTime = unit.toMillis(waitTime);
}
long lockWaitTime = calcLockWaitTime(remainTime);
//步驟2:計算可以容忍接受加鎖失敗節點個數限制(N-(N/2+1))=(3-(3/2+1))=1
int failedLocksLimit = failedLocksLimit(); ==1
List<RLock> acquiredLocks = new ArrayList<>(locks.size());
for (ListIterator<RLock> iterator = locks.listIterator(); iterator.hasNext();) {
RLock lock = iterator.next();
boolean lockAcquired;
try {
if (waitTime == -1 && leaseTime == -1) {
lockAcquired = lock.tryLock();
} else {
long awaitTime = Math.min(lockWaitTime, remainTime);
lockAcquired = lock.tryLock(awaitTime, newLeaseTime, TimeUnit.MILLISECONDS);
}
} catch (RedisResponseTimeoutException e) {
unlockInner(Arrays.asList(lock));
lockAcquired = false;
} catch (Exception e) {
lockAcquired = false;
}
//步驟3:拿鎖成功後,統計成功的redis實例數
if (lockAcquired) {
acquiredLocks.add(lock);
} else {//步驟4:拿鎖失敗的話,那就複雜了
計算可以容忍接受加鎖失敗節點數,是否達到?(N-(N/2+1))=(3-(3/2+1))=1
如果已經達到,就認定最終申請鎖失敗,則沒有必要繼續從後面的節點申請了
因爲紅鎖算法要求至少N/2+1=3/2+1=2個節點都加鎖成功了,纔算最終的鎖申請成功。
//剛好N/2+1=3/2+1=2是成功的,例如 3個或2個都是成功,就退出,代表獲取鎖成功。
if (locks.size() - acquiredLocks.size() == failedLocksLimit()) {
break;
}
//步驟B:例如redis3個節點,第一個成功,第二失敗,第三失敗,,就直接進入failedLocksLimit=0
if (failedLocksLimit == 0) {
//解鎖第一個成功的
unlockInner(acquiredLocks);
//等待時間已經消化完就,就直接返回false,獲取鎖失敗。
if (waitTime == -1) {
return false;
}
//計算可以容忍接受加鎖失敗節點個數限制(N-(N/2+1))=(3-(3/2+1))=1
failedLocksLimit = failedLocksLimit();
//清除成功的list,例如redis3個節點,第一個成功,第二失敗,第三失敗,就是吧第一個清除,然後重新來。
acquiredLocks.clear();
//把iterator的座標重新初始化,然後重新進入新的循環
// reset iterator
while (iterator.hasPrevious()) {
iterator.previous();
}
//總結:failedLocksLimit == 0的設計原理,就是讓for循環一直循環下去,除非出現2種情況才退出死循環
1.waitTime 或 remainTime消化完,退出循環
2.redis實例恢復正常 (例如啓動 docker start redis-master-3 redis-master-2)就正常拿到鎖了
} else {
//步驟A:加鎖失敗節點個數限制-1,然後繼續循環,例如redis3個節點,第一個成功,第二失敗,限制failedLocksLimit=0
failedLocksLimit--;
}
}
//步驟5: 只有等到remainTime消化完,退出死循環
if (remainTime != -1) {
remainTime -= System.currentTimeMillis() - time;
time = System.currentTimeMillis();
if (remainTime <= 0) {
unlockInner(acquiredLocks);
return false;
}
}
}
if (leaseTime != -1) {
List<RFuture<Boolean>> futures = new ArrayList<>(acquiredLocks.size());
for (RLock rLock : acquiredLocks) {
RFuture<Boolean> future = ((RedissonLock) rLock).expireAsync(unit.toMillis(leaseTime), TimeUnit.MILLISECONDS);
futures.add(future);
}
for (RFuture<Boolean> rFuture : futures) {
rFuture.syncUninterruptibly();
}
}
return true;
}
基於以上代碼,我們分爲5個步驟分析,得出一個結論:
1.整個算法過程中都是圍繞N/2+1=3/2+1=2的成功個數來設計的,如果拿不到N/2+1=3/2+1=2,算法會一直死循環下去,,除非出現2種情況才退出死循環
1.waitTime 或 remainTime消化完,退出循環
2.redis實例恢復正常 (例如啓動 docker start redis-master-3 redis-master-2)就正常拿到鎖了
2.這種設計,要特別小心waitTime,這個waitTime一定要設置短,不然redis宕機的情況下,會出現死循環,整個系統卡死。