Concept: In a single-process system, when multiple threads can change a variable (variable shared variable) at the same time, it is necessary to synchronize the variable or code block so that the variable can be modified linearly to prevent data inconsistency or data pollution caused by concurrent modification of variables. In order for multiple threads to have only one thread executing a block of code at a time, there needs to be a flag somewhere that each thread can see and that can be set when the flag doesn’t exist. The remaining subsequent threads find that the tag already exists and wait for the thread that owns the tag to terminate the synchronization block and untag it before attempting to set the tag. This tag can be understood as a lock.
Note: In the case of a single machine (single JVM), where memory is shared between threads, concurrency can be resolved by simply using local locks. However, if the situation is distributed (multiple JVMS), thread A and thread B are probably not in the same JVM, so the thread lock will not work, and the distributed lock will be used to solve the problem
Let’s test the case without locking in a multi-threaded single machine,
One: the code example is as follows
public Map<String, List<Catelog2Vo>> getCatalogJson() {
String category = redisTemplate.opsForValue().get("category");
Map<String, List<Catelog2Vo>> map=null;
if(StringUtils.isEmpty(category)){
System.out.println("Missed cache --");
// Query the database
map= getCatalogJsonWithDb();
// Insert the cache
String s = JSONObject.toJSONString(map);
redisTemplate.opsForValue().set("category",s);
}else{
System.out.println("Cache hit --");
map= JSON.parseObject(category, new TypeReference<Map<String, List<Catelog2Vo>>>(){
}); / / inner classes
}
return map;
}
Copy the codeTwo: start 200 threads and loop for 3 times
3. Modify the code to add synchronized lock mechanism
public Map<String, List<Catelog2Vo>> getCatalogJsonWithDb() {
synchronized (this){
System.out.println("Start query database --");
// Query all categories to reduce database IO
List<CategoryEntity> selectList=this.baseMapper.selectList(null);// Do not pass the query criteria to query all
//List<CategoryEntity> level1 = this.listWithTree1();
// Query all level 1 categories, avoid direct query in database
List<CategoryEntity> level1 =getParentCid(selectList,0L);
Map<String,List<Catelog2Vo>> map=level1.stream().collect(Collectors.toMap(k->k.getCatId().toString(),v->{
// Query all sub-categories corresponding to the current sub-category
List<CategoryEntity> categoryEntities= getParentCid(selectList,v.getCatId());
log.info("The current primary classification ID is: {}, and all corresponding secondary classifications are: {}",v.getCatId(),categoryEntities);
// Encapsulate the result
List<Catelog2Vo> catelog2Vos=null;
if(categoryEntities! =null){
catelog2Vos=categoryEntities.stream().map(l2->{
Catelog2Vo catelog2Vo=new Catelog2Vo(v.getCatId().toString(),l2.getCatId().toString(),l2.getName(),null);
// Select vo as vo
List<CategoryEntity> level3Catelog= getParentCid(selectList,l2.getCatId());
log.info("The current secondary classification ID is: {}, and all corresponding tertiary classifications are: {}",l2.getCatId(),level3Catelog);
if(level3Catelog! =null){
List<Catelog2Vo.Category3Vo> collect=level3Catelog.stream().map(l3->{
Catelog2Vo.Category3Vo category3Vo=new Catelog2Vo.Category3Vo(l2.getCatId().toString(),l3.getCatId().toString(),l3.getName());
return category3Vo;
}).collect(Collectors.toList());
catelog2Vo.setCatalog3List(collect);
}
return catelog2Vo;
}).collect(Collectors.toList());
}
return catelog2Vos;
}));
returnmap; }}Copy the codeThe locking mechanism is very simple, directly add synchronized keyword to the database query method, remember to delete the key data in Redis before the test, because my back-end memory setting is not high, this time set 50 threads to run twice, start the JMeter test
Four: solve the locking time sequence problem
Sample code is as follows
public Map<String, List<Catelog2Vo>> getCatalogJsonWithDb() {
synchronized (this) {
// Get the lock from the cache again
String category = redisTemplate.opsForValue().get("category");
if(! StringUtils.isEmpty(category)) {Map<String, List<Catelog2Vo>> stringListMap = JSON.parseObject(category, new TypeReference<Map<String, List<Catelog2Vo>>>() {
});
return stringListMap;
} else {
System.out.println("Start query database --");
// Query all categories to reduce database IO
List<CategoryEntity> selectList = this.baseMapper.selectList(null);// Do not pass the query criteria to query all
//List<CategoryEntity> level1 = this.listWithTree1();
// Query all level 1 categories, avoid direct query in database
List<CategoryEntity> level1 = getParentCid(selectList, 0L);
Map<String, List<Catelog2Vo>> map = level1.stream().collect(Collectors.toMap(k -> k.getCatId().toString(), v -> {
// Query all sub-categories corresponding to the current sub-category
List<CategoryEntity> categoryEntities = getParentCid(selectList, v.getCatId());
log.info("The current primary classification ID is: {}, and all corresponding secondary classifications are: {}", v.getCatId(), categoryEntities);
// Encapsulate the result
List<Catelog2Vo> catelog2Vos = null;
if(categoryEntities ! =null) {
catelog2Vos = categoryEntities.stream().map(l2 -> {
Catelog2Vo catelog2Vo = new Catelog2Vo(v.getCatId().toString(), l2.getCatId().toString(), l2.getName(), null);
// Select vo as vo
List<CategoryEntity> level3Catelog = getParentCid(selectList, l2.getCatId());
log.info("The current secondary classification ID is: {}, and all corresponding tertiary classifications are: {}", l2.getCatId(), level3Catelog);
if(level3Catelog ! =null) {
List<Catelog2Vo.Category3Vo> collect = level3Catelog.stream().map(l3 -> {
Catelog2Vo.Category3Vo category3Vo = new Catelog2Vo.Category3Vo(l2.getCatId().toString(), l3.getCatId().toString(), l3.getName());
return category3Vo;
}).collect(Collectors.toList());
catelog2Vo.setCatalog3List(collect);
}
return catelog2Vo;
}).collect(Collectors.toList());
}
return catelog2Vos;
}));
// Write cache to redis
String s = JSONObject.toJSONString(map);
redisTemplate.opsForValue().set("category", s);
returnmap; }}}Copy the codeThere are two key codes: one is that the thread obtains the lock and then goes to Redis to confirm whether there is a cache; the other is to write the data into the cache after the query and release the lock
Five: Restart the service and enable the Jemeter for testing
