I have a list that is accessed by multiple background threads to update/read. Updates actions include both insertions and deletions.
To do this concurrently without synchronization problems, I am using a lock on a private readonly object in the class.
To minimize the time I need to lock the list when reading its data, I do a deep clone of it and return the deep clone and unlock the dictionary for insert/delete updates.
Due to this every read of the list increases the memory consumption of my service.
One point to note is that the inserts/deletes are internal to the class that contains the list. But the read is meant for public consumption.
My question is:
Is there any way, I can avoid cloning the list and still use it concurrently for reads using read/write locks?
public class ServiceCache
{
private static List<Users> activeUsers;
private static readonly object lockObject = new object();
private static ServiceCache instance = new ServiceCache();
public static ServiceCache Instance
{
get
{
return instance;
}
}
private void AddUser(User newUser)
{
lock (lockObject)
{
//... add user logic
}
}
private void RemoveUser(User currentUser)
{
lock (lockObject)
{
//... remove user logic
}
}
public List<Users> ActiveUsers
{
get
{
lock (lockObject)
{
//The cache returns deep copies of the users it holds, not links to the actual data.
return activeUsers.Select(au => au.DeepCopy()).ToList();
}
}
}
}
It sounds like you need to use the ConcurrentDictionary class, and create a key for each of the Users objects you are storing. Then it becomes as simple as this for adding / updating a user:
_dictionary.AddOrUpdate("key", (k, v) =>
{
return newUser;
}, (k, v) =>
{
return newUser;
});
And then for removing, you would do this:
Users value = null;
_dictionary.TryRemove("key", out value);
Getting the list of people would be super easy as well, since you would just need to do:
return _dictionary.Values.Select(x => x.Value).ToList();
Which should return a copy of the dictionary contents at that very moment.
And let the .NET runtime take care of the threading for you.
You can use a reader-writer lock to allow simultaneous reads.
However, it would be much faster to use a ConcurrentDictionary and thread-safe immutable values, then get rid of all synchronization.
Due to this every read of the list increases the memory consumption of
my service.
Why? Are the callers not releasing the reference? They need to, since the content of the dictionary can change.
What you are doing with copying is I think very close to how a Concurrent data structure, e.g. copy-on-write collection works, except that the caller cannot hold on to the reference.
A couple of other approaches:
Return the same copy to all callers till the collection gets modified. The returned collection should be immutable
Expose all the functionality the caller would want from the copy and use a single lock to work with the original list
Related
I have a memory cached object-cache, said cache can hold multiple types, and i want to add a lock on said {T} whenever given {T} is accessed.
My implementation:
readonly static IDictionary<Type, List<object>> _cache = new ConcurrentDictionary<Type, List<object>>();
private static List<object> FindTypeInCache(Type type)
{
List<object> list;
if (_cache.TryGetValue(type, out list))
{
return list;
}
else
{
_cache[type] = new List<object>();
}
return new List<object>();
}
public static T FindFirstBy<T>(Func<T, bool> predicate) where T : class
{
// Is this a valid lock locking only _cache[T] ? And not _cache as whole?
lock (_cache[typeof(T)])
{
return FindTypeInCache(typeof(T)).Cast<T>().Where(predicate).FirstOrDefault();
}
}
public static bool AddOrUpdate<T>(Func<T, bool> predicate, T entity) where T : class
{
lock (_cache[typeof(T)])
{
// Find Type cache.
List<object> list = FindTypeInCache(typeof(T));
// Look for old entity.
var e = list.Cast<T>().Where(predicate).FirstOrDefault();
// If no old record exists no problem we treat this as if its a new record.
if (e != null)
{
// Old record found removing it.
list.Remove(e);
}
// Regardless if object existed or not we add it to our Cache.
list.Add(entity);
_cache[typeof(T)] = list;
}
}
Is my implementation correct only locking down _cache[T] and not entire _cache as a whole when accessed?
There's a lot of things weird (or outright wrong) with your code.
First, you're using a ConcurrentDictionary, but you're not using it as a concurrent dictionary. For example, to initialize the list, you'd use the GetOrAddMethod:
private static List<object> FindTypeInCache(Type type)
{
return _cache.GetOrAdd(type, () => new List<object>());
}
Simple and thread-safe :)
Second, you're locking on _cache[type] - but even when there's no such type in the cache. This means a KeyNotFoundException.
Third, the only code you're protecting with the lock is the reading. But that quite likely isn't enough - at the very least, you need to also protect the writing with the same lock (especially tricky given the point above), and depending on your actual usage of the return value, the return value's mutations (and if it is indeed mutable, any reads of that mutable value as well).
In other words, you've only managed to protect the code that doesn't actually need protecting (if you use the correct method to update the dictionary)! The extra lock around the Where etc. might help slightly, but it certainly doesn't make the List access safe.
All that said, maybe there's a better solution anyway. You're using the cache using generic methods. Why not make the cache itself generic? This way, you'll avoid using a dictionary in the first place, because each of the generic types you're storing in the dictionary will get their own type - it also means you can initialize the List<T> in a static constructor safely. Any locking can than also be safely applied to all access to a specific generic cache rather than the "aggregate" cache you have now.
I have a Visual Studio 2008 C# .NET 3.5 project where I want to have a thread-safe pool of Foo objects.
public class FooPool
{
private object pool_lock_ = new object();
private Dictionary<int, Foo> foo_pool_ = new Dictionary<int, Foo>();
// ...
public void Add(Foo f)
{
lock (pool_lock_)
{
foo_pool_.Add(SomeFooDescriminator, f);
}
}
public Foo this[string key]
{
get { return foo_pool_[key]; }
set { lock (pool_lock_) { foo_pool_[key] = value; } }
}
public IEnumerable<Foo> Foos
{
get
{
lock (pool_lock_)
{
// is this thread-safe?
return foo_pool_.Select(x => x.Value);
}
}
}
}
Is the public IEnumerable<Foo> Foos { get; } function thread-safe? Or, do I need to clone the result and return a new list?
No, it isn't.
If another thread adds to the dictionary while your caller enumerates that, you'll get an error.
Instead, you can do:
lock (pool_lock_) {
return foo_pool.Values.ToList();
}
Is the IEnumerable<Foo> Foos { get; } function thread-safe?
No.
Or, do I need to clone the result and return a new list?
No, because that's not right either. A threadsafe method that gives the wrong answer is not very useful.
If you lock and make a copy then the thing you are returning is a snapshot of the past. The collection could be changed to be completely different the moment the lock is released. If you make this threadsafe by making a copy then you are now handing a bag full of lies to your caller.
When you are dealing with single-threaded code, a reasonable model is that everything is staying the same unless you take specific measures to change a thing. That is not a reasonable model in multi-threaded code. In multi-threaded code, you should assume the opposite: everything is constantly changing unless you take specific measures (such as a lock) to ensure that things are not changing. What is the good of handing out a sequence of Foos that describe the state of the world in the distant past, hundreds of nanoseconds ago? The entire world could be different in that amount of time.
It is not thread safe. You need to return ToList():
return foo_pool_.Select(x => x.Value).ToList();
Careful of deferred execution!
The fact is the actual code runs after the lock has exited.
// Don't do this
lock (pool_lock_)
{
return foo_pool_.Select(x => x.Value); // This only prepares the statement, does not run it
}
You may want to consider a SynchronizedCollection,
SynchronizedCollection Class
Provides a thread-safe collection that contains objects of a type specified by the generic parameter as elements.
http://msdn.microsoft.com/en-us/library/ms668265.aspx
If you'll lock on every read access you'll end with very bad performance. And in suggestions to use toList you'll also allocate memory every time.
If you using .NET 4 just use ConcurrentDictionary class from new thread safe collections. They will provide very fast (lock free) mechanisms for accessing data from multiple threads.
http://msdn.microsoft.com/en-us/library/dd997305.aspx
If you are using old .NET version I would suggest you to use for cycle with count variable instead of foreach it will work if you only add elements without removing them (as in your example)
I know that is wrong to use lock(this) or any shared object.
I wonder if this usage is OK?
public class A
{
private readonly object locker = new object();
private List<int> myList;
public A()
{
myList = new List<int>()
}
private void MethodeA()
{
lock(locker)
{
myList.Add(10);
}
}
public void MethodeB()
{
CallToMethodInOtherClass(myList);
}
}
public class OtherClass
{
private readonly object locker = new object();
public CallToMethodInOtherClass(List<int> list)
{
lock(locker)
{
int i = list.Count;
}
}
}
Is this thread safe? In OtherClass we lock with a private object so if the class A lock with its private lock can the list still change in the the lock block in OtherClass?
No, it's not thread safe. Add and Count may be executed at the "same" time. You have two different lock objects.
Always lock your own lock object when passing the list:
public void MethodeB()
{
lock(locker)
{
CallToMethodInOtherClass(myList);
}
}
No this is not thread safe. To make it thread safe you can use lock on static objects because they are shared between threads, this may cause deadlocks in the code but it can be handle by maintaining proper order for locking. There is a performance cost associated with lock so use it wisely.
Hope this helps
No, this is not thread-safe. A.MethodeA and OtherClass.CallToMethodInOtherClass are locking on different objects, so they're not mutually exclusive. If you need to protect the access to the list, don't pass it to external code, keep it private.
No, that is not thread-safe.
Your 2 methods are locking on 2 different objects, they will not lock out each other.
Because CallToMethodInOtherClass() only retrieves the value of Count nothing will go horribly wrong. But the lock() around it is useless and misleading.
If the method would make changes in the list you would have a nasty problem. To solve it, change MethodeB:
public void MethodeB()
{
lock(locker) // same instance as MethodA is using
{
CallToMethodInOtherClass(myList);
}
}
No, they have to lock the same object. With your code they both lock on a different and each call could be executed simultaneous.
To make the code thread safe place a lock in MethodeB or use the list itself as lock object.
It actually is thread-safe (purely as a matter of an implementation detail on Count), but:
Thread-safe snippets of code do not a thread-safe application make. You can combine different thread-safe operations into non-thread-safe operations. Indeed, much non-thread-safe code can be broken down into smaller pieces all of which are thread-safe on their own.
It's not thread-safe for the reason you were hoping, which means that extending it further would not be thread-safe.
This code would be thread-safe:
public void CallToMethodInOtherClass(List<int> list)
{
//note we've no locks!
int i = list.Count;
//do something with i but don't touch list again.
}
Call it with any list, and it'll give i a value based on the state of that list, regardless of what other threads are up to. It will not corrupt list. It will not give i an invalid value.
So while this code is also thread-safe:
public void CallToMethodInOtherClass(List<int> list)
{
Console.WriteLine(list[93]); // obviously only works if there's at least 94 items
// but that's nothing to do with thread-safety
}
This code would not be thread-safe:
public void CallToMethodInOtherClass(List<int> list)
{
lock(locker)//same as in the question, different locker to that used elsewhere.
{
int i = list.Count;
if(i > 93)
Console.WriteLine(list[93]);
}
}
Before going further, the two bits I described as thread-safe are not promised to be by the spec for List. Conservative coding would assume they are not thread-safe rather than depending upon implementation details, but I'm going to depend on the implementation details because it affects the question of how to use locks in an important way:
Because there is code operating on list that is not acquiring the lock on locker first, that code is not prevented from running concurrently with CallToMethodInOtherClass. Now, while list.Count is thread-safe and list[93] is tread-safe,* the combination of the two where we depend on the first to ensure that the second works is not thread-safe. Because code outside the lock can affect list, it's possible for code to call Remove or Clear in between Count assuring us that list[93] would work, and list[93] being called.
Now, if we know that list is only ever added to, that's fine, even if a resize is happening concurrently we'll end up with the value of list[93] either way. If something is writing to list[93] and it's a type that .NET will write to atomically (and int is one such type), we'll end up with either the old one or the new one, just as if we'd locked correctly we'd get the old or the new depending on which thread go the lock first. Again, this is an implementation detail not a specified promise, I'm stating this just to point out how the thread-safety given still results in non thread-safe code.
Moving this toward real code. We shouldn't assume that list.Count and list[93] is threadsafe because we weren't promised they would be and that could change, but even if we did have that promise, those two promises won't add up to a promise that they'd be thread-safe together.
The important thing is to use the same lock to protect blocks of code that can interfere with each other. Hence, consider the variant below that is guaranteed to be threadsafe:
public class ThreadSafeList
{
private readonly object locker = new object();
private List<int> myList = new List<int>();
public void Add(int item)
{
lock(locker)
myList.Add(item);
}
public void Clear()
{
lock(locker)
myList.Clear();
}
public int Count
{
lock(locker)
return myList.Count;
}
public int Item(int index)
{
lock(locker)
return myList[index];
}
}
This class is guaranteed to be thread-safe in everything it does. Without depending on any implementation details, there is no method here that will corrupt state or give incorrect results because of what another thread is doing with the same instance. The following code still doesn't work though:
// (l is a ThreadSafeList visible to multiple threads.
if(l.Count > 0)
Console.WriteLine(l[0]);
We've guaranteed the thread-safety of each call 100%, but we haven't guaranteed the combination, and we can't guarantee the combination.
There's two things we can do. We can add a method for the combination. Something like the following would be common for many classes specifically designed for multi-threaded use:
public bool TryGetItem(int index, out int value)
{
lock(locker)
{
if(l.Count > index)
{
value = l[index];
return true;
}
value = 0;
return false;
}
}
This makes the count test and the item retrieval part of a single operation which is guaranteed to be thread-safe.
Alternatively, and most often what we need to do, we have the lock happen at the place where the operations are grouped:
lock(lockerOnL)//used by every other piece of code operating on l
if(l.Count > 0)
Console.WriteLine(l[0]);
Of course, this makes the locks within ThreadSafeList redundant and just a waste of effort, space, and time. This is the main reason that most classes don't provide thread-safety on their instance members - since you can't meaningfully protect groups of calls on members from within the class, it's a waste of time trying to unless the thread-safety promises are very well specified and useful on their own.
To come back to the code in your question:
The lock in CallToMethodInOtherClass should be removed unless OtherClass has its own reason for locking internally. It can't make a meaningful promise that it won't be combined in a non-threadsafe way and adding more locks to a program just increases the complexity of analysing it to be sure there are no deadlocks.
The call to CallToMethodInOtherClass should be protected by the same lock as other operations in that class:
public void MethodeB()
{
lock(locker)
CallToMethodInOtherClass(myList);
}
Then as long as CallToMethodInOtherClass doesn't store myList somewhere it can be seen by other threads later on, it doesn't matter that CallToMethodInOtherClass isn't thread-safe because the only code that can access myList brings its own guarantee not to call it concurrently with other operations on myList.
The two important things are:
When something is described as "thread-safe", know just what it's promising by that, as there are different sorts of promise that fall under "thread-safe" and on its own it just means "I won't put this object into a nonsensical state", which while an important building block, is not a lot on its own.
Lock on groups of operations, with the same lock for each group that'll affect the same data, and guard the access to objects so that there can't possibly be another thread not playing ball with this.
*This is a very limited definition of thread-safe. Calling list[93] on a List<T> where T is a type that will be written and read atomically and we don't know whether it actually has at least 94 items is equally safe whether or not there are other threads operating on it. Of course, the fact that it can throw ArgumentOutOfRangeException in either case is not what most people would consider "safe", but the guarantee we have with multiple threads remains the same as with one. It's that we obtain a stronger guarantee by checking Count in a single thread but not in a multi-thread situation that leads me to describe that as not thread-safe; while that combo still won't corrupt state it can lead to an exception we'd assured ourselves couldn't happen.
Probably the easiest way to do the trick
public class A
{
private List<int> myList;
public A()
{
myList = new List<int>()
}
private void MethodeA()
{
lock(myList)
{
myList.Add(10);
}
}
public void MethodeB()
{
CallToMethodInOtherClass(myList);
}
}
public class OtherClass
{
public CallToMethodInOtherClass(List<int> list)
{
lock(list)
{
int i = list.Count;
}
}
}
Many of the answers have mentioned using a static readonly lock.
However, you really should try to avoid this static lock. It would be easy to create a deadlock where multiple threads are using the static lock.
What you could use instead is one of the .net 4 concurrent collections, these do provide some thread synchronisation on your behalf, so that you do not need to use the locking.
Take a look at the System.collections.Concurrent namespace.
For this example, you could use the ConcurrentBag<T> class.
Ass all the answers say these are different lock objects.
a simple way is to have a static lock object f.ex:
publc class A
{
public static readonly object lockObj = new object();
}
and in both classes use lock like:
lock(A.lockObj)
{
}
I've a class that contains a static collection to store the logged-in users in an ASP.NET MVC application. I just want to know about the below code is thread-safe or not. Do I need to lock the code whenever I add or remove item to the onlineUsers collection.
public class OnlineUsers
{
private static List<string> onlineUsers = new List<string>();
public static EventHandler<string> OnUserAdded;
public static EventHandler<string> OnUserRemoved;
private OnlineUsers()
{
}
static OnlineUsers()
{
}
public static int NoOfOnlineUsers
{
get
{
return onlineUsers.Count;
}
}
public static List<string> GetUsers()
{
return onlineUsers;
}
public static void AddUser(string userName)
{
if (!onlineUsers.Contains(userName))
{
onlineUsers.Add(userName);
if (OnUserAdded != null)
OnUserAdded(null, userName);
}
}
public static void RemoveUser(string userName)
{
if (onlineUsers.Contains(userName))
{
onlineUsers.Remove(userName);
if (OnUserRemoved != null)
OnUserRemoved(null, userName);
}
}
}
That is absolutely not thread safe. Any time 2 threads are doing something (very common in a web application), chaos is possible - exceptions, or silent data loss.
Yes you need some kind of synchronization such as lock; and static is usually a very bad idea for data storage, IMO (unless treated very carefully and limited to things like configuration data).
Also - static events are notorious for a good way to keep object graphs alive unexpectedly. Treat those with caution too; if you subscribe once only, fine - but don't subscribe etc per request.
Also - it isn't just locking the operations, since this line:
return onlineUsers;
returns your list, now unprotected. all access to an item must be synchronized. Personally I'd return a copy, i.e.
lock(syncObj) {
return onlineUsers.ToArray();
}
Finally, returning a .Count from such can be confusing - as it is not guaranteed to still be Count at any point. It is informational at that point in time only.
Yes, you need to lock the onlineUsers to make that code threadsafe.
A few notes:
Using a HashSet<string> instead of the List<string> may be a good idea, since it is much more efficient for operations like this (Contains and Remove especially). This does not change anything on the locking requirements though.
You can declare a class as "static" if it has only static members.
Yes you do need to lock your code.
object padlock = new object
public bool Contains(T item)
{
lock (padlock)
{
return items.Contains(item);
}
}
Yes. You need to lock the collection before you read or write to the collection, since multiple users are potentially being added from different threadpool workers. You should probably also do it on the count as well, though if you're not concerned with 100% accuracy that may not be an issue.
As per Lucero's answer, you need to lock onlineUsers. Also be careful what will clients of your class do with the onlineUsers returned from GetUsers(). I suggest you change your interface - for example use IEnumerable<string> GetUsers() and make sure the lock is used in its implementation. Something like this:
public static IEnumerable<string> GetUsers() {
lock (...) {
foreach (var element in onlineUsers)
yield return element;
// We need foreach, just "return onlineUsers" would release the lock too early!
}
}
Note that this implementation can expose you to deadlocks if users try to call some other method of OnlineUsers that uses lock, while still iterating over the result of GetUsers().
That code it is not thread-safe per se.
I will not make any suggestions relative to your "design", since you didn't ask any. I'll assume you found good reasons for those static members and exposing your list's contents as you did.
However, if you want to make your code thread-safe, you should basically use a lock object to lock on, and wrap the contents of your methods with a lock statement:
private readonly object syncObject = new object();
void SomeMethod()
{
lock (this.syncObject)
{
// Work with your list here
}
}
Beware that those events being raised have the potential to hold the lock for an extended period of time, depending on what the delegates do.
You could omit the lock from the NoOfOnlineUsers property while declaring your list as volatile. However, if you want the Count value to persist for as long as you are using it at a certain moment, use a lock there, as well.
As others suggested here, exposing your list directly, even with a lock, will still pose a "threat" on it's contents. I would go with returning a copy (and that should fit most purposes) as Mark Gravell advised.
Now, since you said you are using this in an ASP.NET environment, it is worth saying that all local and member variables, as well as their member variables, if any, are thread safe.
Just looking for a code review of this code. ASP.net Cache is not an option. The static list will be accessed a lot on a website that gets well over 10K page views per day and concurrent read attempts is likely. On app restart when the list is rebuilt I was wondering if there are any issues I may be overlooking? Is locking on the list being instantiated best practice?
public class MyClass
{
private static List<Entry> _listCache = null;
protected static List<Entry> ListCache
{
get
{
if (_listCache == null)
{
_listCache = new List<Entry>();
lock (_listCache)
{
//Add items to the list _listCache from XML file
}
}
return _listCache;
}
}
//....Other methods that work with the list
}
10k views - that's one every 8 seconds... not sure you need to worry too much... ;-p
But re the code - that is overcomplicating things, and you could still end up initializing it twice. I'd just use a static constructor to do this; it'll be more robust. If you must have full isolated lazy loading (even with other static methods on the type), there is a trick with an inner class to achieve the same:
public class MyClass
{
static class InnerCache {
internal static readonly IList<Entry> _listCache;
static InnerCache() {
List<Entry> tmp = new List<Entry>();
//Add items to the list _listCache from XML file
_listCache = new ReadOnlyCollection<Entry>(tmp);
}
}
protected static IList<Entry> ListCache {
get {return InnerCache._listCache;}
}
}
I would also be concerned about the chance of somebody mutating the list - might want to use a readonly list!
There's not really a reason this wouldn't work for you. However, if you want to do it the way your sample code is, you want to lock before you check to see if _listCache is null. So you would need a separate monitor to lock on. Something like this:
public class MyClass
{
private static object _listCacheMonitor = new object();
private static List<Entry> _listCache = null;
protected static List<Entry> ListCache
{
get
{
lock (_listCacheMonitor) {
if (_listCache == null)
{
_listCache = new List<Entry>();
//Add items to the list _listCache from XML file
}
}
return _listCache;
}
}
//....Other methods that work with the list
}
A static constructor may be your best bet here. A static constructor will block all threads that depend on it while it's running, and it will only run once. As you have the code here, the lock doesn't really do anything, and there are lots of ways that bad things can happen, including multiple Lists being initialized from XML at the same time. In fact, one thread could create a new List then lock and load a different list and then return a third list, depending on when the thread switching occurs.
Multiple threads can initalize _listCache. Depending on the code generation optimizations and runtime execution optimization, this may result in multiple threads locking and updating different objects. And besides, you can't expose the list as a property allowing anyone to adds/remove object w/o a lock.
You'd be better using an immutable list that multiple readers can safely parse in read-only mode. Alternatively you can use a Read-Write lock, but things will get pretty hairy, between the initialization control and the access r-w control.