I am playing with a standard optimistic concurrency control scenario with extended session / automatic versioning. I have an entity which I load in the first transaction, present to user for modification and save in the second one, both transactions sharing the same session. After the entity is somehow modified session.flush() at the end of the second transaction may throw a StaleObjectStateException in case a version inconsistency is detected meaning that a concurrent transaction has saved a next version of the entity in between.
I want to handle such an error in a most simple way -- just to reload entity losing current changes and continue with editing and saving again. First I tried this:
session.refresh(entity);
but after I modify and attempt to save this refreshed entity, I still get the same StaleObjectStateException, even though it does get refreshed and the version number appears consistent; yes I know that using refresh() in extended sessions is discouraged, but don't understand why. Is this behavior related to the reason it is discouraged?
Next I tried the following way to avoid using session.refresh():
session.evict(entity);
entity = session.load(MyEntity.class, id);
but it still results in StaleObjectStateException being raised at saving the entity which is not indeed stale.
The only way I managed to cope with the exception is this:
session.clear();
entity = session.load(MyEntity.class, id);
but isn't session.clear() the same as session.evict() pertaining to my concrete entity?
To resume, my questions are:
Why is StaleObjectStateException still thrown on a reloaded entity unless session.clear() is done?
What is the correct way to reload an entity which has already been loaded in the same session and why is refresh() bad? Is there something wrong with this approach to implement conversation?
I'm using Hibernate 4.1.7.Final, with no second-level cache.
My apologies if my question is repeating, but I fail to find a profound explanation...
When you get an exception in a session, than that session instance is broken. You cannot use that instance any more and you have to throw it away and create a new instance. The exception is not reset (as you can see you get the same exception again thought logically this should not happen). This is a general rule for using hibernate sessions. The reason for this is, hibernate does not always see why an exception appears and the state of the session instance may be inconsistent.
I don't know why it works after clear(). This may be accidentally. It is more prudent to use a new instance.
If you use a StatelessSession, then you don't have this restriction, but stateless sessions have other disadvantages, for example no caching.
Related
I was getting an error like this
a different object with the same identifier value was already associated with the session:
I've searched and found that it could be fixed with CascadeType.MERGE or refactoring a lot of code to prevent that a same database object becomes two instances within the session.
I can't refactor it.
CascadeType.MERGE worked, but that means I would have to code a lot to resolve remove problems, since it was .ALL before, right?
I got it working putting
#Transactional(propagation = Propagation.REQUIRES_NEW)
above a method, of a class annoted with #Service, that query database, which was the one that was throwing the exception I mentioned.
I need help to understand if this new annotated method can bring me any future headache like it is now.
It is being called from some cron jobs beside the action I'm fixing.
In fact you create a separate transaction for the method call by annotating the method with #Transactional(propagation = Propagation.REQUIRES_NEW)
It means in case of an Exception thrown out of the method all DB changes (saving etc.) are applied and are not rolled back. This could significantly damage the business logic and could be source of inconsistent data in DB.
I would reconsider applying the Propagation.REQUIRES_NEW.
Merge sounds much more suitable in this case.
None of the solution you listed are acceptable IMHO.
Deferring some part of treatment to a new transaction will break the atomicity (all or nothing) of your unit of work, and changing the cascading type will imply that you manually handle all operation automatically cascaded before.
The right approach is to understand why hibernate encounters 2 different object instances with the same identifier, the most common causes is because you manually persists (save) a detached / transient object with a fixed identifier while this one already exists in the session (a managed object with the same identifier is already in the session).
You could try to manually re-attach (merge / update /saveOrUpdate) the detached object instance causing the problem.
You have to be aware of the entity lifecycle to properly understand what happens here.
This seems like it would come up often, but I've Googled to no avail.
Suppose you have a Hibernate entity User. You have one User in your DB with id 1.
You have two threads running, A and B. They do the following:
A gets user 1 and closes its Session
B gets user 1 and deletes it
A changes a field on user 1
A gets a new Session and merges user 1
All my testing indicates that the merge attempts to find user 1 in the DB (it can't, obviously), so it inserts a new user with id 2.
My expectation, on the other hand, would be that Hibernate would see that the user being merged was not new (because it has an ID). It would try to find the user in the DB, which would fail, so it would not attempt an insert or an update. Ideally it would throw some kind of concurrency exception.
Note that I am using optimistic locking through #Version, and that does not help matters.
So, questions:
Is my observed Hibernate behaviour the intended behaviour?
If so, is it the same behaviour when calling merge on a JPA EntityManager instead of a Hibernate Session?
If the answer to 2. is yes, why is nobody complaining about it?
Please see the text from hibernate documentation below.
Copy the state of the given object onto the persistent object with the same identifier. If there is no persistent instance currently associated with the session, it will be loaded. Return the persistent instance. If the given instance is unsaved, save a copy of and return it as a newly persistent instance.
It clearly stated that copy the state(data) of object in database. if object is not there then save a copy of that data. When we say save a copy hibernate always create a record with new identifier.
Hibernate merge function works something like as follows.
It checks the status(attached or detached to the session) of entity and found it detached.
Then it tries to load the entity with identifier but not found in database.
As entity is not found then it treat that entity as transient.
Transient entity always create a new database record with new identifier.
Locking is always applied to attached entities. If entity is detached then hibernate will always load it and version value gets updated.
Locking is used to control concurrency problems. It is not the concurrency issue.
I've been looking at JSR-220, from which Session#merge claims to get its semantics. The JSR is sadly ambiguous, I have found.
It does say:
Optimistic locking is a technique that is used to insure that updates
to the database data corresponding to the state of an entity are made
only when no intervening transaction has updated that data since the
entity state was read.
If you take "updates" to include general mutation of the database data, including deletes, and not just a SQL UPDATE, which I do, I think you can make an argument that the observed behaviour is not compliant with optimistic locking.
Many people agree, given the comments on my question and the subsequent discovery of this bug.
From a purely practical point of view, the behaviour, compliant or not, could lead to quite a few bugs, because it is contrary to many developers' expectations. There does not seem to be an easy fix for it. In fact, Spring Data JPA seems to ignore this issue completely by blindly using EM#merge. Maybe other JPA providers handle this differently, but with Hibernate this could cause issues.
I'm actually working around this by using Session#update currently. It's really ugly, and requires code to handle the case when you try to update an entity that is detached, and there's a managed copy of it already. But, it won't lead to spurious inserts either.
1.Is my observed Hibernate behaviour the intended behaviour?
The behavior is correct. You just trying to do operations that are not protected against concurrent data modification :) If you have to split the operation into two sessions. Just find the object for update again and check if it is still there, throw exception if not. If there is one then lock it by using em.(class, primary key, LockModeType); or using #Version or #Entity(optimisticLock=OptimisticLockType.ALL/DIRTY/VERSION) to protect the object till the end of the transaction.
2.If so, is it the same behaviour when calling merge on a JPA EntityManager instead of a Hibernate Session?
Probably: yes
3.If the answer to 2. is yes, why is nobody complaining about it?
Because if you protect your operations using pessimistic or optimistic locking the problem will disappear:)
The problem you are trying to solve is called: Non-repeatable read
Background
I have a java/spring system where transactions are managed manually via a custom HandlerInterceptor. That is to say:
at the begining of every request a transaction is opened (an unfortunate part of the system is that any request might result in a write to the db)
an EntityManager instance joins the transaction
the entity manager is used to load entities which are modified. The EntityManager tracks all changes
at the end of every request the EntityManager is flushed and committed
Yes this is not ideal, but I did not create this system and it's simple enough to allow us to work within it's confines - I'm not looking to change it without good reason.
I am not used to commit-all-tracked-entities-on-flush behavior and so have been doing something like:
//change entity
if(ovalValidator.isValid(entity))
em.persist(entity);
I need to fix this to work with my new understanding and switching the above to this seems to work:
//change entity
if(!ovalValidator.isValid(entity))
em.detach(entity);
My question
It is my understanding that this just removes the entity from the flush queue even if it IS marked as dirty. Is this correct? Is there a better way to achieve what I am trying to (don't save changes to that entity)? Is there anything I need to look out for if I'm doing this?
detache removes the entity from the session (changeTracking, lazyloading, ...) it does what you want. You could also implement en interceptor removing the dirty mark of the invalid entities but i think your solution would work as well
In my web application I have several threads that potentially access the same data concurrently why I decided to implement optimistic (versioning) and pessimistic locking with Hibernate.
Currently I use the following pattern to lock an entity and perform write operations on it (using Springs Transaction manager and transaction demarcation with #Transactional):
#Transactional
public void doSomething(entity) {
session.lock(entity, LockMode.UPGRADE);
session.refresh(entity);
// I change the entity itself as well as entites in a relationship.
entity.setBar(...);
for(Child childEntity : entity.getChildren()) {
childEntity.setFoo(...);
}
}
However, sometimes I am getting StaleObjectException when the #Transactional is flushing that tells me that a ChildEntity has been modifed concurrently and now has a wrong version.
I guess I am not correctly refreshing entity and its children so I am working with stale data. Can someone point out how to achieve this? Some thoughts of me included clearing the persistence context (the session) or calling session.lock(entity, LockMode.READ) again, but I am not sure what is correct here.
Thanks for your help!
You may want to take at look at this Hibernate-Issue: LockMode.Upgrade doesn't refresh entity values.
In short: Hibernat does NOT perform a select after a successful lock if the given entity was already preloaded. You need to call refresh for the entity for yourself after you received the lock.
Why do you make "LockMode.UPGRADE" and optimistic locking live together? Seem like controversial things.
Hibernate never lock objects in memory and always use the locking mechanism of the database. Also, "if the requested lock mode is not supported by the database, Hibernate uses an appropriate alternate mode instead of throwing an exception. This ensures that applications are portable.". It means, that if your database doesn't support SELECT ... FOR UPDATE, most probably, you will get these exceptions.
Another possible reason is that you haven't used "org.hibernate.annotations.CascadeType.LOCK" for children.
We are using Hibernate Spring MVC with OpenSessionInView filter.
Here is a problem we are running into (pseudo code)
transaction 1
load object foo
transaction 1 end
update foo's properties (not calling session.save or session.update but only foo's setters)
validate foo (using hibernate validator)
if validation fails ?
go back to edit screen
transaction 2 (read only)
load form backing objects from db
transaction 2 end
go to view
else
transaction 3
session.update(foo)
transaction 3 end
the problem we have is if the validation fails
foo is marked "dirty" in the hibernate session (since we use OpenSessionInView we only have one session throughout the http request), when we load the form backing objects (like a list of some entities using an HQL query), hibernate before performing the query checks if there are dirty objects in the session, it sees that foo is and flushes it, when transaction 2 is committed the updates are written to the database.
The problem is that even though it is a read only transaction and even though foo wasn't updated in transaction 2 hibernate doesn't have knowledge of which object was updated in which transaction and doesn't flush only objects from that transaction.
Any suggestions? did somebody ran into similar problem before
Update: this post sheds some more light on the problem: http://brian.pontarelli.com/2007/04/03/hibernate-pitfalls-part-2/
You can run a get on foo to put it into the hibernate session, and then replace it with the object you created elsewhere. But for this to work, you have to know all the ids for your objects so that the ids will look correct to Hibernate.
There are a couple of options here. First is that you don't actually need transaction 2 since the session is open you could just load the backing objects from the db, thus avoiding the dirty check on the session. The other option is to evict foo from the session after it is retrieved and later use session.merge() to reattach it when you what your changes to be stored.
With hibernate it is important to understand what exactly is going on under the covers. At every commit boundary it will attempt to flush all changes to objects in the current session regardless of whether or not the changes where made in the current transaction or any transaction at all for that matter. This is way you don't actually need to call session.update() for any object that is already in the session.
Hope this helps
There is a design issue here. Do you think an ORM is a transparent abstraction of your datastore, or do you think it's a set of data manipulation libraries? I would say that Hibernate is the former. Its whole reason for existing is to remove the distinction between your in-memory object state and your database state. It does provide low-level mechanisms to allow you to pry the two apart and deal with them separately, but by doing so you're removing a lot of Hibernate's value.
So very simply - Hibernate = your database. If you don't want something persisted, don't change your persistent objects.
Validate your data before you update your domain objects. By all means validate domain objects as well, but that's a last line of defense. If you do get a validation error on a persistent object, don't swallow the exception. Unless you prevent it, Hibernate will do the right thing, which is to close the session there and then.
What about using Session.clear() and/or Session.evict()?
What about setting singleSession=false on the filter? That might put your operations into separate sessions so you don't have to deal with the 1st level cache issues. Otherwise you will probably want to detach/attach your objects manually as the user above suggests. You could also change the FlushMode on your Session if you don't want things being flushed automatically (FlushMode.MANUAL).
Implement a service layer, take a look at spring's #Transactional annotation, and mark your methods as #Transactional(readOnly=true) where applicable.
Your flush mode is probably set to auto, which means you don't really have control of when a DB commit happens.
You could also set your flush mode to manual, and your services/repos will only try to synchronize the db with your app when you tell them to.