- Upgrading from 5.0 to 6.0
- Detailed release notes
- Feature199869Spring Boot 3 Starter
- Feature197500Support JakartaEE libraries and packages
- Feature199608Hibernate 6 support
- Feature85601JMS 2.0 support
- Feature20711Support for TransactionSynchronizationRegistry
- Feature199530Make the transaction template thread-safe
- Feature192016Better support for borrowConnection timeouts in the pool
- Feature189885Refactor shutdown: improve waiting for in-flight transactions
- Feature182107Remove user / password properties from MessageDrivenContainer
- Feature199528Drop reaping functionality from the pools
- Bug197505Avoid that defaults in the Spring Boot starter override jta.properties
- Bug197362Clarify message if rest port not set in client
- Bug197506Support diamond case architectures for readOnly remoting
- Bug200555Also close Callable- and PreparedStatements when JDBC connection is closed
- Bug197253Suspend/resume should not change local sibling count
- Bug197240Allow spring.jta properties in Spring Boot starter
- Bug200925Improve handling of null xaresource
- Bug201164Avoid ConcurrentModificationException during trace logging in the pool
- Issue201646Spring Boot 2 starter: also add jta and jms versions of Spring Boot's POM to the starter project
-
- Bug181871Avoid that ConnectionPoolWithConcurrentValidation grows beyond maxPoolSize
- About Severity
- Available to customers only. Want to become a customer?
Upgrading from 5.0 to 6.0
This release was very hard to do. Not because we wanted many new features included (there are not that many, actually), but rather because it was a balancing act between stability / backwards compatibility and new "breaking change" emerging platforms (specifically: Spring Boot 3, Hibernate 6 and their dependencies on JakartaEE and very recent Java versions).
We think we did a good job: just like the previous release 5.0, this new release can run on "good old" Java 8 for most modules, except for Spring Boot 3 integration (which requires Java 17 as per Spring Boot) and Hibernate 6. So if your application worked with release 5.0 then it should still work, provided that you do the following:
Add extra dependencies explicitly to your POM
We now support both javax and jakarta namespaces for the Java enterprise APIs, by offering regular jars as well as "jakarta" jars (with the corresponding classifier in the jar names). These jakarta jars are generated at build time with the Eclipse transformer utility.In order to do this, we had to break the transitive dependency mechanism (which, incidentally, also avoids pulling in vulnerable 3rd party code libs). So: except for Spring Boot 3 apps, this means that you will now have to add the following dependency to your pom file (or you risk facing ClassNotFoundExceptions):
The traditional javax style
<dependency>
<groupId>com.atomikos</groupId>
<artifactId>transactions-jta</artifactId>
<version>6.0.109</version> <!-- Commercial, for open source use version 6.0.0 -->
</dependency>
<dependency>
<groupId>jakarta.jms</groupId>
<artifactId>jakarta.jms-api</artifactId>
<!-- NOTE: despite "jakarta" in the name, this version is still a javax jar -->
<version>2.0.3</version>
</dependency>
<dependency>
<groupId>javax.transaction</groupId>
<artifactId>jta</artifactId>
<version>1.1</version>
</dependency>
The new jakarta style
<dependency>
<groupId>com.atomikos</groupId>
<artifactId>transactions-jta</artifactId>
<version>6.0.109</version> <!-- Commercial, for open source use version 6.0.0 -->
<classifier>jakarta</classifier>
</dependency>
<dependency>
<groupId>jakarta.jms</groupId>
<artifactId>jakarta.jms-api</artifactId>
<version>3.1.0</version>
</dependency>
<dependency>
<groupId>jakarta.transaction</groupId>
<artifactId>jakarta.transaction-api</artifactId>
<version>2.0.1</version>
</dependency>
Override the new JMS behaviour to revert to prior release behaviour
We now support JMS 2. This required some changes to the JMS session creation behaviour, so if you rely on the "old" JMS behaviour then set the following property on theAtomikosConnectionFactoryBean:
AtomikosConnectionFactoryBean cf = new AtomikosConnectionFactoryBean();
// 1 sets behaviour like in releases 3.9-5.0, 0 sets behaviour like in releases pre-3.9
cf.setSessionCreationMode(1);
For more details, see the javadoc of the AtomikosConnectionFactoryBean class.
Drop any reapTimeout property
We've removed reap functionality from the pool, so remove any reapTimeout property in your datasource or connection factory config.
Detailed release notes
If you like the nitty gritty details, then the following sections are for you!
Feature199869 Spring Boot 3 Starter
We added a starter for Spring Boot 3.
Technical details
Spring Boot 3 has made a "big bang" upgrade to Jakarta EE and Java 17, with all complications involved. You can now use Atomikos with Spring Boot 3 by means of our new Spring Boot 3 starter module.
Changes impacting client API
Spring Boot 3 requires Java 17 or higher, plus all JakartaEE libraries (and for course Atomikos releases 6.0 or higher). Note that this means you will also need JakartaEE JMS drivers, and JakartaEE persistence providers. Hurray.
You need the following dependency to use Spring Boot 3:
<dependency>
<groupId>com.atomikos</groupId>
<artifactId>transactions-spring-boot3-starter</artifactId>
<version>6.0.109</version>
</dependency>
For the JakartaEE JMS and JTA APIs: we've added those to the Spring Boot 3 starter module already - so you don't have to bother.
Feature197500 Support JakartaEE libraries and packages
Description
We now also support JakartaEE (in addition to the "old" JEE with javax namespaces).Technical details
All relevant modules are now available in both a traditional variant and a JakartaEE variant, the latter generated with the Eclipse transformer utility during the build. JakartaEE jars carry the same name as their traditional counterparts, but have the additional classifier "jakarta" in the jar file name to distinguish them.The choice for this strategy was heavily driven by our desire to still support Java 8 applications that can't upgrade to Java 17 or JakartaEE right now.
The previous statement probably deserves some explanation: after some initial attempts it was painfully clear to us that upgrading from Java 8 to Java 17 is nothing short of a nightmare.
[Oh and by the way, we figure that this is the reason why extended support for Java 8 is available until 2030, the longest of all Java versions at the time of publication. Yes, that means Java 8 should be supported for longer than Java 17 - go figure that!] Our chosen approach should allow existing installations to upgrade to our 6.0 release without the need to upgrade to JakartaEE and/or Java 17 at the same time.Past experience with our customers has shown that even a "simple" upgrade from Java 6 to Java 8 can be a huge roadblock - especially if the development team is gone and only operations teams are left. So we prefer to keep upgrading a seamless and straightforward process, without requiring needless Java updates at the same time. We are sure that you will appreciate that.
The "downside" of this choice is that you can no longer count on transitive dependencies as much as you did, because many modules now have 2 variants available (and hence the transitive dependencies would not necessarily match the right variant). While this may seem a disadvantage, it actually increases security because there are no longer transitive dependencies on 3rd party jars either (so you avoid pulling in vulnerabilities).
So: you have to add some dependencies manually that were transitive before. But the effort required is significantly lower than the effort it would take to upgrade your code base towards Java 17 and JakartaEE all at once.
Changes impacting client API
As explained in the upgrade section above: you need to declare some extra dependencies. Also, you have to be careful to use the right combination of Jakarta-enabled jars in your application. To this end, we have defined BOM files that list all the jars designed to work together:
- transactions-essentials versus transactions-essentials-jakarta
- extreme-transactions versus extreme-transactions-jakarta
You can check these BOMs for inspiration on which jars to use / combine in your application. If you are curious where to find them: check groupId=com.atomikos and artifactId=name_of_BOM_file.
Feature199608 Hibernate 6 support
Description
You can now easily use Hibernate 6 in combination with our product!
Technical details
Hibernate 6 is also in the JakartaEE camp, so of course you need JakartaEE support to be able to use it. Well, as we've explained above: we have done just that.
To use Hibernate 6, it suffices to use transactions-hibernate4 in the jakarta flavour. For a working sample: check the supplied examples project called examples-hibernate6 or see the summary POM snippet below.Changes impacting client API
No real changes are needed, but you do have to add the correct POM dependencies:
<dependency>
<groupId>com.atomikos</groupId>
<artifactId>transactions-hibernate4</artifactId>
<version>6.0.109</version>
<classifier>jakarta</classifier>
</dependency>
<dependency>
<groupId>com.atomikos</groupId>
<artifactId>transactions-jdbc</artifactId>
<version>6.0.109</version>
<classifier>jakarta</classifier>
</dependency>
<dependency>
<groupId>com.atomikos</groupId>
<artifactId>transactions-jta</artifactId>
<version>6.0.109</version>
<classifier>jakarta</classifier>
</dependency>
<dependency>
<groupId>jakarta.transaction</groupId>
<artifactId>jakarta.transaction-api</artifactId>
<version>2.0.1</version>
</dependency>
<dependency>
<groupId>jakarta.persistence</groupId>
<artifactId>jakarta.persistence-api</artifactId>
<version>3.1.0</version>
</dependency>
Of course, you also need the Hibernate 6 dependencies themselves - but we'll leave that up to you…
Feature85601 JMS 2.0 support
Description
We now have (minimal) support for JMS 2.0.
Technical details
As required by JakartaEE and Spring Boot 3, we needed some support for JMS 2.0. For the sake of releasing on time, this support is only partial: we do not yet support the newJMSContext API. Spring Boot 3 does not seem to need it, so that should be fine for now. Attempting to call any of the createContext methods on the AtomikosConnectionFactoryBean will throw UnsupportedOperationException.
Changes impacting client API
Session creation behaviour has changed due to the clarification of session creation behaviour in JMS 2.0 (which is incompatible with what we had in prior releases). As you can see in the table below, the differences are in the combined interpretation of:
- a JTA transaction being present on the calling thread or not,
- the value set for
localTransactionModeand - the value of the sessionTransacted flag when creating a session.
Prior to JMS 2.0, session creation had some fuzzy border cases where the interpretation was left to the party implementing the specs (like Atomikos). In JMS 2.0 this has been addressed. Needless to say, the probably of a perfect match with our interpretation was next to zero - so we had to make some changes.
To preserve compatibility your instances ofAtomikosConnectionFactoryBean now offer an extra method setSessionCreationMode(int mode) which changes behaviour as described below. The default is value 2 (JMS 2.0) so for compatibility with 5.0, try value 1. For compatibility with pre-3.9 installations, try value 0.
- SessionCreationMode.JMS_2_0 (the default as of release 6.0), resolving to constant value 2
-
existing JTA transaction for thread localTransactionMode resulting session true ignored XA session false false XA session false true non-XA session according to sessionTransacted/acknowledgeMode parameters - SessionCreationMode.PRE_6_0 (optional, for backward compatibility) - resolving to constant value 1
-
localTransactionMode sessionTransactedFlag resulting session false ignored XA session true ignored non-XA session according to sessionTransacted/acknowledgeMode parameters - SessionCreationMode.PRE_3_9 (optional, for backward compatibility and equivalent to ignoreSessionTransactedFlag = false) - resolving to constant value 0
-
localTransactionMode sessionTransactedFlag resulting session false true XA session other other non-XA session according to sessionTransacted/acknowledgeMode parameters
AtomikosConnectionFactoryBean.
Feature20711 Support for TransactionSynchronizationRegistry
Description
We now offer an implementation ofTransactionSynchronizationRegistry defined in the more recent JTA specifications.
Technical details
You can use an instance ofcom.atomikos.icatch.jta.TransactionSynchronizationRegistryImp contained in module transactions-jta.jar.
Unless you are a vendor of persistence providers, you probably won't ever need this class so we'll keep the explanation to a minimum - since vendors of persistence providers already know all there is to know.
Changes impacting client API
An extra class available in our distribution, for you to use if you want to.
Feature199530 Make the transaction template thread-safe
Description
Instances ofcom.atomikos.icatch.jta.template.TransactionTemplate are now tread-safe.
Technical details
The first implementations of our template were not intended for threaded use cases. Based on customer feedback, we have changed that - so you can now reuse the same instance in different threads - just like Spring's template.
Changes impacting client API
Instead of having to create multiple template instances, you can now reuse the same instance in different places of your code base. This should simplify your configuration and code base.
Feature192016 Better support for borrowConnection timeouts in the pool
Description
We've improved the pool logic so that borrow requests can better respect the timeout.
Technical details
When a connection is requested and none is available, new connections would be created in the application's thread. In case of network issues, this would block the application's thread, possibly for much longer than theborrowConnectionTimeout setting would (and should) allow. That's because there is no easy way to interrupt a blocked IO request.
We now improved this as follows:
- There is a separate background thread that grows the pool when needed.
- The application's thread waits for this thread to grow the pool, but no longer than specified by
borrowConnectionTimeout.
Changes impacting client API
No real changes are needed, except that it should work better.
Feature189885 Refactor shutdown: improve waiting for in-flight transactions
Description
We have simplified and improved the shutdown logic.
Technical details
Shutdown used to wait for a timeout bound by the value ofcom.atomikos.icatch.default_max_wait_time_on_shutdown. Recovery itself is already bound by a different parameter: com.atomikos.icatch.default_max_timeout. The previous logic used to mix those two and the results were not always clear.
We have simplified this to the following:
- Shutdown waits for all active transactions to finish in the JVM.
- Then it performs a recovery pass to clean up as much as possible.
- If both previous steps worked fine then there are no pending transactions and the transaction log files can be deleted.
- On the other hand, if either step has issues then the transaction log files have to be kept (we log a warning for that).
Changes impacting client API
You no longer needcom.atomikos.icatch.default_max_wait_time_on_shutdown in your jta.properties file. Beware that shutdown duration is bound by the value of com.atomikos.icatch.default_max_timeout - unless you use the "forceShutdown" option.
Feature182107 Remove user / password properties from MessageDrivenContainer
Description
The user and password properties have been removed fromcom.atomikos.jms.extra.MessageDrivenContainer because they were never used anyway.
Technical details
Instances of this class delegate tocom.atomikos.jms.AtomikosConnectionFactoryBean to create connections. The latter is also configured with a user and password, and those are the values that are actually being used.
As a result, keeping two unused properties in the com.atomikos.jms.extra.MessageDrivenContainer class was confusing - so we have removed those.
Changes impacting client API
You need to remove any references to these properties from your configuration.
Feature199528 Drop reaping functionality from the pools
Description
We've removed the reaping functionality from the pools.
Technical details
Reaping was when our pools would take away connections that were being held onto for too long.
Historically, reaping functionality has caused more problems / confusion than it solved. Moreover, it was intended to fix application-level connection leaks, meaning bugs in the application (as opposed to bugs in our code base). We felt that became a source of needless complexity.
We believe that the correct way of fixing application-level connection leaks is by fixing the application, rather than abruptly reaping connections away from that application. So our 6.0 release seemed a good time to remove that feature.
Changes impacting client API
Remove any references toreapTimeout from your configuration.
Bug197505 Avoid that defaults in the Spring Boot starter override jta.properties
| Severity: | 4 |
|---|---|
| Affected version(s): | 5.0.x |
Description
You can now count on the property values from jta.properties to be taken into account even with Spring Boot.
Technical details
The Atomikos JTA properties implementation in our Spring Boot starter would define default values for many properties, meaning that their value specified jta.properties would not be taken into account. This has now been fixed.
Changes impacting client API
Your properties specified in jta.properties should now work.
Bug197362 Clarify message if rest port not set in client
| Severity: | 1/2/3/4 |
|---|---|
| Affected version(s): | 5.0.x |
Description
You can now count on a better error message when theAtomikosRestPort URL is not set (for transactions across remoting calls).
Technical details
When theAtomikosRestPort URL was not set, the client template would report a misleading message saying that there is no transaction for the thread. Instead, the root cause is a URL that is missing - so we fixed that for you.
Changes impacting client API
None.
Bug197506 Support diamond case architectures for readOnly remoting
| Severity: | 4 |
|---|---|
| Affected version(s): | 5.0.x |
Description
You can now use transitive readOnly remoting transactions in all cases.
Technical details
As outlined in this GitHub issue there was a problem with readOnly invocations when:- the same shared service was called over different paths, and
- all invocations were readOnly
This is known as a "diamond case" because the invocation diagram looks like a diamond.
This issue has been fixed in the following way: our product will now avoid the readOnly optimisation in this specific scenario. This is still correct, at a minor performance overhead in the exotic cases where this does happen.
Changes impacting client API
None.
Bug200555 Also close Callable- and PreparedStatements when JDBC connection is closed
| Severity: | 4 |
|---|---|
| Affected version(s): | 5.0.x |
Description
Now you can close a JDBC connection and also have the Callable- and PreparedStatements closed automatically.
Technical details
When a JDBC connection is closed by the application (returned to the pool) then the JDBC specification requires all pending statements to be closed as well.
We supported this, but apparently this was not done for all statement types. We fixed this now.
This issue is marked as severity 4 (development use) because we had no real bug reports from any customers. It was something we noticed during a code review.
Changes impacting client API
None.
Bug197253 Suspend/resume should not change local sibling count
| Severity: | 2 |
|---|---|
| Affected version(s): | 5.0.x |
Description
You can now use @RequiresNew with imported transactions.
Technical details
As explained in this GitHub issue, there were problems when a remote transaction was imported and then subsequently called local logic that was marked with @RequiresNew. As required by the specification of @RequiresNew, this would suspend any active transaction - and resume it later.Our code had a side effect of suspend/resume that changed the local sibling count.
Sibling counts help detect orphaned invocations (and their transactional updates to persistent storage) that arise out of lost replies. For instance, consider this scenario with services A and B:
- A starts a transaction.
- A calls B as part of that transaction.
- B does work and returns a result.
- A does not receive the result due to a network timeout (so B now has an orphaned invocation).
- A tries again, so B performs the changes again and returns a new result (in the same transaction).
- A commits the transaction thinking it only updated B once.
- B commits the same transaction with two sets of updates.
This risk here is the different views of A and B regarding the scope of the transaction: A thinks it commits one update to B, whereas B commits two different updates. This can be a problem for data consistency, so we avoid this by keeping sibling counts at B and A. A constructs its sibling count picture with each result it actually receives with its replies from A. Before commit, A passes on the "count" it has for invocations at B, and if B finds that there is no match then it refuses to commit.
This would avoid the problem outlined above, because in step 4 service A will miss a count, so in step 6 service A will pass a count of 1 for service B, whereas B will see 2 and refuses the commit process.
In short, sibling counts have their purpose. However due to a bug, this was affected by a suspend/resume at service B (when it has @RequiresNew logic inside).
Changes impacting client API
You should now be able to configure @RequiresNew on any service that needs it.
Bug197240 Allow spring.jta properties in Spring Boot starter
| Severity: | 4 |
|---|---|
| Affected version(s): | 5.0.x |
Description
Any spring.jta properties in Spring Boot should now be taken into account.
Technical details
After the Spring Boot team contributed the source code for their Atomikos starter we thought it was safer to ignore any properties with the "legacy" prefix spring.jta.atomikos.properties (in your application.properties file for Spring Boot).This has proven to confuse users, so we now take such spring.jta properties into account.
Changes impacting client API
Any spring.jta properties in Spring Boot should now be taken into account.
Bug200925 Improve handling of null xaresource
| Severity: | 3 |
|---|---|
| Affected version(s): | 3.9.x, 4.0.x, 5.0.x |
Description
We've refined how we deal with lost XA connections during the commit phase.
Technical details
Some connection issues would lead to needless heuristic exceptions during commit, in particular when the commit was 1-phase (where failure to reach the backend results in resource-internal rollback anyway).We have now refined this so there are no needless heuristic exceptions any more in these cases. At the same time, we have also tried to minimise the number of cases where such connection issues remain.
Changes impacting client API
None.
Bug201164 Avoid ConcurrentModificationException during trace logging in the pool
| Severity: | 4 |
|---|---|
| Affected version(s): | 5.0.x |
Description
You can now use our pools with TRACE logging and get less exceptions.
Technical details
The issue details are described in this GitHub issue.Changes impacting client API
None.
Issue201646 Spring Boot 2 starter: also add jta and jms versions of Spring Boot's POM to the starter project
| Severity: | 4 |
|---|---|
| Affected version(s): | 6.0.x |
Description
Unlike Spring Boot 3 support, we did not yet add the JMS and JTA dependencies in the Spring Boot 2 starter.
Technical details
We still have to add the JMS and JTA dependencies in the Spring Boot 2 starter. This is needed because of the removal of transitive dependencies. Until then, you will have to add them yourself.
Free Download Bug181871 Avoid that ConnectionPoolWithConcurrentValidation grows beyond maxPoolSize
| Severity: | 2 |
|---|---|
| Affected version(s): | 5.0.x |
Description
We've ported a fix from our commercial release that prevents the pool from exceeding its maxPoolSize.
Technical details
The concurrent pool does not enforce synchronization so when it:
- checks if the pool can grow and then
- actually grows it
… a different thread may have done the same and the pool grows too much.
Solution: we've made growPool (with existing synchronized block) check maxSize again, and don't grow if already at the maximum.
Changes impacting client API
None.

Add a comment