NPE in ConnectorXAResource.getResourceHandle causes connection pool problems #3025
Comments
|
I tested changing the code to: I replaced the /glassfish/modules/connectors-runtime.jar and my remote bean calls keep working / I cannot reproduce the problem anymore. This might not be the correct fix, but it avoids the NullPointerException and remote bean calls keep working. |
|
I will try to create a test case to reproduce this. I cannot provide the current code. This could take a while. |
|
@escay I was the poster of #2310. I actually have seen this issue go away since using more default connection pool settings, as what was recommended in my ticket. I'm not sure if it will help you, but I just wanted to add a data point. Here is my current connection pool setting (for Oracle): <jdbc-connection-pool validation-table-name="TABLEVALIDATION"
connection-validation-method="custom-validation"
allow-non-component-callers="true"
statement-cache-size="200"
associate-with-thread="false"
statement-timeout-in-seconds="-1"
idle-timeout-in-seconds="0"
lazy-connection-association="false"
max-wait-time-in-millis="0"
connection-creation-retry-attempts="5"
validate-atmost-once-period-in-seconds="30"
lazy-connection-enlistment="false"
max-pool-size="96"
res-type="javax.sql.DataSource"
name="MyPool"
is-connection-validation-required="true"
validation-classname="org.glassfish.api.jdbc.validation.OracleConnectionValidation"
datasource-classname="oracle.jdbc.xa.client.OracleXADataSource">
<property name="DataSourceName" value="OracleDataSource"/>
<property name="ImplicitCachingEnabled" value="false"/>
<property name="NetworkProtocol" value="tcp"/>
<property name="DatabaseName" value="transitem"/>
<property name="LoginTimeout" value="0"/>
<property name="User" value="myuser"/>
<property name="Password" value="mypass"/>
<property name="URL" value="jdbc:oracle:thin:@localhost:1521:ORCL"/>
<property name="PortNumber" value="1521"/>
<property name="ExplicitCachingEnabled" value="false"/>
<property name="dynamic-reconfiguration-wait-timeout-in-seconds" value="960"/>
<property name="MaxStatements" value="0"/>
</jdbc-connection-pool> |
|
@marcreichman Thank you for the information. In the meantime I have created a reproduction scenario and created a support ticket at Payara directly using a support contract. They are able to reproduce the problem. I hope the cause of the problem can be found. |
|
I believe this is fixed in 5.184 |
|
See #3344 for the commits |
Description
I run very often into similar issues as described in #769, especially the situation described in issue #2310 seems to be the same as the one in my server.log file:
Reproduction is very difficult but I can reproduce it often when:
Expected Outcome
No NPEs.
Remote bean calls that keep working.
Current Outcome
NPEs and the exception:
Steps to reproduce (Only for bug reports)
I debugged Payara and added a breakpoint on NPE.
This eventually shows a NPE occurs in com.sun.enterprise.resource.ConnectorXAResource on line 255
Payara/appserver/connectors/connectors-runtime/src/main/java/com/sun/enterprise/resource/ConnectorXAResource.java
Line 255 in 189529b
In my case the values are:
The handling of the 'h == null' situation is missing:
As far as I can see this line of code was introduced in the fix for #769
https://github.com/payara/Payara/blame/189529b7fb3b00f282ab7c3315ce53f96a6f93a7/appserver/connectors/connectors-runtime/src/main/java/com/sun/enterprise/resource/ConnectorXAResource.java#L255
Perhaps the code should be:
To make it null pointer safe? But I do not have enough knowledge to understand why there would not be a j2eetran.getNonXAResource() is some cases. EJB Pool or database pool limit reached in an earlier stage?
The stacktrace at the moment the NPE occurs is:
Eventually this NPE is handled in handleResourceException and the server.log shows the PoolingException: java.lang.NullPointerException:
Environment
The text was updated successfully, but these errors were encountered: