FAQ
List of frequently asked questions and often encountered issues with nevisIDM.
Database problems and errors
List of questions about issues with the database or connectivity to the database.
The Oracle database connection is very slow and the next errors appear randomly in my logs: java.sql.SQLRecoverableException: IO Error: Connection reset by peer, java.io.IOException: Connection reset by peer or*java.sql.SQLException: ORA-12592: TNS:bad packet.*
How can I fix this?
One reason could be a bad connection between nevisIDM and the Oracle database server. Therefore, ensure that you have a steady, good connection between nevisIDM and the Oracle database server, without a firewall that randomly terminates idle connections.
Another reason could be a slow random number generation. Oracle connections require a certain amount of entropy from the Linux kernel. When the pool is drained and there is not enough entropy to generate good random numbers, the Oracle driver is not able to open new connections. This leads to sporadic connection errors. To fix this, apply one of the following solutions:
- Ensure that the host does not run out of random entropy. This sometimes happens, especially on virtual machines. To prevent this, follow the guidelines of your VM and Linux OS.
For example, the nevisAppliance makes use of haveged and rngd to always have enough entropy available. 2. Use /dev/urandom as random number source for your JVM. To do so, use the JAVA_OPTS expression as shown in the code snippet below. Either adjust this expression in the file env.conf (for Standalone deployments) or add it to the file vmargs.conf (for WildFly):
# Extend JAVA_OPTS with the following
JAVA_OPTS="-Djava.security.egd=file:///dev/urandom"