IO Error 'Connection Reset' and 'Connection Timed Out' on DB jobs/varas

Products

• CA Automic Workload Automation - Automation Engine
• CA Automic One Automation

Issue/Introduction

Users may encounter jobs or workflows failing with the following JDBC-related errors:

• U02012033 Query could not be executed: 'IO Error: Connection reset
• U02012033 Query could not be executed: 'IO Error: Connection timed out

Environment

This can happen on any version

Cause

These errors typically originate outside of the Automic application at the network or database layer:

• Connection Timed Out: Usually caused by a 'quiet close' where a firewall or network device silently drops an idle TCP connection without notifying the client (JDBC driver).
• Connection Reset: Indicates that the connection was actively terminated by the remote host, firewall, or database, which then sent a Reset (RST) packet to the JDBC driver.

Automic is behaving as designed by accurately reporting the error returned by the JDBC driver.

Resolution

As these are infrastructure-level issues, they must be investigated by Network or Database Administration teams. However, the following Automic-side configurations can help mitigate the impact:

1. Upgrade to v24.4.4 HF1 or Higher

Starting with v24.4.4 HF1, the database agent has been enhanced to ensure that idle connection cleanup happens asynchronously. This prevents a single hung connection from blocking other downstream job requests or variable resolutions.

2. Configure TCP Keep-Alive

To prevent firewalls from silently reaping idle connections, enable keep-alive at the connection object level to maintain 'chatty' traffic:

• Add oracle.net.keepAlive=true to the JDBC connection properties.

3. Adjust Connection Pooling

If connections are being dropped due to idle timeouts, configure the agent or connection pool to prune them faster than the network's timeout limit. For example, setting a low retention_time ensures idle connections are closed cleanly by the application before the firewall intervenes.

4. Additional Troubleshooting

As the errors are being returned by jdbc and reported by Automic, true root cause of the issue will need to be done together with a DBA and/or network team.  It is possible to lower the re-use of idle connections by setting a low retention_time to ensure idle connections are closed cleanly by the application before the firewall intervenes.