Vulnerable undertow version detected in AIQUM 9.18 causing high CPU usage and GUI unresponsiveness
Applies to
- Active IQ Unified Manager (AIQUM) 9.18 and above.
- RHEL/OVA/Windows
- CVE-2023-1108
Issue
After upgrading to AIQUM 9.18(P2), the following symptoms may occur:
- High CPU usage on the AIQUM Java/JBoss process
- A single Undertow worker thread (for example
default I/O-2) consumes approximately 100% of one CPU core continuously - AIQUM Web GUI is slow or partially unresponsive
- The condition does not self-resolve and may require a service restart to recover
- Excessive growth to audit-log.log during vulnerability scanning.
- Audit-log.log growth may result in insufficient disk space
- The issue may start after upgrading to AIQUM 9.18P2D2
- The issue may be triggered when HTTPS/TLS connections are abnormally terminated (for example TCP reset during handshake), which can occur during monitoring, load balancer health checks, or security scanning activity
On Linux, high CPU can be confirmed with:
top -H -p <jboss_pid>
Example output:
default I/O-2 R 99.7 ...
Thread dumps may show a stack similar to:
sun.security.ssl.SSLEngineImpl.wrap at io.undertow.protocols.ssl.SslConduit.wrapAndFlip(SslConduit.java:1009) at io.undertow.protocols.ssl.SslConduit.doWrap(SslConduit.java:940) at io.undertow.protocols.ssl.SslConduit.doHandshake(SslConduit.java:673) at io.undertow.protocols.ssl.SslConduit$5$1.run(SslConduit.java:1153)
Configuring Undertow read, write, or request-parse timeouts does not resolve this issue because the thread is actively spinning in user-space SSL code rather than waiting on socket I/O.
