larger JVM heap size in Gateway 10 vs Gateway 9.4

Products

• CA API Gateway

Issue/Introduction

We see a larger JVM heap size means cost of major garbage collections go up.  Please note to them that under v9.4 we did not note this large size. But in 10.0 we are seeing high jvm apparently due to garbage collection. It has been noted in certain cases like large message sizes, and audits of respose/request (large message) that the new GC uses more resources. Where smaller messages the new gateway 10 Garbage Collection method is efficient.

Environment

Release : 10.0

Component : API GATEWAY

Resolution

1) download the appliancedefs.sh I attached to the case its from a 10.0 system but I modified it as we did on another case that the customer was having issues with the NEW 10.x GC method causing issues.

2) Scp this to your 10.0 test host(s).

 

3) cp /opt/SecureSpan/Gateway/runtime/etc/profile.d/appliancedefs.sh /opt/SecureSpan/Gateway/runtime/etc/profile.d/appliancedefs.orig

4) Copy the scp file cp /home/ssgconfig/appliancedefs.sh /opt/SecureSpan/Gateway/runtime/etc/profile.d/appliancedefs.sh

5) chmod 555 /opt/SecureSpan/Gateway/runtime/etc/profile.d/appliancdefs.sh

6) chown  layer7:layer7 /opt/SecureSpan/Gateway/runtime/etc/profile.d/appliancedefs.sh

7) restart the gateway. Validate your jvm usage.

appliancedefs is attached to this article.

Changed lines of appliancedefs below for reference,

# Setting larger permsize for java 1.6   <=== Exist in both versions 4 lines below are updated(3 are new 1 is now a comment).
# Remove new 10.0 GC Can cause issue in certain size messages
#NODE_OPTS="-Xss256k -XX:+UseG1GC -Xmx${java_ram}k -Xms${java_ram}k -XX:G1HeapRegionSize=${G1HeapRegionSize} -XX:+AlwaysPreTouch -XX:+ParallelRefProcEnabled -XX:+UnlockExperimentalVMOptions -XX:InitiatingHeapOccupancyPercent=70 -XX:G1NewSizePercent=20 -XX:G1MaxNewSizePercent=60"
# Add Old 9.4 GC Garbage Collection
NODE_OPTS="-Xmx${java_ram}k -Xss256k -XX:+UseParallelOldGC"