Like in a NetWeaver 2004s Java system, the automated kernel distribution works with the kernel program SAPCPE which checks, and if necessary, copies newer version of the kernel binary files from the central kernel directory to the instance-individual executable directories. Die individual instances start from their own executable directories. You need perform the configuration service which comes with SAPinst of the SP level 19 or higher only once to enable SAPCPE. As the screen-shot below shows, you have to start SAPinst on the central instance host and choose the service Enable Automated Kernel Distribution Using SAPCPE. With some further specifications the function is configured for all future updates (You can download the Support Package Stack Guide for NW2004 for detail information about the configuration steps). Every time when the kernel binaries on the central instance have been updated, SAPCPE will replicate them to the dialog instances upon restart.
You should also take following restrictions or conditions that are significant for the proper functioning of the kernel replication into account:
- The automated kernel replication and the configuration service for SAPCPE works only in Java-only systems, not in double-stack systems (ABAP+Java).
- The Java-system cluster must be homogeneous. Dialog instances running on hosts with operation systems other than the central instance require different versions compiled for the individual operating systems. Due to file format incompatibility, kernel files for different operating systems are located on appropriate hosts. Therefore SAPCPE cannot simply replicate the version from the central instance.
- All instances must be at least of SPS17. As mentioned earlier, the kernel replication was impossible for NW04 due to some architectural limitations. With SPS17, these limitations are remedied. Dont be confused with the replication configuration service delivered with SPS19. The configuration service is included in patch tool (SAPinst) of SPS19 and above, but runtime systems of SPS17 already have the appropriate architecture to be configured for the automated replication. Therefore you can use the configuration service included in SPS19 to configure an SPS17 Java system.
I hope, the kernel replication function eases your daily work a little 😉