This is my first blog post in SCN and I hope this information helps. The phase ACT_UPG is a phase where ABAP Dictionary customer objects are activated and SPDD adjustments are done. As can be seen from the Web, there seems to be a common issue in this phase – the performance of the phase. More often than not, customers would find themselves stuck in this phase for a very long time. Here are some notes and important points to ponder, before proceeding with ACT_UPG, which may help lessen the time for execution of this phase.
The first thing to note in this regard is to NOT ignore non-severe errors – this may result in errors in a later phase of the upgrade. It is always best to consult SAP and get a word of confirmation regarding this. Since non-severe errors might sometimes result in inconsistencies, it is best to rectify them in this phase and proceed, unless otherwise specified by SAP. The next thing is to download the latest tp and R3trans executables from the service marketplace – it is always recommended and considered as a best practise by SAP to update the kernel to the latest patch level before proceeding further with the upgrade. In this way, several bug fixes are delivered by the latest kernel patch. Errors such as “unresolved requests in buffer” are most commonly resolved by updating the tp and R3trans to the latest levels. The note
19466 – Downloading SAP kernel patches;
would help in locating the correct SAP kernel patches for a system.
It is also observed that a sufficient amount of memory in the application servers increase the overall performance of the upgrade. This is an obvious fact and can be easily followed. This also implies and stresses upon the fact that the upgrade directory DIR_PUT, the trans directory DIR_TRANS be kept in the local disk so that access to these directories is faster than would be if they were mounted on the local server and located on a remote server. If the upgrade directory is placed in a remote disk, the performance of not only ACT_UPG, but of the overall upgrade would depend on the network connection speed. Sometimes, however, despite having a local upgrade directory, the network layer is used, and thus, the configuration of the network is important. A common case that highlights this is as follows – a production system is a high availability system while a test system is not. Hence, the production system would use virtual hostnames that have to niping back to the same system. Here, the network setup is different to that of a non-production system. Hence, the location of these directories would be decisive factors in determining the performance of the phase.
The hardware is to be looked into, in this regard, as the speed of the hardware proportionates directly to the speed of the upgrade. Here, the values for abap/heap_area*non*dia, abap/heap_area*total and ztta/rool_area could be increased. Such parameters could be tweaked according to the requirements. If the database is on the host where the upgrade tool and R3trans run, or on a different host, might influence the speed, but even a local database access uses the network layer and thus network configuration for the database host is also an issue. This was discussed in the previous paragraph. Finally, here are some notes that provide more detail about the commonly encountered errors in ACT_UPG and include performance tweaks.
1630256 – Usage of the “memory-optimized activator” during Upgrades and Enhancement Packages Installations
1674812 – Performance improvement for ACT_UPG
1563641 – TSV_NEW_PAGE_ALLOC_FAILED dump occurs
1387739 – Out of memory in ACT_UPG phase during EHP installation
744343 – Tips for importing Support Packages with minimized downtime;
558197 – Upgrade hangs in PARCONV_UPG, XPRAS_UPG, SHADOW_IMPORT_UPG2
589124 – Performance improvements when Support Package imported;
132536 – Improved semaphore mechanism in tp;
1341975 – Add. info. on upgrading to SAP ERP 6.0 including EHP5 ABAP