Additional Blogs by Members
cancel
Showing results for 
Search instead for 
Did you mean: 
Former Member
0 Kudos

TMS: Large number of transports – things to be checked

We have just completed one of the upgrade projects in which as a correction we had more than 10K transport request (or transports or just requests) to be imported in prod system into the system as a post facto.  We had to spend lot of time on understanding and accelerating the transports to minimize the overall business downtime. It was a good experience which I am sharing here. Thanks to all SAP guys and my colleagues who gave positive response to my questions and extended their help.

Challenge

Starting with the correction imports after and upgrade, we understood that the imports are taking more than 32 hours to complete and yes these were not all the corrections, something more was also planned as an addition.

Analysis and solution  *Location of transport files: <p style="margin: 0in 0in 0pt 0.25in; text-indent: -0.25in; text-align: justify" class="MsoListParagraphCxSpMiddle">o   All the transport files (/usr/sap/trans) – data files (R files), co files (K files) and log files should physically reside on the production server. This helps in reducing overall time for imports. We had all the transport files on the production host. If this is not possible then there are specifications given at location http://www.ietf.org/rfc/rfc3530.txt for NFS to be used during TMS import run.</p><p style="margin: 0in 0in 0pt 0.25in; text-indent: -0.25in; text-align: justify" class="MsoListParagraphCxSpMiddle">o   In case of use of transport files over NFS, Set parameter lock_check_wait_time as low as possible (1); the value must never exceed 10 and its value must exceed the values of the NFS parameters "acregmin," "acregmax," "acdirmin" and "acdirmax" which can be set for each "mount."</p> *Changes in the transport profile:

o   We set the parameter "acc_import=true" for large number of small transports. This activated the "accelerated import". Setting up this parameter has not shown us any disadvantages but added a good value.

o   As we had lot of ABAP Dictionary objects (for example, domains, data elements, tables, indexes, views, and so on), we set the parameter "activatemerged=true". This has ensured the system did not activate the ABAP dictionary for each request. It performed a combined activation for all ABAP Dictionary objects at the end in the queue. Setting up this parameter has not shown us any disadvantages but added a good value.

o   Parallel process for import been done by parameter ‘parallel =n’. Based upon our experience and calculations I kept n as 4 for final run.

 Points not considered