Last Update: 6th of June 2019
Upgrade to S/4HANA 1809 FPS01/FPS02
The short Version – SAP S/4HANA On Premise 1809 Upgrade Guide
So, Upgrading to S/4 1809 FPS01 and no (really) news to know? Actually I mentioned a lot of things already in the Blogs about Upgrading to S/4HANA FPS01, the Import of FPS02 and FPS03 to S/4HANA 1709.
The Software Stack comes with a complete set of files, including also the DVD’s which had to be downloaded separately with the content of the stack.xml (around 217 files with 47,4 GB, FPS02 comes with 238 files with around 48,2 GB)
Creation of CISI and stack.xml file
as mentioned in the Blog – S/4HANA 1709 FPS03 – back on the Mothership again … the correction if the installed software information (CISI) will make sure that the stack.xml for the upgrade process will be as consistent as possible.
The big picture of planning landscape changes is shown in the picture above. Click it to find more details on the landscape management process with involved roles, steps, and tools.
Verification in the Maintenance Planner – Part I: Principle and Evolution
Verification in the Maintenance Planner – Part II: How-to & Examples
Recommendation for Landscape Data Gathering: Use HTTP(S) for All Data Suppliers including RZ70
(Blogs created by Wolf Hengevoss)
Nevertheless due to former software uploads to the EPS/in Box it can happen, that the content of the stack.xml become inconsistent, even that everything is correct. So before the first upload of the new software stack will be done, it is suitable to delete the old files from EPS Inbox, even when the necessary files are already available.
In this Example there was the Problem, that two identical files caused a duplicate entry in table TEPSIN.
Run Report /SDF/RC_START_CHECK
Before you start the Upgrade (and I should better read my own Blogs … ;-)), you should check with the Report /SDF/RC_START_CHECK the overall consistency of the S/4 System.
For that you have to make sure, that the necessary SAP Note Correction are applied, before you reach the SUM Phase where the error occurs.
Note 2708143 – SAP S/4HANA 1809 SP01 Conversion & Upgrade – TCI Note #6
(contains the TCI file K102008CPS4CORE.SAR)
Starting the SUM Process
Latest Information on the Software Update Manager (SUM)
As of SUM 2.0 SP03, a new user interface (UI) based on SAP Fiori is available for the Software Update Manager 2.0 and will replace the current SUM 2.0 UI with one of the next releases. You have already the option to use and get to know it. However, bear in mind that this user interface may be still enhanced or maintained, if needed.
To start the Software Update Manager 2.0 with the new UI, just replace the last word “sluigui” in the known URL with “slui” and enter the following:
https://<fully qualified hostname>:1129/lmsl/<mode>/<SID>/slui
Note 2644862 – Central Note – Software Update Manager 2.0 SP04 [lmt_020]
Note 2690522 – Central Note – Software Update Manager 2.0 SP05 [lmt_021]
Note 2691785 – SAP HANA: Additional information – SUM of SL Toolset 1.0 SP25
Note 2735060 – SAP HANA 2.0 SPS 03 Database Revision 036
As I learned from the last Upgrade, I have increased the SUM Process Parameters to improve the long runtimes for certain phases …
You can expect a very long runtime here, depending on the amount of Support Packages which are included. It can also be the case that you are running out available HANA Memory which leads to the same error message here …
Fazit: provide as much Memory for HANA as you can afford, but at least 350 GB in this Phase.
It can be easily the double mentioned amount needed. Unfortunately there is no SAP Note which fits to the actual problem here.
See also the Wiki – SHADOW_IMPORT* phases
some nice things to know, if you ignored inactive Objects which will be shown again in Phase REPACHK2:
- if there are Problems with inactive objects which are either not available or fragments from older maintenance processes you can first
- if nothing of the above actions will solve the problem, you must either delete or activate the affected objects. In most of the cases this will not necessary, or you played with a S4 IDES implementation around the objects “Z_DEMO_01” … 😉
! If you ignored the activation of DDL Sources here, the follow up error will occur in the Phase PARDIST_CDS. So, despite to BW or BW/4 Upgrade don't miss to activate the Structures right here with the Report RUTDDLSACT mentioned in the log file ACTUPT.LOG/SAV
looks good so far, but …
So what happened here? I have checked Google, CSN and so on but not really a fruitful solution …
Note 2085980 – New features in memory management as of Kernel Release 7.40
nice overview, but affected parameters are missing …
Note 2180736 – TSV_TNEW_PAGE_ALLOC_FAILED
nice root cause analysis, but no answer how to solve the problem …
Note 2417223 – LOAD_NO_ROLL dump and Extended Memory exhausted as of Kernel 74x
nice description, but wrong direction for the root cause …
Note 2706314 – Negative smoothed and averaged values at performance MTE nodes in RZ20, negative values in ST06
Nice to know, error listed in the work process trace, but not related to the root cause …
finally the SAP Help allowed a educational guess to increase the parameter ztta/roll_extension, as the Kernel default is much to low. So I added the following set of parameter’s to the Profile of the Shadow Instance and restarted the Instance:
abap/buffersize = 1400000 abap/heap_area_total = 108197104800 abap/shared_objects_size_MB = 1024 em/initial_size_MB = 65536 em/global_area_MB = 49152 em/proc_max_size_MB = 128000000 em/proc_use_stdheap = TRUE ipc/shm_psize_10 = 198000000 ipc/shm_psize_40 = 1978000000 rsdb/esm/buffersize_kb = 524288 rsdb/obj/buffersize = 524288 rsdb/otr/buffersize_kb = 16384 rsdb/tbi_buffer_area_MB = 2048 rdisp/PG_MAXFS = 128000 rdisp/PG_SHM = 128000 ztta/roll_extension = 40000000000
voila, it worked! this already the next screen …
You have Problems stopping/starting the SAP Instance, e.g. MAIN_PROFSTART/ STOPSAP_TRANS: no Problem …
- ps -ef | grep <sid>adm => kill the orphan processes - in /tmp ls -la | grep <sid>adm => delete the .sapstream* files - start the system normally - continue with the phase => voila, it worked!
now it became a bit strange, never heard about the phase RUN_UCON_REGISTER_FUNCS_DT
the good message: it doesn’t occur with the FPS02 Upgrade …
The SAP Note mention the parameter ucon/rfc/active to ‘0’ which was the case.
So, I checked the affected table: VBAK (uhm, don’t touch it …)
However, after quite a while of compiling at this stage, the SUM procedure continued …
Then: the Phase MAIN_NEWBAS/PARCONV_UPG started …
and what followed was intensive search and activation Phase of a lot of Data Definitions from a Travel Add-On for S/4 how on Earth came this Add-On on the System?
Ah, yes it’s an IDES system …)
After more than three hours of SE11, SE14, and the ABAP tools for Eclipse I eventually found these SAP Notes, which tells you to ignore this and run the activation of the HANA Views after the Upgrade with the Report RUTDDLSACT …
Note 2298340 – SAP HANA DB: CDS views with external views as base objects cannot be created in the DB
Note 2478641 – DMO Upgrade error in phase MAIN_NEWBAS/PARMVNT_UPG View using external HANA view cannot be created
Note 2756802 – DD: DDLS hierarchies in SUM phase MAIN_SWITCH/PARMVNT_VIEW
finally also the Upgrade continued. It seems the new Objects can with the Import as a comparison with an identical system showed some differences here ...
Feedback from the 2nd Upgrade:
Don’t ignore the inactive DDL Sources during ACT_UPG and PARCONV_CDS and tune the shadow Instance as mentioned, then this will be an almost smooth Upgrade ...
Furthermore challenge the SUM Process Parameter.
Again: Don’t ignore the Phases I just mentioned before.
However this is a new Error and it can be solved in the same way.
Check the Short Dump with ST22 and who is the “Last Author”:
SAP (don’t worry SAP, it is not your fault … 😉
We can use our nice Report RUTDDLSACT which can be used to find the cascade to the root cause: the initial DDL/View which causes the Short Dump.
But: How is the relation from a missing table type to inactive DDL/View?
Going down the activation cascade, finally end at the table BNKA
/PF1/CV_ORDER_MANAGE /PF1/CV_ORDER_MG /PF1/I_DB_ORDER /PF1/C_VH_ACCOUNT_ID /PF1/I_FCLM_BAM_ACLINK2 /PF1/P_BNKA => BNKA
Table BNKA were enhanced with two appends with the same content
– Append structure /PECROSS/STR_ROUTING_GPATT
– Append structure /PF1/STR_ROUTING_GPATT
TABL BNKA was not activated Check table BNKA (KRAMER/09.05.19/14:58) Append structure /PECROSS/STR_ROUTING_GPATT appended to table BNKA Append structure /PF1/STR_ROUTING_GPATT appended to table BNKA Enhancement category 2 possible (include or subtype not yet classified) Field FED_FUNDS in table BNKA is specified twice Field FED_OVER_RT_CODE in table BNKA is specified twice Field RT_STATUS in table BNKA is specified twice Group name GPATT in table BNKA is specified twice Index BNKA-HWV must be converted BNKA: Structure change at field level (DB is adjusted) Check on table BNKA resulted in errors
As the Table belongs the the Payment Engine, I decided to use the append from the Add-On …
voila: Phase XPRAS_AIMMRG continues and finish successfully.
Update from 4th of June 2019
Implementing the two SAP Notes also causes trouble with inactive views/DDLs
Note 2796473 – Timing field in MtM Current Date Query – CDS + DDIC
Note 2800034 – ATC Error Fixes for CDS Views
- P_VERSIONEDLOGISTICDOCUMENT - P_VLOGP_DENDRUN - P_VLOGP_DENDRUN_BPO - P_VLOGP_DENDRUN_BPOX - P_VLOGP_DENDRUN_MTM - P_VLOGP_DENDRUN_NA - P_POSITIONREP_COMPOSITEKEYFIG - C_CMMDTYPOSCURRENTDATEQRY ICMMUNIONPFCUBE ICMMUNIONPFEXT PCMMUNIONPF PVLOGPPOSINDOD PVLOGPPOSRCKF PVLOGPPOSRSKF PVLOGPPOSREPEKF PVLOGPKONVDPR - P_POSITIONREP_SIMPLEKEYFIG - P_UNIONVIEWPHYSFIN PNETPOSBINERRCT PKONVDERCTDEND PKONVDERCTPD PKONVDNDUPDTDDT PKONVDNDVALIDREC PVLOGPKONVDUNQ - P_VLOGP_DENDRUN_POSITIONREP - P_VLOGP_POSITIONREP - P_VLOGP_PRICED_UNPRICED - P_VLOGP_QTYSPLIT_QTYPERDAY_ADJ - P_VLOGP_QUOTATION_TIMESERIES - P_VLOGP_REPORTINGDATE - P_VLOGP_REPORTINGDATE_YRMNTH - I_VERSIONEDLOGISTICDOCUMENT
btw: the mentioned DDL View is now active as well during the table activation
if you forgot to lock the system again after this maintenance (it is possible as you are already in the Phase XPRAS_AIMMRG) you have lock the system manually as it is not automatically locked again.
even that the error message a bit misleading, I also encountered that the client 000 didn’t had a logical system assigned and the corresponding RFC connection was missing as well.
This is another example of an inconsistency in a working system which was never recognized before and causes problems once the system is in an Upgrade Process.
Not recognizing errors, doesn’t mean a system run’s properly …
Ups, the System was started but the SUM process reported an error. root cause:
somehow the files for the ICM were missing in the 7.73 Kernel (didn’t happen in the second upgrade)
Business as usual and I can see already the finish line …
And sometimes some DDL Views cannot be cleaned out … (In this example the developer left the company … ;-))
Save the Logs …
Any finally (Drum Roll …)
System looks good so far, S/4 1809 FPS01 …
And the final status of S/4 1809 FPS02 …
At this stage I’m not sure if I like the overall run time with the resources …
btw: I used resource-minimized instead of downtime-minimized …
the second run with more support packages included and increased amount of processes shows a better result ...
S/4 on premise is back!
Roland Kramer, SAP Platform Architect for Analytics SAP SE
“I have no special talent, I am only passionately curious.”