cancel
Showing results for 
Search instead for 
Did you mean: 

solman 7.0Ehp1 homogenous syscopy - JMT shows wrong src cluster ID

francois_keen
Participant
0 Kudos

Dear Community.
I'm doing a homegenous system copy of a solman 7.0 EhP1 SP25 system as our source hardware is going to be out of support.
We are keeping the SID the same = CEN , same system number = 00, only hostname changes from source=usorlu95 to target=smprd
I'm facing the symptom described in SAP note 966752 section "The source central instance ID is wrong"

The jmt jmt system.log gives:
"#######Thread [main,5,main]
##0#0#Error#1#com.sap.sdt.mc.systemcopy#Plain###The source or target cluster ID is not present on the system! The current (source) cluster ID is 30340 and the new (target) cluster ID is 55530#"

the jmt system.log is right because the existing source cluster ID is 124123 and not 30340

so i went in the configtool of the source, on the source host usorlu95 and got the box number:

Property sheet instance.properties.ID124123
box number = CENDVEBMGS00usorlu95


the data in the exported java stack is correct, the src.ci.host is correct=usorlu95
usorlu95:cenadm 33% more SOURCE.PROPERTIES
src.ci.sid=CEN
src.ci.instance.number=00
src.ci.instance.name=DVEBMGS00
src.apps=AS,AAS,SM-APP,SM
src.export.version=2.0
db.copyWithDBtools=true
src.export.typeIsCopy=false
src.ci.host=usorlu95


so following the note 966752, I should edit /tmp/sapinst_instdir/SOLMAN/LM/COPY/ORA/SYSTEM/CENTRAL/AS/jmt/cluster_id_switch.properties and adjust the line 'src.ci.host=<host name>'.

grep src.ci.host cluster_id_switch.properties
src.ci.host=usorlu95


Please note that the JRE on the source and target is exactly the same [an old 1.4.2_09-b05, yes this is very old, I plan to migrate to SAPJVM
4 once we have moved to the new server].

The media used for the source java stack export and the target installation is the same master DVD [and latest available]: 51040437
SolMan 7.0 EhP1 SR1

It looks like there would be some "calculation" problem when determining the cluster ID of the source. I got no idea what is the
algorythm behind this cluster ID calculation [except that the calculation is based on the boxnumber as per note 966752]

the source and target platforms remain the same : Solaris 10 on SPARC as this is a homogemous system copy. Source platform Sun-Fire-V210 [CPU =SUNW,UltraSPARC-IIIi] and the target platform is SPARC-Enterprise-T5220 [CPU = SUNW,UltraSPARC-T2]. The DB is the same on source and target oracle 10.2.0.4 same SBP SAP_102044_201005_SOL64 [yes I know, we need to upgrade to Oracle 11, this is planned once we will have moved to the new hardware]

So I do not know wherelse I should look to solve the problem.

I opened an OSS message yesterday but as it is a medium priority it is going to take time to get feedback from SAP Support.

Have you ever experienced this problem before? Could you please provide some guidance?

Many thanks for your help and expertise

Best regards

Francois

Accepted Solutions (1)

Accepted Solutions (1)

francois_keen
Participant
0 Kudos

Hi

With SAP support help, it was found that my source system used to have a system number 01 for a couple of months 6 years ago.

so following sap note 966752 by adjusting src.ci.instance.number and src.ci.instance.name did the trick.

and utimatelly on another system copy, i just modified the SOURCE.PROPERTIES of the java stack export of the source so that the taget install process wouldnt complain at all.

Why the export doesnt take the real isntance number and box number into account is a mistery...

Answers (1)

Answers (1)

Former Member
0 Kudos

For my case, we ran a restore of the database which includes the java and abap stack. however, we did not actually a java export which would export the java in the filesystem of the source. what we did in the CI install, we used java export with a different SID. I changed the source.properties SID and the hostname.