on 08-25-2009 10:12 AM
Hi all,
I'm facing a problem during installation ECC 6 SR 3 in Solaris 5.10 & Oracle.
When importing ABAP (task SAPSSEXC), it is running too long, and not continue till table REPOSRC....I searched for some note and found note 454368 saying that I should run R3load with option -merge_bck.
As <SID>adm, I tried these : 1. R3load -merge_bck <TASK> <TASK.bck>, 2. R3load -merge_bck <TASK>, 3. R3load -merge_bck <BCK>, none of them worked.
Has anyone experience this ? Please advice, thanks.
Regards
This is a strange issue because every time the process reached 6 hours processing, the import was just stop there without a trace or error log. Running import command as daemon was work for my case.
You must be a registered user to add a comment. If you've already registered, sign in. Otherwise, register and sign in.
>acing a problem during installation ECC 6 SR 3 in Solaris 5.10 & Oracle.
>
> When importing ABAP (task SAPSSEXC), it is running too long, and not continue till table REPOSRC....I searched for some note and found note 454368 saying that I should run R3load with option -merge_bck.
This is necessary if the import was interrupted and a .bck file exist.
> As <SID>adm, I tried these : 1. R3load -merge_bck <TASK> <TASK.bck>, 2. R3load -merge_bck <TASK>, 3. R3load -merge_bck <BCK>, none of them worked.
What did you expect them to do?
What database version and patchlevel are you using and what is your hardware?
Markus
You must be a registered user to add a comment. If you've already registered, sign in. Otherwise, register and sign in.
Thanks Markus.
I've tried merge_only previously to re-merge the bck file so the installation running properly.
The thing is, R3load runs too long like hours and the last status stop on REPOSRC in the task file, just as per explain in the note, so I used the merge_bck hoping it works faste, my oracle version is 10.0.2.0 and all hardware & sw requirement already met.
Regards,
> The thing is, R3load runs too long like hours and the last status stop on REPOSRC in the task file, just as per explain in the note, so I used the merge_bck hoping it works faste, my oracle version is 10.0.2.0 and all hardware & sw requirement already met.
It's nothing uncommon that loading a database using R3load takes some time.
I was asking about your hardware because there are known issues/problems with Txxx servers from Sun concerning single threaded applications as Oracle and SAP.
Markus
Hi Markus,
My server is T5220.
For me it is uncommon, since I'm waiting almost 1 day just for that 1 task, considering the other task with much more bigger size has done within 2 hours. Thanks.
Actually I still don't get my answer, what are the parameter to use the -merge_bck option ?
Regards,
Edited by: sprout009 on Aug 25, 2009 11:42 AM
> My server is T5220.
REPOSRC is a table with LOBs which takes longer than normal transparent tables.
> For me it is uncommon, since I'm waiting almost 1 day just for that 1 task, considering the other task with much more bigger size has done within 2 hours. Thanks.
The "T-Server" series is known to have performance issues, that server architecture is not really suited for single threaded SAP applications (as ABAP + Oracle is).
and also
Oracle Metalink 781763.1 "Migration from fast single threaded CPU machine to CMT UltraSPARC T1 and T2 results in increased CPU reporting"
> Actually I still don't get my answer,
I don't know what you did expect. I have no access to your machine, I don't know how your database is configured, whether you use e. g. "-loadprocedure fast", I know nothing about your system and you expect me to look in to a magic ball to tell you why your load is running "long"? I'm sorry, I'm trying to help but I need more input.
> what are the parameter to use the -merge_bck option ?
No parameter, it's used if you stop the import to merge the current imported part with the part not-yet-imported. R3load -h would give you the options.
I doubt that there's much you can do about the speed. It's just how the server and it CMT-CPU is designed --> not good for ABAP. Others had that problem too and since there is an official Oracle statement about this you can consider it true.
Markus
> Your explanation is helpful. Do you know if SAP releases a note around this performance issue on Txxx server? I guess I just have to wait a bit longer then.
No SAP not - but an Oracle metalink note.
> I haven't tried the -loadprocedure fast option, does it accelerate the process ?
Yes, that option uses a different algorithm for loading the data (Oracle fastloader):
Note 1045847 - ORACLE DIRECT PATH LOAD SUPPORT IN R3LOAD
Markus
Hi all,
My bad that I have to raise this topic again. After waiting for so long the import for this REPOSRC table still doesn't move.
The first error saying that no enough space, my harddisk allocation is fine (free), then i provide the memory 4Gb. I continued with the import after that, but still it doesn't move. The weird thing is that the process stopped and gone ( I ps -ef | grep R3load ), but there is no error log. No update on log file showing any error.
Anyone has experience this ?
Regards
Edited by: sprout009 on Aug 27, 2009 4:27 AM
Edited by: sprout009 on Aug 27, 2009 4:27 AM
Edited by: sprout009 on Aug 27, 2009 4:29 AM
> The first error saying that no enough space
Where did you get that error?
>, my harddisk allocation is fine (free), then i provide the memory 4Gb. I continued with the import after that, but still it doesn't move. The weird thing is that the process stopped and gone ( I ps -ef | grep R3load ), but there is no error log. No update on log file showing any error.
The logs won't be written until the table is imported.
Check with
prstat -a 1
if you see activity.
Markus
Hi Markus,
The no enough space log is stated on dbase alert file. The proses was there for hours, and then at 1 point, it stops and no error or additional changes to the log file.
Now, as SIDadm I used this command :
/usr/sap/<SID>/SYS/exe/run/R3load -i SAPSSEXC.cmd -dbcodepage 4102 -merge_bck -loadprocedure fast LOAD -l SAPSSEXC.log -stop_on_error
During the process, the screen output is stay as follow, nothing more :
sapparam: sapargv( argc, argv) has not been called.
sapparam(1c): No Profile used.
sapparam: SAPSYSTEMNAME neither in Profile nor in Commandline
Is this the way that it should be? Or there is something that I should fix with this output ?
In SAPSSEXC.log the last output is as follow and no more changes:
.......
DbSl Trace: my_free1: size=4216, sum=mem_dbms=600611
DbSl Trace: my_malloc1: size=10240128, sum=mem_dbms=10840755
Thanks.
Regards
User | Count |
---|---|
79 | |
9 | |
9 | |
7 | |
7 | |
7 | |
6 | |
5 | |
5 | |
4 |
You must be a registered user to add a comment. If you've already registered, sign in. Otherwise, register and sign in.