on 12-06-2008 1:18 AM
Does SAP make available some where a real procedure
for NetWever 7.0 ABAP stack only KERNEL UPGRADE ?
a lot of people have questions ? answer are even available... but still... even experienced
basis admin having trouble now with Kerne Upgrade
Main line :
stopsap, clean shared memory
stopsap
saposcol -k
ps -ef|grep sapstartsrv
kill <sapstartsrv_pids>
cleanipc <inst_nr>
lsnrctl stop
backup your kernel
cd /sapmnt/<SID>
cp -pr exe exe.001
extract new kernel
cd exe
SAPCAR -xvf <YOUR_DW.SAR>
or
. Login as root
2. Make new backup directory
cd /
mkdir SIDkernel_10102007
3. Create latest kernel backup
cp -pr /sapmnt/SID/exe/* /SIDkernel_10102007
4. Compare the original kernel and the backup
du -sk /sapmnt/SID/exe/ 840413 /sapmnt/SID/exe
du -sk /SIDkernel_10102007 841992 /SIDkernel_10102007
5. Make new shadow kernel directory
cd /
mkdir newkernel_175
6. Copy the new downloaded kernel files into /newkernel_175
7. Unpack the new downloaded kernel in folder /newkernel_175
SAPCAR -xvf SAPEXE_175-20000221.SAR
SAPCAR -xvf SAPEXEDB_175-20000223.SAR
8. Login as sidadm
9. Stop the SAP System
stopsap
10. Stop saposcol
saposcol -k
11. Login as root
12. Go to the current kernel directory
cd /sapmnt/SID/exe
13. Copy all extracted support packages into this folder
cp -pr /newkernel_175/* /sapmnt/SID/exe
14. Copy back the original file saproot.sh into this folder.
cp -p /SIDkernel_10102007/saproot.sh /sapmnt/SID/exe
15. Go to the current kernel directory
cd /sapmnt/SID/exe
Then run:
./saproot.sh SID
16. Login as sidadm
17. Start the saposcol service
saposcol
18. Start the SAP system
startsap
AND STILL ... FOLLOWING ALL THE STEPS... I have the work process going down
as soon as i put back old kernel is working fine again
I'm trying to upgrade to :
SAPEXE_179-20000978.SAR
SAPEXEDB_179-20000979.SAR
DBATL700O10_39-20000979.SAR
on an AIX 64 bit with Oracle 10G
Error I'm getting is :
FAIL: NIECONN_REFUSED (Connection refused), NiRawConnect failed in plugin_fopen()
and even the dpmon cannot work
dpmon pf=/usr/sap/H1D/SYS/profile/H1D_DVEBMGS00_dehrdb01
DpMonInit failed - possibly no dispatcher running
So, my QUESTION is : DOES SAP have a detailled procedure to do the Kernel Upgrade on NW7.0
for an ABAP stack only installation ?
And what I see is:
. Login as root
Then run:
./saproot.sh SID
16. Login as sidadm
17. Start the saposcol service
saposcol
18. Start the SAP system
startsap
It´s clear that it can´t work. You omitted one important step. If you extract/copy files using the user "root" the system can´t open file due to missing permissions.
I would just add an
cd /sapmnt
chown -Rf <sid>adm:sapsys exe
and then run ./saproot.sh .
Markus
You must be a registered user to add a comment. If you've already registered, sign in. Otherwise, register and sign in.
Error I'm getting is :
FAIL: NIECONN_REFUSED (Connection refused), NiRawConnect failed in plugin_fopen()
What is the full error message from the developer trace?
I´m doing kernel upgrades since more than a decade...
Markus
You must be a registered user to add a comment. If you've already registered, sign in. Otherwise, register and sign in.
Error found in work logs :
06.12.2008 03:06:47
ShmDetach
FAIL: NIECONN_REFUSED (Connection refused), NiRawConnect failed in plugin_fopen()
(2850872) Starting: local ms.sapH1D_DVEBMGS00 pf=/usr/sap/H1D/SYS/profile/H1D_DVEBMGS00_dehrdb01
(2732220) Starting: local dw.sapH1D_DVEBMGS00 pf=/usr/sap/H1D/SYS/profile/H1D_DVEBMGS00_dehrdb01
(868592) Starting: local co.sapH1D_DVEBMGS00 pf=/usr/sap/H1D/SYS/profile/H1D_DVEBMGS00_dehrdb01 -F
(2953220) Starting: local se.sapH1D_DVEBMGS00 pf=/usr/sap/H1D/SYS/profile/H1D_DVEBMGS00_dehrdb01 -F
(2412770) Starting: local ig.sapH1D_DVEBMGS00 -mode=profile pf=/usr/sap/H1D/SYS/profile/H1D_DVEBMGS00_dehrdb01
(2748518) Waiting for Child Processes to terminate.
Instance on host dehrdb01 started
(2748518) **** 2008/12/06 03:08:56 Child 2732220 terminated with Status 0 . ****
(2732220) **** 2008/12/06 03:08:56 No RestartProgram command for program 1 ****
here another one:
M NiWait: sleep (1000ms) ...
M Sat Dec 6 03:08:53 2008
M NiWait: sleep (1000ms) ...
M Sat Dec 6 03:08:54 2008
M DpSigInt: caught signal 2
M SigICallExitRoutine: call exithandler (ThSigHandler)
M ThSigHandler: call ThErrHandle
M in_ThErrHandle: 1
M ThSigHandler: signal (step 1, th_errno 11, action 2)
M ThIErrHandle: new stat of W1 is WP_SHUTDOWN
M ThIErrHandle: I'm during shutdown
M PfStatDisconnect: disconnect statistics
M ThIErrHandle: action changed to 2
M Entering ThSetStatError
M ThShutDownServer: shutdown server
M PfStatIndInit: Initializing Index-Record
M PfWriteIntoFile: copied shared buf (0 bytes) to local buf
M ThIErrHandle: do not call ThrCoreInfo (no_core_info=0, in_dynp_env=0)
M Entering ThReadDetachMode
M call ThrShutDown (1)...
A Saving PXA statistics.
A No programs loaded
B B dbtbxbuf: Saving TBX statistics:
B dbtbxbuf: 0 object entries written.
M ***LOG Q02=> wp_halt, WPStop (Workproc 1 512038) [dpuxtool.c 260]
M Good Bye .....
and another...
[Thr 1] MsHttpOwnDomain: own domain[1] = emea.int.tenneco.com
[Thr 1] ms/icf_info_server : deleted
[Thr 1] *** I listen to port sapmsH1D (3600) ***
[Thr 1] *** I listen to internal port 3900 (3900) ***
[Thr 1] CUSTOMER KEY: >P1592267744<
[Thr 1] Sat Dec 6 03:08:55 2008
[Thr 1] *** ERROR => MsSClientHandle: no server provides service ENQ (4), requested from dehrdb01_H1D_00 [ms
xxserv_mt. 4684]
and this one:
more sapcpe.log
================ >>> main =======================
INFO: sapcpe version: @(#) $Id: //bas/700_REL/src/ccm/sapstart/sapxxcpe.c#18 $ SAP
=================================================
INFO: Try to find file table sapcpeft (default:sapcpeft)
INFO: Try to find file table in the central directory /usr/sap/H1D/SYS/exe/ctrun/sapcpeft
INFO: not found any file table.
=================================================
Update Local Programs (dir-sync + file-path + ln)
=================================================
Starting at 2008/12/06 03:06:47
ERROR: can't change to central Executable-Dir /usr/sap/H1D/SYS/exe/ctrun (No such file or directory) [sap
xxcpe.c:1488]
or this one too :
[Thr 1] * LOCATION ICM runing on host dehrdb01
[Thr 1] * ERROR dp lib version mismatch, please check installation
[Thr 1] *
TIME Sat Dec 6 03:06:52 2008
[Thr 1] * RELEASE 700
[Thr 1] * COMPONENT DP
[Thr 1] * VERSION 241000
[Thr 1] * RC -9999
[Thr 1] * MODULE dpxxtool2_mt.c
[Thr 1] * LINE 688
[Thr 1] * DETAIL version of external and kernel dp lib mismatch (ext lib
version is 241000 / kernel lib version is 232000)
[Thr 1] * COUNTER 1
[Thr 1] *
[Thr 1] *****************************************************************************
[Thr 1] *** ERROR => DpIPCInit2: DpSysAdmExtCreate [dpxxtool2_mt 369]
[Thr 1] *** ERROR => IcmInit: IcmIPCInit failed (rc=-9999) [icxxman_mt.c 1741]
[Thr 1] *** ICM shutdown completed (pid: 2539700) ***
and sappfpar has no errors
we 1st tried this without success:
SAPEXE_179-20000978.SAR
SAPEXEDB_179-20000979.SAR
DBATL700O10_39-20000979.SAR
the we add the following with same result and revert back to kernel 146 (that one works fine)
SAPCPE_3-20000978.SAR
LIBRFC_41-20000978.SAR
CPICTLIB_11-20000978.SAR
DW_187-20000978.SAR
IGSEXE_13-10003077.SAR
or this one too :
* LOCATION ICM runing on host dehrdb01
* ERROR dp lib version mismatch, please check installation
TIME Sat Dec 6 03:06:52 2008
* DETAIL version of external and kernel dp lib mismatch (ext lib
version is 241000 / kernel lib version is 232000)
Does the same error also occur if you run "slibclean" as root just before you start the instance?
If yes then I´d say the kernel is wrong. I would try to uncar
- SAPEXE
- SAPEXEDB
- DBTOOLS*
and then use the latest
DW*
and try again.
Markus
root> /usr/sbin/slibclean
yes this command was used in all the attempts
this upgrade has been tried with 2 of the differents latest available kernels...
are you suggesting that maybe the problem could be the SAPCAR itself ?
but then someone would have to explain why the backup of the old kernel
made with the same SAPCAR was packed and unpacked again and still works fine
This is what we do with old kernel...
Logon as root
> cd /sapmnt/[SID]/exe
we create an archive with all files and directories from actual directory
into specified path without point files:
SAPCAR -cvf /sapmnt/[SID]/old_SAP_kernel.SAR *
Then as new kernel was not a success we brought back our archive
SAPCAR -xvf /sapmnt/H1D/old_SAP_kernel.SAR
It does not cost a lot to try donwload the latest SAPCAR but I'm not convinced
Kind regards,
R.S.
Here is the full procedure for SID = H1D
PROCEDURE for system H1D
h1dadm>
stopsap
saposcol -k
ps -ef|grep adm
kill <remaining sap process>
cleanipc <inst_nr>
to revert back to old kernel
cd /sapmnt/H1D/exe
/sapmnt/SAPCAR -xvf /sapmnt/H1D/old_SAP_kernel.SAR
root>
/usr/sbin/slibclean
cd /sapmnt/H1D
cp -pr exe exe.old
OR use SAPCAR to backup old Kernel
[ download latest SAPCAR and copy it to /sapmnt/ ]
cd /sapmnt/H1D/exe
/sapmnt/SAPCAR -cvf /sapmnt/H1D/old_SAP_kernel.SAR *
(extract new kernel - user root)
cd /sapmnt/H1D
mkdir kernel179
cd /sapmnt/H1D/kernel179
/sapmnt/SAPCAR -xvf /sapmnt/SAPEXE_179-20000978.SAR
/sapmnt/SAPCAR -xvf /sapmnt/SAPEXEDB_179-20000979.SAR
/sapmnt/SAPCAR -xvf /sapmnt/DBATL700O10_39-20000979.SAR
/sapmnt/SAPCAR -xvf /sapmnt/SAPCPE_3-20000978.SAR
/sapmnt/SAPCAR -xvf /sapmnt/LIBRFC_41-20000978.SAR
/sapmnt/SAPCAR -xvf /sapmnt/CPICTLIB_11-20000978.SAR
/sapmnt/SAPCAR -xvf /sapmnt/DW_187-20000978.SAR
/sapmnt/SAPCAR -xvf /sapmnt/IGSEXE_13-10003077.SAR
cd ..
chown -Rf h1dadm:sapsys kernel179
cp -pr ./kernel179/* /sapmnt/H1D/exe/
cd /sapmnt/H1D/exe
./saproot.sh H1D
/usr/sbin/slibclean
h1dadm>
saposcol -l
startsap
Despite the fact this was done again with the latest SAPCAR available for AIX 64 / Netwearver 7.0
the result is still the same : the work processes (dw.*) are dieing
this makes me MAD the old kenel that still work fine is 146 - the new is 179 plus "dw 187"
Kind regards,
Raoul
MORE details about the error itself :
The message server disconnect, the error seems to be in the dispatcher :
The Error :
Versions of dplib are different (sys_adm lib= / lib 2410)
dpxxtool 681# lib=241000 krn=232000 DpSysAd bad version
The dplib version defined in the structure sys_adm dplib is different
to your version. This is because the version status of the logged on
program is different to the SAP kernel. Check the version of
the execute programs.
Error (Function,Module,Row) Q0 Z lib=241000 krn=232000 DpSysAdbad verdpxxtool68
ipcs only shows a few root entry, nothing from sidadm, memory was cleared
I created a OSS message and I'm suggesting to focus on the error message
with the dispatcher :
The Error :
Versions of dplib are different (sys_adm lib= / lib 2410)
dpxxtool 681# lib=241000 krn=232000 DpSysAd bad ver
The dplib version defined in the structure sys_adm dplib is different
to your version. This is because the version status of the logged on
program is different to the SAP kernel. Check the version of
the execute programs.
Error (Function,Module,Row) Q0 Z lib=241000 krn=232000 DpSysAd bad ver dpxxtool 681
Please note that the procedure remains the same for getting back to old kernel 146
and there we have no problems at all
Here answer to your question about memory:
ipcs only shows a few root entry, nothing from sidadm, memory was cleared
I created a OSS message and I'm suggesting to focus on the error message
with the dispatcher :
The Error :
Versions of dplib are different (sys_adm lib= / lib 2410)
dpxxtool 681# lib=241000 krn=232000 DpSysAd bad ver
The dplib version defined in the structure sys_adm dplib is different
to your version. This is because the version status of the logged on
program is different to the SAP kernel. Check the version of
the execute programs.
Error (Function,Module,Row) Q0 Z lib=241000 krn=232000 DpSysAd bad ver dpxxtool 681
> I created a OSS message and I'm suggesting to focus on the error message
> with the dispatcher :
> The Error :
> Versions of dplib are different (sys_adm lib= / lib 2410)
That is a kernel internal problem. The versions in the kernel itself are not consistent. Only the development will be able to comment on this.
Markus
I've got an answer from SAP to my customer message and what they suggest is this :
cp -r ./kernelNEW/ /sapmnt/[SID]/exe/* (that is what I did till now)
and add the extra step just after :
cp -r ./kernelNEW/ /usr/sap/[SID]/DVEBMGS<NR>/exe*
This is suggesting that sapcpe it's not working or at least does not
do the required copy from /sapmnt/SID/exe into
usr/sap/SID/DVEBMSG<nr>/exe ?
as the only real big difference in the steps is that they did add a
manual copy to usr/sap/SID/DVEBMSG<nr>/exe
I will try this extra manual copy. I see that not all the files
in /sapmnt/SID/exe are duplicated in usr/sap/SID/DVEBMSG<nr>/exe ... is
there any specif rule to be aware ? br* files from DBATOOL are not there
Here we go, now my problem is solved.
Explanation for any one interested in it is :
the SAP instance where I was trying to do a Kernel Upgrade is one that has been upgraded
from 46C to NW 7.0 / ERP 6.0 by some external consultant.
NW 7.0 have a new structure for the EXE directory
a program is supposed to synchronize /sapmnt/SID/exe with /usr/sap/SID/DVEBMGS<num>/exe
the program used for this is "sapcpe" and it needs some extra parameters in the SAP "start profile"
When this is working fine sapcpe at start of SAP will replicate the kernel in the instance directory
this was not happening in my case :
Solution in case of emergency :
just add a manual copy of the kernel files into usr/sap/SID/DVEBMGS<num>/exe
Final fix:
Check OSS Note 1104735, and make sure the sapcpe replicate mechanism works fine in the future
Hope this will help some of you
Kind regards,
Raoul
You must be a registered user to add a comment. If you've already registered, sign in. Otherwise, register and sign in.