Upgrade Portal from NW 7.0 to NW 7.31

[georg_salmen]

Hi,

I am currently running into this error in SUM when attempting to upgrade from NW 7.0 to NW 7.31.

I am currently using SUM ver 1.0 SP8_1

Stack XML file was generated on a SOLMAN 701

An error has occurred during the execution of the Preprocessing (start shadow instance SCS)

Trouble Ticket Report:

An error has occurred during the execution of the Start shadow instance SCS step with action pre-execute. Service com.sap.sdt.j2ee.services.servicesimpl.EngineControllerService failed with the following message:

Expected to find an instance located on host n/a, but none was detected before that

You can find more information in the files /usr/sap/EPX/SUM/sdt/log/SWITCH/START-SCS-SHADOW_05.LOG and ProcessOverview.html.

Use the information provided to trouble-shoot the problem. There might be an OSS note providing a solution to this problem. Search for OSS notes with the following search terms:

SWITCH,pre-execute,install-shadow,start-scs-shadow,com.sap.sdt.j2ee.services.servicesimpl.EngineControllerService,class com.sap.sdt.util.diag.DiagException

Error:

AM [Error ]: The following problem has occurred during step execution: com.sap.sdt.util.diag.DiagException: Expected to find an instance located on host $(0), but none was detected before that

Has anyone had this error? about every helpful tip would be very grateful!

Thanks

Georg

[georg_salmen · Accepted solution]

Hi all,

Sorry for the late reply, the problem is fixed and the upgrade was successful.

Solution for the first error

Error: Expected to find an instance located on host $(0), but none was detected before that

SAP has sent me a modified file (sdt_sapjup_services.jar)

1. Revert the system

2. Unpack SUM SP8 PL1 and replace the old ...\SUM\sdt\lib\sdt_sapjup_services.jar with the new      one

3. Start SUM

The second error

Error: Could not start SAP instance with number 79. Could not send the command to start the instance with number 79 on host fmgepx. No instance was found in the system instances list with number 79 and host name fmgepx

The solution is to add the following rows:

SAPLOCALHOST = fmgepx

SAPLOCALHOSTFULL = fmgepx.munich-airport.de

in

/SUM/sdt/EPX/SYS/profile/EPX_JC78_fmgepx

and

/SUM/sdt/EPX/SYS/profile/EPX_SCS79_fmgepx

Then restart sapstartsrv

sapcontrol -nr 78 -function RestartService

sapcontrol -nr 79 -function RestartService

Check if the correct host is in the list.

sapcontrol -nr 78 -function GetSystemInstanceList

sapcontrol -nr 79 -function GetSystemInstanceList

Then repeat the failed step in SUM.

The third error

Error: Could not update components. Deployment of queue Queue_2013.08.13 at 10:45:37 completed with error Could not connect to Deploy Controller. Make sure that  AS Java and Deploy Controller service are  running. com.sap.engine.services.dc.api.ConnectionException: [ERROR CODE DPL.DCAPI.1144] Could not establish connection to AS Java on [fmgepx:57804]. No alive connection. Check state of the server. Check if the host and port specified are correct and AS Java is up and running. javax.naming.NamingException: Exception during getInitialContext operation. Cannot establish connection to the remote server. [Root exception is com.sap.engine.services.security.exceptions.BaseLoginException: Exception in creating new RemoteLoginContext instance.] com.sap.engine.services.security.exceptions.BaseLoginException:

Exception in creating new RemoteLoginContext instance. com.sap.engine.services.rmi_p4.P4IOException: No alive connection.Check state of the server

In the SAP Management Console I've stopped and restarted the system, the upgrade could continue without error.

Thanks again to all who have contributed so many ideas!

Regards,

Georg

[Former Member]

Just an FYI. In our company we previously used SUM Sp07 patch05 to upgrade our sandbox EP system (NW 7.0) to target 7.31 and we didn't have this issue with the "host $(0)" message.

Now we were upgrading our DEV EP system using SUM SP08 patch02 and is when we encountered this issue, so apparently yes ... the SP08 patch02 is buggy.

We have a message opened with SAP to see if they could provide us a solution or the sdt_sapjup_services.jar file that Georg mentioned as there is no note for this. However we might attempt to restart the SUM with SP07 patch05 and will let you know the results.

Cheers

[Former Member]

Update. We just downloaded and used SUM1.0 SP08 patch 3 (which was just released yesterday) and we were able to bypass the error:

                    Expected to find an instance located on host $(0)

So far no additional issues for this SUM specific version.

Regards

[Former Member]

I can also confirm the same result.

The SUM handling of the virtual hosts is significantly better.

Currently in the process of two upgrades. Will update when they are complete and OK.

But none the less the issue reported in this blog is resolved for now.

[Former Member]

Hi Ross, so I guess you are using SUM SP08 patch3 correct? If so how is it going?

We encountered an error in the EXECUTION phase maybe like 70% through. Error follows:

Could not load usage type type data from data source jdbc/pool/EPD. com.sap.sl.ut.infoprovider.exceptions.UTInfoException: The SQL Statement is wrong : SELECT * FROM BC_SL_PRODUCTS com.sap.sl.ut.infoprovider.exceptions.UTInfoException: The SQL Statement is wrong : SELECT * FROM BC_SL_PRODUCTS java.sql.SQLException: Invalid column name

Has anybody seen this error before? We checked at DB level and the table exists, we are even able to query it without issues. I believe by some reason the jdbc is having issues to reach our SAPSR3DB database schema. We opened a message to SAP but if someone has any idea I'd appreciate if you could help us.

Thanks

[matus_petrek2]

Hi Jesus,

what about your Oracle client? Are you on latest version 11, or still using version 10?

Also check if the SAPSR3DB user is not locked and/or if OPS$ mechanism works correctly.

Regards,

Matus

[Former Member]

Hi Matus, thanks for the reply.

We have Oracle 11.2.0.2 with instant client 11202. SAPSR3DB was not locked and OPS$ mechanism was fine.

Reattempting to run the SUM from the point we where (repeating the step) wasn't working so we stopped and restarted SUM and then it worked. I feel this could be a new bug with SUM 08 patch 03 as we never have this issue in the previous upgrade we did to an exact copy of the system on which we are working on, on which we used SUM 07 patch 05.

The important thing is that the upgrade completed after a really painful process. LMDB+SUM08 ..... oh man!

[Former Member]

Hi Jesus

I agree lmdb + SUM8 p1 and then SUM8p3 have been a pain, im running the upgrade for EP but to NW7.4.

Luckily for me im running this one week after you guys, so it was great to find that the SP3 solves the detect instance issue.

For the Could not load usage type type data from data source error that you mentioned, i remember doing the same thing you did (stop SUM and relaunch it) couple of months ago for PI upgrade to 7.31, I havent faced this yet for EP, but ill confirm if it happens to me as well.

Thanks!

G

[Former Member]

Hi Jesus,

Oh yes SUM is just a whole lot of fun. Actually I can't be too disparaging as I mostly like SUM IF I can work around the bugs. I do believe SAP is developing a good tool that covers most scenarios. Which in the end makes it easier to do a range of tasks. However releasing a tool for us to test - which is basically what the early versions of SP08 were - that is not so cool.

And like you I have found the LMDB not so user friendly. Oh setting up for a BW upgrade or other work - I have it sorted now but getting there is painful. Once again I quite like the new setup but to get it to work is just such a task that it does leave a somewhat bad taste in ones mouth.

Glad you got through the Upgrade. I normally repeat a process at least three times before I consider it an issue and go looking for a resolution. And like you - Stopping SUM and restarting it - also restarting the instance can really help.

I have appreciated everyone's input.

[Former Member]

Hi Georg,

If the issue still exists, please look into the folowing:

1. Add IP and hostname to /etc/hosts
2. Check if you have enough memory
3. Check Env. Vars. "PATH"
4. Update Kernel to 720

Regards,

Mishal

[georg_salmen]

Hi,

Thanks for the many tips, unfortunately, the problem still exists.

From the SAP I finally get a constructive answer, it is an error in SUM.

I installed the patch from the SAP, unfortunately now occurs a similar error.

The following problem has occurred during step execution: com.sap.sdt.util.diag.DiagException: Could not start SAP instance with number 79.

Could not send the command to start the instance with number 79 on host fmgepx.

No instance was found in the system instances list with number 79 and host name fmgepx.

I tried to start the Shadow SCS instance manually, but does not help.

sapstart pf=/usr/sap/EPX/SUM/sdt/EPX/SYS/profile/EPX_SCS79_fmgepx

Has anyone had this error? about every helpful tip would be very grateful!

Thanks

Georg

[Former Member]

Hi,

Please the below note to check the issue.

1897758 - Troubleshooting abap instance startup failure issue

Thanks

Rishi Abrol

[Former Member]

Hi,

Can you please update the log also.

Start the shadow SCS instance using same command lines from Upgrade directory log
        ...java/log/START_SHD_SCS_THS_0X.LOG:

          Example:
          /<upgrade directory>/java/data/kernel/sapcontrol
          -prot NI_HTTP -nr XY -function Start

       If any issue with running sapcontrol refer to Note 1565645 -SAP composite note: sapcontrol & KBA 1664214 - Upgrade is not able to start the shadow instance because of an "HTTP error 401

1797346 - To start AS JAVA shadow instance during Phase DEPLOY_ONLINE_SHD_DM or START_SHD_J2EE_DM manually

Thanks

Rishi Abrol

[Former Member]

Even if you update the kernel and stop the sapstartsrv and fix hostnames (since you need to if you have a logical host it still does not work.

Running the command  

/<upgrade directory>/java/data/kernel/sapcontrol  -prot NI_HTTP -nr XY -function Start

it works but SUM fails to recoginise that the system has started. you can work around this by modifying the control files. Basically just telling SUM this step is completed

[Former Member]

Hi Georg,

Did you fix the issues.

Thanks

Rishi Abrol

[Former Member]

Hello,

I had a similar issue with SUM SP08. I have used SUM SP07 for upgrade from NW 7.01 to NW .31 and it works:)

Regards

Ankit Jain

[Former Member]

I have had exactly the same issue with SUM SP08_0, SP08_1 and SP08_2 in each case SUM SP07_7 resolved this issue. I logged the call with SAP and just could not wait for their response. Thought they would have a resolution for this by now.

[Former Member]

Actually I take that back. I thought SP07_7 would work for the portal upgrade (7.02 to 7.31) but it failed for other reasons. As such I have logged a call with SAP for resolution.

Using SP07_7 worked for a 7.0 to 7.31 PI upgrade

For portal upgrade using SP08_2 I am getting the same error.....

Error: Expected to find an instance located on host $(0), but none was detected before that

Hopefully I can get a quick resolution.

[matus_petrek2]

Hello Ross,

my coleague had the same issues. Check out the note 1898687 - is this also your case?

Also, please try to edit profiles and unify the hostname, check out SAPLOCALHOST parameter and if possible set it to virtual hostname (if you use one).

Also, do the same for the shadow instance profiles - it is possible that it is not set there correctly or missing.

Regards,

Matus

[Former Member]

Hi Matus,

Thanks for the ideas. In our case here we have already checked and converted the profiles with the virtual hostnames for the shadow instance profiles. For the original instance - we are just up to preprocessing so merging those profiles would be a little premature.

SAPLOCALHOST is set in all profiles - also SAPLOCALHOSTFULL has been set.

Once all these changes have been made it is possible to start the shadow SCS manually. Problem is SUM does not recognise that it is running. You get the same original error

Error: Expected to find an instance located on host $(0), but none was detected before that

Since I have followed the same procedure as before when it worked on SP07_7 - I really think this is an issue with the SUM program. SAP is looking at it now.

As mentioned I also reverted and started again with SP07_7 but had different issues so I have decided to persist with the newer version (SP08_2) as I have a lot of upgrades to complete. I expect SP08_3 will be released very soon. Perhaps this will resolve things.

[matus_petrek2]

Hi Ross,

I know is late for the step of merging profiles, just wanted to know if SUM did this step correctly.

And I completely agree, seems that SUM SP8 is full of bugs.

Regards,

[Former Member]

Hi Georg,

I found another thread from a few months back, issue was not resolved guy switched from SUM to usual sapinst method for his upgrade. However one of the guys did mention at the end of the thread:

http://scn.sap.com/thread/3175233

I also faced the same issue.I stopped the system and deleted all the shared memory including the memory and semaphores using the ipcrs command and after that the SUM tool worked fine.

I think it was due to the shared memory entry that the sapcontrol was detecting the wrong instance

Have you tried maybe stopping your SUM processes, doing cleanipc of all revelant ports and instance numbers and then perhaps retrying?

Kind Regards.,

Johan

[georg_salmen]

Hi johan,

can you explain to me precisely what commands you've executed exactly.

I've rebooted the server, but that should also deleted all of the shared memory including the memory and semaphores, or?

Regards,

Georg

[Former Member]

hi,

you will have to do a cleanipc of your instance numbers.

cleanipc 00 remove (if your system number is 0 you use this)

cleanipc 10 remove (if your system number is 10 you do this)

cleanipc 99 remove (this removes shared memory for saposcol make sure all is stopped before)

It should list all areas to be clean example:

http://scn.sap.com/docs/DOC-27773

Kind Regards,

Johan

[matus_petrek2]

Hello Georg,

is this pure Java or Dual-stack? Which platform is it?

It seems the problem is related to the hostname - is this virtual machine?

Can you post more details from the logs?

Regards,

Matus

[georg_salmen]

Hi Matus,

it is only Java, the platform ist Linux on x86_64 64bit, it is a virtual machine.

Which logs you need exactly?

Thanks for your answer!

Regards,

Georg

I have attacht the TroubleTicket & START-SCS-SHADOW Log file.

TroubleTicket_36.txt.zip

START-SCS-SHADOW_31.LOG.zip

[matus_petrek2]

Hi Georg,

you can find more logs in /usr/sap/EPX/SUM/sdt/log/. From the logs you've sent one cannot see any details.

As this is linux, as adm user execute command "niping -v" and see if the hostname is resolved correctly.

In addition check parameter SAPLOCALHOST in your profile, the problem is most probably caused by virtual hostname set instead of physical hostname.

Regards,

Matus

[Former Member]

Hi,

Can you please try to stop all the processes and deleted all the shared memory including the memory and semaphores using the ipcrs command and after that the SUM tool worked fine.

I think it was due to the shared memory entry that the sapcontrol was detecting the wrong instance.

can you please try to get the output of the below command from sidadm

sapcontrol -nr 00 -function GetVersionInfo

Thanks

Rishi Abrol

[georg_salmen]

Hi Matus,

I obtained the following results when I run the command with the sidadm user.

niping -v

Hostname/Nodeaddr verification:
===============================

Hostname of local computer: las9                             (NiMyHostName)

Lookup of hostname: las9                                     (NiHostToAddr)
    --> IP-Addr.: 1**.***.***.***
Lookup of IP-Addr.: 1**.***.***.***                          (NiAddrToHost)
    --> Hostname: las9

Lookup of hostname: localhost                                (NiHostToAddr)
    --> IP-Addr.: 1**.***.***.***
Lookup of IP-Addr.: 1**.***.***.***                                (NiAddrToHost)

The parameter SAPLOCALHOST I have checkt, is always entered fmgepx.

Regards,

Georg

[georg_salmen]

Hi,

Thanks for your help!

I reboot the server, the same result.

The result of the command.

sapcontrol -nr 00 -function GetVersionInfo

07.08.2013 14:57:32
GetVersionInfo
OK
Filename, VersionInfo, Time
/usr/sap/EPX/JC00/exe/sapstartsrv, 700, patch 310, changelist 1261844, optu (Jul 10 2011, 20:16:40), linuxx86_64, 2011 07 10 20:27:28
/usr/sap/EPX/JC00/exe/disp+work, 700, patch 310, changelist 1261844, optu (Jul 10 2011, 20:05:13), linuxx86_64, 2013 05 28 13:26:58
/usr/sap/EPX/JC00/exe/gwrd, 700, patch 310, changelist 1261844, optu (Jul 10 2011, 20:05:13), linuxx86_64, 2013 05 28 13:26:59
/usr/sap/EPX/JC00/exe/msg_server, 700, patch 310, changelist 1261844, optu (Jul 10 2011, 20:16:40), linuxx86_64, 2013 05 28 13:26:59
/usr/sap/EPX/JC00/exe/dbadaslib.so, 700, patch 310, changelist 1261844, optu (Jul 10 2011, 20:05:13), linuxx86_64, 2013 05 28 13:26:58
/usr/sap/EPX/JC00/exe/dboraslib.so, 700, patch 310, changelist 1261844, optu (Jul  8 2011, 20:48:19), linuxx86_64, 2013 05 28 13:26:58
/usr/sap/EPX/JC00/exe/enserver, 700, patch 310, changelist 1261844, optu (Jul 10 2011, 20:16:40), linuxx86_64, 2013 05 28 13:26:58
/usr/sap/EPX/JC00/exe/icman, 700, patch 310, changelist 1261844, optu (Jul 10 2011, 20:16:40), linuxx86_64, 2013 05 28 13:26:59
/usr/sap/EPX/JC00/exe/jcontrol, 700, patch 310, changelist 1261844, optu (Jul 10 2011, 20:16:40), linuxx86_64, 2011 07 10 20:44:49
/usr/sap/EPX/JC00/exe/jlaunch, 700, patch 310, changelist 1261844, optu (Jul 10 2011, 20:16:40), linuxx86_64, 2011 07 10 20:44:50

Regards,

Georg

[matus_petrek2]

Hi,

I understand that fmgepx is your virtual hostname and las9 is your physical hostname, but is fmgepx also defined in /etc/hosts? Can you try to ping this host to see if it is known to your server?

You have an option to adjust /etc/hosts if the host is unknown or to change SAPLOCALHOST parameter.

Regards,

Matus

[georg_salmen]

Hi,

las9 is also virtual, we use a NIS, las9 and fmgepx I can ping.

I do not know if the info is Helpful?

SUM created the shadow profiles on the basis of las9 than fmgepx, attacht the profiles.

Regards,

Georg

EPX_JC78_las9.txt.zip

DEFAULT.txt.zip

EPX_SCS79_las9.txt.zip

[Former Member]

Hi,

Is there any reason why your system is stil having the old kernel.

Kernel 7000 is out of maintenance . Can you try to upgrade the kernel to 720 and then try running the SUM tools.

Note 1636252 - Installing a 7.20 kernel in SAP Web AS 7.00/7.01/7.10/7.11

Thanks

Rishi Abrol

[matus_petrek2]

Hello Georg,

I found it weird to use two virtual hostnames, what is their connection? Is this cluster environment?

Why you are using fmgepx while being on las9?

I've reviewed your profiles but you are not using SAPLOCALHOST, this is taken by default as fmgepx.

Please add the parameter SAPLOCALHOST=las9 to your DEFAULT profile, restart the system and retry in SUM.

Regards,

Matus