Java stack does not start after system copy

[Former Member]

Hi!

I've recently performed an homogeneous copy (Productive System to Quality System) of a dual stack system (JAVA + ABAP). The copy was not by SAP's method (sapinst, migration controller), but simply by restoring database, generating control file... This works flawlessly for ABAP stack, but we're having issues with JAVA stack.

As expected the Java instance ID stored in database from Productive System, differs from the one in Development System, and Java can't be started.

Apr 16, 2009 4:56:56 PM -

---

Apr 16, 2009 4:56:56 PM Bootstrap MODE:

Apr 16, 2009 4:56:56 PM <INSTANCE GLOBALS>

Apr 16, 2009 4:56:56 PM determined by parameter ID0093287.

Apr 16, 2009 4:56:56 PM -

---

Apr 16, 2009 4:56:59 PM Exception occurred:

com.sap.engine.frame.core.configuration.NameNotFoundException: A configuration with the

path "cluster_data/dispatcher/ID9328700" does not exist.

Is there a way to get around this, to change the instance ID, without having to use sapinst export method? If not what would be the quickest way to get Java back?

Thanks in advance.

Best regards,

Daniel Garrido

[markus_doehr2]

> I've recently performed an homogeneous copy (Productive System to Quality System) of a dual stack system (JAVA + ABAP). The copy was not by SAP's method (sapinst, migration controller), but simply by restoring database, generating control file... This works flawlessly for ABAP stack, but we're having issues with JAVA stack.

This is unsupported and technically not documented.

For Java system copies you have to use sapinst; java stores lots of its configuration on the filesystem. If you just copy that you'll have the old hostname and SID all spread over, the secure store is not accessible etc.

So do the Java part as documented in the system copy guide.

Markus

[Former Member]

Hello Daniel.

we're using the same method than you for our system copy.

It's not supported by SAP but it works if you know exactly how to reconfigure the complete java stack.

One major issue is to set up the instance ID. Normally, you just have to configure it at your first refresh. The next time, your quality system ID would be the same.

First of all, change all occurences ( name of the server, SAP instance, cluster ID, etc... ) of your productive system in following files:

/usr/sap/SID/DVEBMGSxx/j2ee/cluster/bootstrap/bootstrap.properties

/usr/sap/SID/DVEBMGSxx/j2ee/clusterinstance.properties

/usr/sap/SID/DVEBMGSxx/SDM/program/config/sdm_jstartup.properties

/usr/sap/SID/DVEBMGSxx/j2ee/cluster/dispatcher/cfg/kernel/<Name of the services>.properties

/usr/sap/SID/DVEBMGSxx/j2ee/cluster/serverX/cfg/kernel/<Name of the services>.properties

Then, launch Configtool and change your connexion parameters. ( Just answer ' No ' where you're prompted to use the default connexion ).

Now, using the search option of configtool, search for all occurences of your production SID, server, and system number ( if it's not the same than your production ), and replace it with the corresponding information of your server.

( Note that you can bypass this step if you export your Configtool configuration before the refresh and import it after restoring the sapdatas ).

You must switch to Configuration editor mode to edit propertysheet instance.properties.IDxxxxxx ( under Cluster_data ) replacing productive occurences with quality informations.

If the applications doesn't start afterall, check the properties files again at OS level, because they could have been replaced during the restart, and find informations in dev_ files and server logs to solve remaining issues.

Don't forget to reinstall your license before restarting...

Hope this could help you.

Seb.

[Former Member]

Hi Sebastien,

One major issue is to set up the instance ID. Normally, you just have to configure it at your first refresh. The next time, your quality system ID would be the same.

Can you please provide the steps required to configure it at the first refresh. There is no instance ID available and I am not sure how to generate one.

Please advise. Thank you.

Edited by: One Idea on May 24, 2009 11:15 PM

[Former Member]

Hi Gurus

I am facing the same issue but my target system is also on a cluster with two nodes.

Every thing went fine and the central instance installation was completed successfully but i cannot see java processes at all.

We followed the recommended method of system copy and it works every time.

But when we restore the database again then java doesn't come up.

Can anyone help on this issue.

Thanks and Regards

Khalid

[markus_doehr2]

> I've recently performed an homogeneous copy (Productive System to Quality System) of a dual stack system (JAVA + ABAP). The copy was not by SAP's method (sapinst, migration controller), but simply by restoring database, generating control file... This works flawlessly for ABAP stack, but we're having issues with JAVA stack.

This is your problem.

As soon as Java comes into place you must use sapinst.

An AS-Java (or an Addin) is very different to ABAP. ABAP stores EVERYTHING in the database, so copying that and providing a filesystem structure and a profile is sufficient to get a system back up again.

Java stores parts of its configuration in the database only, most of the Java AS is on the filesystem - including configurations in .properties files that are synchronized with the database when the AS is starting. It's VERY important that filesystem and database are backed up at the same time and that during the time the backup is running nobody is deploying/changing anything on the Java side to keep database and filesystem in sync.

That is one major drawback when you have a double stack installation, copies can't be done easily any more by just restoring database backups.

Markus

[Former Member]

Hello Daniel,

there is one more thing that you can do to resolve this error

since you have copied production to qa,I assume there will be more server processes in prod as compared to qa,so what you can do is remove on server process in qa through config tool and then start SAP,this may help in removing the memory issues

Rohit

[ganesh_s7]

This message was moderated.

[Former Member]

Hi Ganesh,

>What might possibly work is that you can try changing the NodeID (or Instance ID) from the Productive System Id to the Quality sytem id manually in all Cluster & local Server and Dispatcher process using 'config tool'. But then that is a Long List of Folders/sub-folders that you have to manually change. For Practical reasons, i would suggest that you repeat the process using R3load(Migration Monitor) for export in PDN & use the same for import in QAS.

Please be aware that this method is not supported by SAP.

Also, please read the Rules of Engagement and do not ask for points in Forums.

Best Regards,

Matt

[Former Member]

Hi,

Please check SAP notes 966752 & 831812. It may help you.

Also check below thread.

Thanks

Sushil

[Former Member]

Hi,

Been there, done that, got caught also....

We use now a mixed way of system refresh for Oracle based dual stack systems.

We restore the database fot the ABAP stack as usual and we reimport with sapinst an export of the Jva schema owner.

This one more reason to NOT use dual stack systems and to install separate java stacks !

It's too late for us now for our BI systems....

Regards,

Olivier

[Former Member]

Hi Oliver:

I'm curious. For copying your Java stack you chose the following menu path (or something similar) in SAPINST for the JAVA export from your dual-stack source?

System Copy -> Oracle -> Source System -> Distributed System -> Based on AS Java -> Java Instance Export

Regards,

Wayne

[markus_doehr2]

> System Copy -> Oracle -> Source System -> Distributed System -> Based on AS Java -> Java Instance Export

Use Central System - based on ABAP + Java

Markus

[debasissahoo]

Hi Daniel,

System copy on ABAP+Java system with simple database restore is not supported by SAP. so even if you will be able to make the java up, SAP might not support any future error on your quality system!

The safest way to bring java back is to do a system copy based on SAPs supported method.

But however, You can try this, as you've already done the system copy.. you can test if this works or not. I see in your post two different cluster IDs.. ID0093287, ID9328700.

Not sure which one is from production system.. assuming ID9328700 is of your production, you 've to go to QAS configtool and here you see a different cluster ID. Change the cluster ID in QAS using the configuration editor. restart the java cluster.

Regards,

Debasis.

[Former Member]

The java stack always does have components in the filesystem located in /usr/sap/SID/<inst>/j2ee. The supported sapinst method takes care of this.

Well you can try to copy the files manually, but as said you are on your own then.

Best regards, Michael