Universe Promotion from BIAR issue

[former_member272418]

Hi experts !

I am facing an issue regarding the promotion of universes.

My landscape consist of 4 systems: DEVELOPMENT, TEST, PRE-PRODUCTION, PRODUCTION.

Software versions are:

• Web application server: Tomcat 6.0.35
• Business Objects Enterprise 14.0.5.957
• DBMS: MySQL 5.1.50

and are the same for all systems.

In DEVELOPMENT system I create the LCM BIAR file containing many objects (web intelligence, crystal report, dashboards, universes), then I

import it in the other systems:

in TEST and PRODUCTION systems everything goes fine but in PRE-PRODUCTION system I get an error ONLY on the promotion of the universes (all

other kinds of object are promoted successfully).

In the Promotion Management Tool I get the error (look at the attached screenshot "error.jpg"):

******

Resolution Status=Copied, Dependency Status=All the required

dependencies

are included., Commit Status=Commit attempted and failed., Promotion

Status=Failure : Object Promoted. Update Failed. Error :

com.businessobjects.mds.repository.exceptions.RepositoryException:

[[error.cmsQueryFailedException] 0]

<RepositoryFileImpl.getDataStream()>,<com.businessobjects.mds.sdk.plugin.desktop.SLMetaDataInfoObject.SLMetaDataInfoObjectException:

DSLMetaDataFile.getFrsContent()>

******

I tried also to import, only in this PRE-PRODUCTION system, a BIAR file containing ONLY one of those universes, but I get the same error.

In attach you find traces of Adaptive Processing Server and CMS of the period while this promotion attempt was running.

Any ideas ?

Thank you very much !

cms_cme2eas01v.CentralManagementServer_trace.000002.txt.zip

aps_cme2eas01v.AdaptiveProcessingServer_trace.000644.txt.zip

[Former Member · Accepted solution]

Please let us know whether the indexing has been turned on at Pre-prod, if yes has indexing also been enabled for universes.

Is it possible to pause the platform search indexing and then try to promote the BIAR files consisting of the universes. This will help the DSL service to have more memory rather than platform index engine and thus might solve the issue.

[former_member272418 · Accepted solution]

Hi Arvind, thank you very much for you answer.

I repeated the promotion with the Indexing stopped but I got the same error.

[Former Member · Accepted solution]

Please share the newly generated logs for APS. It might provide better insights

[Former Member · Accepted solution]

To add please let us know "cme2eas04v:6400" is part of which box.

[former_member272418 · Accepted solution]

Attached you find the two logs generated by the splitted Adaptive Processing Server, created as Asma indicated me.

Adaptive Processing Server 1 supplied the services:

• Adaptive connectivity services
• Bex web application service
• DSL bridge Service
• Multi dimensional analysis service

and Adaptive Processing Server 2 all other services supplied by our unique original Adaptive Processing Server.

Thank you very much for your care !

aps_cme2eas04v.AdaptiveProcessingServer1_trace.000001.glf.txt.zip

aps_cme2eas04v.AdaptiveProcessingServer2_trace.000001.glf.txt.zip

[former_member272418 · Accepted solution]

In this pre-production environment we have 9 machines:

4 running Tomcat

4 running BO4

1 running MySQL

cme2eas04v is the fourth node on which BO4 is running.

The tests I performed with the splitted APS run only on this node !

[Former Member · Accepted solution]

While going through the logs of I find there is a communication drop occurring. Issue can be resolved very easily as updating hosts file entry with FQDN and IP address of all nodes at the same time if it this does not workout, issue would be with how theenvironment has been sized up.

To better understand and for solution we will need to understand how the cluster has been setup exactly. Please explain for each node what would be the APS services in them and explain whether the cms is available in all BO4 nodes or not.

[former_member272418 · Accepted solution]

Hi Arvind

the system is structured as follow:

• 4 machines running Tomcat
• 4 machines running BO4
• 1 node running MySQL
• an hardware load balancer (usually I don't pass thru it because I log directly in one of the tomcats).

In every of the 9 machines (all running Linux Red Hat 5.6 Tikanga as operating system) the file /etc/hosts is filled with every of the other 8 machines "ip-hostname-fqdn".

Eveyone of the 4 BO4 nodes has a CMS and an APS.

Every APS supplies the following sub-services:

• Adaptive Connectivity Service
• BEx Web Applications Service
• Client Auditing Proxy Service
• Custom Data Access Service
• DSL Bridge Service
• Data Federation Service
• Document Recovery Service
• Excel Data Access Service
• Insight to Action Service
• Lifecycle Management ClearCase Service
• Lifecycle Management Service
• Monitoring Service
• Multi Dimensional Analysis Service
• Platform Search Service
• Publishing Post Processing Service
• Publishing Service
• Rebean Service
• Security Token Service
• TraceLog Service
• Translation Service
• Visual Difference Service
• Visualization Service
• Web Intelligence Monitoring Service

[denis_konovalov · Accepted solution]

you should open a support Incident with SAP, something is not right in your pre-prod system.

[former_member272418 · Accepted solution]

What do you mean Denis ?

Unfortunately I didn't perform the installation of this system.

However, I opened an OSS message on 14 of November but, till now, I haven't had a great help from them !

[denis_konovalov · Accepted solution]

you have same content and I'm assuming same biar file importing fine in 2 environments and failing in 3rd.

This means its not the content of the file or file itself, its something that's unique to 3rd env.

[former_member272418 · Accepted solution]

Yes Denis, you're right !

In fact I'm searching for some advices to understand what's wrong with the pre-production environment !

Unfortunately, as I said, the installation was not performed by me.

[Former Member · Accepted solution]

I see that the APS is a deefault being used which includes all services. Could you split the APS and try to promote. Creating new APS including the service mentioned in note  1819897 - Firewall considerations for Promotion Management should help with the sizing.

Disable the remaining APS temporarily to check this activity.

[former_member272418 · Accepted solution]

Hi Arvind

I already tried to split the APS as I said in a previous message in this thread (Nov 21, 2013 2:21 PM) but I got however the issue.

In that case I splitted the APS in a node and stopped the APS in other nodes.

[Former Member · Accepted solution]

What would happen if a Live - Live promotion is done? Is it successfull?

Also can we create a BIAR file from pre-prod and transfer the contents to any other test environment. It would help us know wheter the files related with BAIR might have got corrupted at Repo level and hence a possible repair of the installation would be needed.

Message was edited by: Arvind Pandalai

[former_member272418 · Accepted solution]

Hi Arvind

sorry for being so late !

A live-to-live promotion is impossibile to do because firewalls block communications between environents !

Now I am trying to set up a clone of PRE-PROD env on a virtual machine.

Then I'll import MySQL dump and FSR copy from PRE-PROD on it. I hope to get an (almost) identical clone of this environment.

On it I'll try the same promotions and if it will be unsuccesfull with the same error, I think there won't be no more doubt regading the corruption of the repository or inconsistency between FRS and DB repositories.

I'll let you know as soon as possibile

Thank ALL OF YOU very much !

[Former Member · Accepted solution]

I guess it would be easier if you run repository diagnostic tool rather than simulate your box. follow the wiki How to use Repository Diagnostic Tool - Business Intelligence (BusinessObjects) - SCN Wikito get an idea.

For 4.0 hope it helps http://help.sap.com/businessobject/product_guides/boexir4/en/xi4_bip_repository_diagnostic_tool_en.p...

[former_member272418 · Accepted solution]

Hi guys

according with your suggestion as well as the OSS messagge I tried to run the RDT (repository diagnostic tool).

I had a problem... to run the RDT is mandatory insert the cluster key parameter but I haven't it (and it is impossibile to trace it). So I tried to change it according to SAP note 1613058.

Everything went fine on the first node (I remeber you this is a 4 nodes cluster) but on the other three nodes, trying to change the cluster key, I get the error :

/BOXI/sap_bobj//ccm.sh: line 84:  9585 Segmentation fault (core dumped) $CE_CMDLINE_PREFIX mozjsshell "$CCMFILE" "$@"

Now the Central Management Server of node 2,3 and 4 won't start ! (all other Server Process are ok)

How can I fix that now ???

Please help me !

[Former Member · Accepted solution]

Hello,

The error received when running the ccm.sh script is not really a problem and it is documented in the link below:

http://service.sap.com/sap/support/notes/1340285

Best regards,

Insaf

[former_member272418 · Accepted solution]

I don't understand what you (and SAP) mean with "it is not a real problem" but, after the error, the tool cmsdbsetup.sh quits and DOESN'T CHANGE the cluster key setted on nodes 2,3 and 4 and their CMS don't start anymore !

[Former Member · Accepted solution]

some time I met the unkown java error too, but it will be successfully after trying several times. but if you have this same universe(older version) and the the universe is developed by Designer Tools, the import will be hanged; if you have the older version universe, please delete the oder version universe before importing.

[former_member272418 · Accepted solution]

Finally I was able to run the Repository Diagnostic Tool (I had to recreate nodes 2,3 and 4, but now they are all up&running).

I run the tool without the "-repair" option; in attachment the result.

What do you suggest now ?

Thank you very much.

Best regards !

[Former Member · Accepted solution]

Long time no c .

Well I would suggest you to perform a repair then. I could see and probably you also saw the inconsistencies for various objects in FRS.

Take a backup of CMS, IFRS and OFRS and start the repair.

Hope it works out. (fingers crossed)

[former_member272418 · Accepted solution]

Hi guys

I run reposcan WITH "-repair" option... it found many objects and repaired all of them... but the problem isn't solved: universe promotion gets the same error as before !!!

Do you think that a CMS database REINITIALIZATION may solve the problem ?

Please let me know !

Thank you very much !

[Former Member · Accepted solution]

Yu have raised a SAP incident. Check with support

[Former Member · Accepted solution]

I took some pain to read the entire thread again , instead of re-initializing the CMS DB, it would be better to upgrade to a higher Support Package and verify the problem.

You can upgrade to 4.1 SP2 Patch2. I am not sure whether Patch 3 has been released

Message was edited by: Arvind Pandalai

[former_member272418 · Accepted solution]

I am checking with support... but they are very slow 😞

[former_member272418 · Accepted solution]

thank you Arvind but... don't you think upgrade is a bit exaggerated ?

I have 4 system in my landscape (all at the same release, service pack and fix pack) and 3 of them are working correctly at 4.0 sp5.3 so I can't think this is a release issue !

Do you think that reinitializing Data Source may solve the issue ?

[Former Member · Accepted solution]

Rethinking over my suggestion and previous updates, issue is due to inconsistencies existing presently. So I guess you are quite right. Re-initializing the data source should help.

Have you tried to point the CMS to a backup of CMS Db and find if it works? If it works a possible scenario of some corrupt entries at Db level which can be corrected.

[former_member272418 · Accepted solution]

Ok Arvind !

The CMS DB reinitialize worked !

Now Universe Imports work fine !

[Former Member]

Dear Andrea,

Please make sure that both source and destination environments are on the same patch level.

Please perform these steps:

1. Please split the Adaptive processing server with the following services together under one Adaptive Processing server

Adaptive connectivity services
Bex web application service
DSL bridge Service
Multi dimensional analysis service

2. Restart the connection server

Regards,

Asma

[former_member272418]

Hi Asma, thank you very much for you answer.

Yes, I am sure that both source and destination environments are on the same patch level !

I tried to split the Adaptive Processing Server in the way you described then restart the Connection Server but the promotion gave me the same error.