cancel
Showing results for 
Search instead for 
Did you mean: 

Problem with start of DB in distributed environment

Former Member
0 Kudos

Hi all,

I hope someone can help me.

We have a problem with the startup of our sap instances when the database is running on another node.

The problem is, that the startdb doesn't find the remote database and start the database local on the node.

I have a flexframe infrastructure and all sap instances are single services like db, scs, ci, app

The Flexframe starts the services step by step in the correct sequence.

In the first step flexframe starts the database on the remote host correctly and

in the next step flexframe starts the ci.

In the protocols I find the successfull checks with R3Trans and TNSPING.

In the step "sapcontrol starts the instance" is implemented a wait of 10 seconds.

During this windows the startdb-script is running.

The tnsnames adapter resolves the alias correct.

After that step the function "check_db_running"

is running and this function starts the database local.

At the moment we stop the unconnected database-instance and the system works without any problems.

Here the protocols:

The startsap.trc has following infos:

    }09:40:47 ## restore_path() done

    EXEDIR=/usr/sap/SID/DVEBMGS00/exe

    EXECMD=/usr/sap/SID/DVEBMGS00/exe/R3trans

   }09:40:47 ## getExePath() done: 0

   DBCONNECT_TEST=/usr/sap/SID/DVEBMGS00/exe/R3trans -d -w

...

   }09:40:48 ## restore_path() done

   Database is running

   DB_RUNNING=1

   JDB_RUNNING=

  }09:40:48 ## set_db_running() done: 1

  {09:40:48 ## set_sapjdbhost() start

....

   }09:41:02 ## sapstartsrv_do() done: 0

   starting SAP Instance DVEBMGS00

   Startup-Log is written to /home_sap/sidadm/startsap_DVEBMGS00.log

   /usr/sap/SID/DVEBMGS00/exe/sapcontrol -prot NI_HTTP -nr 00 -function Start

   wait sleep 10 while instance is starting

   Instance on host hostname started

  }09:41:14 ## instance_do() done


_______________________________

STARTDB.LOG

------------------------------ Thu Oct 23 09:41:03 CEST 2014

checking V2 connect

TNS Ping Utility for Linux: Version 11.2.0.3.0 - Production on 23-OCT-2014 09:41:03

Copyright (c) 1997, 2011, Oracle.  All rights reserved.

Used parameter files:

/sapmnt/SID/profile/oracle/sqlnet.ora

Used TNSNAMES adapter to resolve the alias

Attempting to contact (DESCRIPTION = (ADDRESS = (COMMUNITY = SAP.WORLD)(PROTOCOL = TCP)(HOST = xxxx)(PORT = x

)) (CONNECT_DATA = (SERVER = DEDICATED) (SERVICE_NAME = SID)))

OK (0 msec)

tnsping: V2 connect to SID

------------------------------ Thu Oct 23 09:41:03 CEST 2014

Connect to the database to check the database state:

-------------------------------------------------------------

Thu Oct 23 09:41:03 CEST 2014

check if Oracle processes are running

database not available

There are no Oracle processes running -

Database is probably already stopped.

check_db_running: check finished with return code: 10

Database not available

____________

Has anybody an idea where is my problem?

Best regards

Peter

Accepted Solutions (0)

Answers (2)

Answers (2)

former_member185239
Active Contributor
0 Kudos

Hi Peter,

You need to check the DEFAULT profile. If sapmnt is not mounted then you might face this issue.

Also maintain the SAPDBHOST as environmemt variable or in DEFAULT profile and then give a retry to start the system.

Also do mainatian the ORACLE_HOME , ORACLE_SID environment variable.

With Regards

Ashutosh Chaturvedi

Former Member
0 Kudos

Hi Ashutosh,

thank you for your feedback.

SAPDBHOST is correct implemented in the DEFAULT profile and the envionment variable of ORACLE_HOME and Oracle_SID are correct.

The system is working correctly after a start, only I have two running database services.

The remote database service is connected to the datafiles.

The database service on the ci server hast no connection to the files

My problem is, why cannot connect the startdb script to the remote database.

Is my problem, that the ssh to the remote server not works?

Best regards

Peter

Former Member
0 Kudos

Hi Peter

I have documented some startup problems in FlexFrame environment.  Maybe some of these can be helpful.

- Startup problems

-DB not started , tnsping failed:  Different Oracle tnsnames/sqlnet config

between <sid>adm and ora<sid> 

                      

- SAP services are tried to start on different node than were virtual IP is

running -> Check that virtual IP is started on right node

                       

- Check that listener name is in /FlexFrame/scripts/ora_listener_names file and listener running

                        

- Hanging TCP-session streams in /var/tmp

                        

- After upgrade update SAP version info to FlexFrame ldap configuration

                       

- After upgrade - copy new .*.csh environment files for <sid>adm and ora<sid>

                        

- Run ff_service.sh  -s sid -a cleanup -t service for every service before starting

              

Ssh is needed for BR-tools to work properly but should not affect startup.

But recommend to activate ssh without password configuration between ci/app and database node.

You can also test to start SAP system via standard start commands (when agent stopped on node).

Br. Jari

Reagan
Product and Topic Expert
Product and Topic Expert
0 Kudos

Are you able to start the DB and SAP separately ? As far as I know on a distributed setup the start of the database and SAP is done separately.

Former Member
0 Kudos

Hi Benjamin,

yes , it is able to start the services separatly.

The flexframe scripts starts every service(db, ci and app-serivce) separatly.

The problem ist, that the database is running and in the next step the ci is starting.

In the new 7.4 profile of the ci we have an execute of startdb script.

This startdb checks first the connect with tnsping. This works and the retuncode is 0.

In the next step the script start the step "check_db_running".

This step can not connect to the database.

....

check if Oracle processes are running

database not available

There are no Oracle processes running -

Database is probably already stopped.

check_db_running: check finished with return code: 10

Database not available

...

A workaround is to comment out the startdb in the profile and with this solution we worked

in the last month.

But in this weekend we start the next upgrade and the same problems are occuring.

Now I will find a correct solution.

Best regards

Peter