Hallo Daniel,

the database setup seems to be ok.
The param file of the service database exists.
The working directory of the service database exists and can be accessed by the 
database owner.
The database kernel executable exists.
But the database kernel of the service database does not start.

You could try to start the database kernel interactively.
To do this, you need a shell prompt as database ownwer.
You can find the name of the database owner in the file '/etc/opt/sdb'.
There is a line SdbOwner=<database owner>. The default name of the database 
owner is 'sdb'.

You cannot login as database owner, because the account is locked.
But 'su - <database owner>' as root should work. If the database owner has no 
shell, this will fail.

If you are on a Linux system, you can try 'su -s <path to shell> - <database 
owner>'.
<path to shell> must exist in '/etc/shells'. I think 'su -s /bin/bash - sdb' 
should work.

If the 'su' cannot handle '-s <shell>', you can modify the login shell in 
'/etc/passwd'.
You will find the login shell in the last field.

Use the command 'id' to check that the user switch was sucessfully.
Don't care about the missing HOME directory, for the moment.

Change the current working directory to the RUNDIRECTORY of the service 
database,
in your setup type:
cd /var/opt/sdb/data/wrk/.M750031

Set the seach path for shared libraries, in your setup type:
LD_LIBARY_PATH=/opt/sdb/7500/lib:/opt/sdb/7500/sap:$LD_LIBRARY_PATH export 
LD_LIBARY_PATH
or
LD_LIBARY_PATH=/opt/sdb/7500/lib/lib64:/opt/sdb/7500/sap:$LD_LIBRARY_PATH 
export LD_LIBARY_PATH
if your try to start a 64-bit database kernel.

Make sure that there are no resources form preceding database starts.
To do this remove the following directories:
/var/opt/sdb/data/dbspeed/.M750031
/var/opt/sdb/data/diag/.M750031
/var/opt/sdb/data/fifo/.M750031
/var/opt/sdb/data/ipc/db:.M750031
/var/opt/sdb/data/ipc/us:.M750031
/var/opt/sdb/data/pid/.M750031
/var/opt/sdb/data/ppid/.M750031

Do this everytime before you start the database.

Start the database kernel, type:
/opt/sdb/7500/pgm/kernel .M750031

See what happends. If you can't see anything, the databse kernel switched into 
ADMIN mode.
If the the program terminates, but you cannot see any error message, try to 
strace the database kernel.
You must use the follow child switch to trace the database kernel.
You can try something like:
strace -f /opt/sdb/7500/pgm/kernel .M750031 > strace.txt 2>&1

May be a look into strace.txt could help.

Best regards,
Christoph Bautz
SAP Labs Berlin


-----Original Message-----
From: Daniel Rosowski [mailto:[EMAIL PROTECTED] 
Sent: Donnerstag, 2. März 2006 11:39
To: [email protected]
Subject: RE: disk full, now kernel dies before reaching admin state

Hi Tilo,
ok, here's the output:

1497-1 ~ # dbmcli db_enum -s
OK
SAG_VIP /opt/sdb/7500   7.5.0.31        fast    offline
SAG_VIP /opt/sdb/7500   7.5.0.31        quick   offline
SAG_VIP /opt/sdb/7500   7.5.0.31        slow    offline
.M750031        /opt/sdb/7500   7.5.0.31        fast    offline
.M750031        /opt/sdb/7500   7.5.0.31        quick   offline
.M750031        /opt/sdb/7500   7.5.0.31        slow    offline
1497-1 ~ # dbmcli -d .M750031 -u service,service param_directget
RUNDIRECTORY
OK
RUNDIRECTORY    /var/opt/sdb/data/wrk/.M750031

Hope it helps!

Daniel

--
MaxDB Discussion Mailing List
For list archives: http://lists.mysql.com/maxdb
To unsubscribe:    http://lists.mysql.com/[EMAIL PROTECTED]

Reply via email to