Re: Oracle 9.2.0 Gentoo Linux - ORA 3113 on DB startup

I have this problem with Linux/Oracle nearly every time I create a db on a new install. You need to increase all of the shm vars before you start as others have said. Also if you are using dbca (ON INITIAL INSTALL, NOT A RUNNING SERVER W/DBS) do an ipcs and delete any segments owned by the oracle account, then try to create the db. This has worked every time for me, do a google search for "Oracle 9i Linux install how-to" and you can find many user written install guides that are far less cryptic than Oracles docs with troubleshooting sections (where I found this workaround).

Where I have a problem is in creating a second database, I get the comm error and nothing will create from dbca. I have then tried creating the db by hand using sqlplus and I get a lower level error about host def not found (something like that, seems like a "root cause" for the 3113 comm error) and Oracle docs define this as a null pointer in an internal structure (i.e. the shmseg it is trying to talk over I'm guessing), I have searched otn forums, google etc etc. and have not been able to find a fix for this. In doing this I have made sure the ORACLE_SID env var is correct for the db I an trying to create, the correct entry is in /etc/oratab, the init<SID>.ora exists and is correct, the pw file is created etc., all before I start the create process. After two days I just gave up and went back to Sybase (regretfully), this problem only seems to rear its ugly head on Linux and NT from the 100's of web pages and forums I have read researching it, none provide any solutions or possible fixes.

"Jasper" <rig0r_at_phreaker.net> wrote in message news:bc94c2df.0302090731.7596e4e4_at_posting.google.com...
> Hi,
>
> I installed Oracle 9.2.0 on my Gentoo Linux box yesterday.
> Installation went
> smooth, but I can't create a database. I get an ORA-3113 on the
> startup nomount command.
>
> SQL> connect / as sysdba
> Connected to an idle instance.
> SQL> startup nomount;
> ... waits for about 30 secs ...
> ORA-03113: end-of-file on communication channel
> SQL>
>
> I've set the kernel parameters as recommended in the installation docs
> :
>
> # cat /proc/sys/kernel/sem
> 250 32000 100 128
> # cat /proc/sys/kernel/shmmax
> 2147483648
>
> The alert log says :
>
> Sun Feb 9 16:19:51 2003
> Starting ORACLE instance (normal)
> LICENSE_MAX_SESSION = 0
> LICENSE_SESSIONS_WARNING = 0
> SCN scheme 2
> Using log_archive_dest parameter default value
> LICENSE_MAX_USERS = 0
> SYS auditing is disabled
> Starting up ORACLE RDBMS Version: 9.2.0.1.0.
> System parameters with non-default values:
> processes = 35
> shared_pool_size = 12582912
> control_files = /data/oradata/DB2/d01/control01.ctl,
> /data/oradata/DB2/d02/control02.ctl,
> /data/oradata/DB2/d03/control03.ctl
> db_block_size = 8192
> db_cache_size = 12582912
> db_file_multiblock_read_count= 16
> db_name = DB2
> optimizer_mode = choose
> star_transformation_enabled= true
> parallel_threads_per_cpu = 4
> parallel_automatic_tuning= TRUE
> query_rewrite_enabled = true
> query_rewrite_integrity = trusted
> pga_aggregate_target = 31457280
>
> Trace file says :
>
> /home/oracle/product/9.2.0/rdbms/log/db2_ora_1717.trc
> Oracle9i Enterprise Edition Release 9.2.0.1.0 - Production
> With the Partitioning option
> JServer Release 9.2.0.1.0 - Production
> ORACLE_HOME = /home/oracle/product/9.2.0
> System name: Linux
> Node name: anubis
> Release: 2.4.20
> Version: #5 Sat Dec 21 16:00:38 CET 2002
> Machine: i686
> Instance name: DB2
> Redo thread mounted by this instance: 0 <none>
> Oracle process number: 0
> 1717
>
> OPIRIP: Uncaught error 1034. Error stack:
> ORA-01034: ORACLE not available
> ORA-27101: shared memory realm does not exist
> Linux Error: 2: No such file or directory
>
> I did an strace of the oracle server process (which is spawned after
> the connect / as sysdba). I repeatedly get this message :
>
> --- SIGALRM (Alarm clock) ---
> rt_sigprocmask(SIG_BLOCK, [], NULL, 8) = 0
> gettimeofday({1044803565, 148746}, NULL) = 0
> setitimer(ITIMER_REAL, {it_interval={0, 0}, it_value={0, 10000}},
> NULL) = 0
> rt_sigprocmask(SIG_UNBLOCK, [], NULL, 8) = 0
> rt_sigreturn(0x1) = -1 EINTR (Interrupted system
> call)
> rt_sigprocmask(SIG_BLOCK, [ALRM], NULL, 8) = 0
> setitimer(ITIMER_REAL, {it_interval={0, 0}, it_value={0, 0}}, NULL) =
> 0
> rt_sigprocmask(SIG_UNBLOCK, [ALRM], NULL, 8) = 0
> gettimeofday({1044803565, 149087}, {4294967236, 0}) = 0
> gettimeofday({1044803565, 149142}, {4294967236, 0}) = 0
> rt_sigprocmask(SIG_BLOCK, NULL, [RTMIN], 8) = 0
> rt_sigprocmask(SIG_BLOCK, [ALRM], NULL, 8) = 0
> gettimeofday({1044803565, 149291}, NULL) = 0
> setitimer(ITIMER_REAL, {it_interval={0, 0}, it_value={1, 20000}},
> NULL) = 0
> rt_sigprocmask(SIG_UNBLOCK, [ALRM], NULL, 8) = 0
> semop(131073, 0xbfffca00, 1) = -1 EINTR (Interrupted system
> call)
>
> Looks like semop() is failing. But why...
>
> Any help ?
>
> Thx !
Received on Sun Feb 09 2003 - 16:09:16 CST