Re: Unable to manually create RAC database

On Dec 16, 3:32 pm, trub3101 <trub3..._at_sky.com> wrote:
> On Dec 16, 3:10 am, gerryt <lepsys..._at_gmail.com> wrote:
>
>
>
> > On Dec 15, 6:24 pm, trub3101 <mich..._at_bassline01.wanadoo.co.uk> wrote:
>
> > > On 16 Dec, 01:03, DA Morgan <damor..._at_psoug.org> wrote:
>
> > > > trub3101 wrote:
> > > > > On 16 Dec, 00:00, "Bob Jones" <em..._at_me.not> wrote:
> > > > >> "trub3101" <mich..._at_bassline01.wanadoo.co.uk> wrote in message
>
> > > > >>news:6522b0fb-d411-4539-bb45-5f6344432ffb_at_e67g2000hsc.googlegroups.com...
>
> > > > >>> On 15 Dec, 19:38, "Bob Jones" <em..._at_me.not> wrote:
> > > > >>>> "trub3101" <mich..._at_bassline01.wanadoo.co.uk> wrote in message
> > > > >>>>news:d09a1798-720c-4d3c-8f91-469f2c40f044_at_x69g2000hsx.googlegroups.com...
> > > > >>>>> Hi Guru's,
> > > > >>>>> I am hoping that someone (anyone!) will be able to provide me with
> > > > >>>>> some assistance with this as my RAC experience is very limited.
> > > > >>>>> I have been trying to manually create a new RAC database <new_RAC_db>
> > > > >>>>> but I keep getting this error message:
> > > > >>>>> create database "<new_RAC_db>"
> > > > >>>>> *
> > > > >>>>> ERROR at line 1:
> > > > >>>>> ORA-01092: ORACLE instance terminated. Disconnection forced
> > > > >>>>> The servers (hosted) came online yesterday with <test_RAC_db> running
> > > > >>>>> (no db build script provided!)however, because of the db_block_size
> > > > >>>>> (too small) this
> > > > >>>>> is of no use to our business needs
> > > > >>>>> I have managed to manually create a single instance database (on
> > > > >>>>> another server) using the script listed (without the RAC thread 2
> > > > >>>>> prerequisites!)
> > > > >>>>> Hopefully, I have provided enough information for someone to point me
> > > > >>>>> in the right direction.
> > > > >>>>> Many thanks in advance
> > > > >>>>> tb3101
> > > > >>>>> ---Architecture.
> > > > >>>>> Two nodes
> > > > >>>>> Linux node1.domain01.com 2.6.9-67.ELsmp #1 SMP Wed Nov 7 13:56:44 EST
> > > > >>>>> 2007 x86_64 x86_64 x86_64 GNU/Linux
> > > > >>>>> Linux node2.domain01.com 2.6.9-67.ELsmp #1 SMP Wed Nov 7 13:56:44 EST
> > > > >>>>> 2007 x86_64 x86_64 x86_64 GNU/Linux
> > > > >>>>> Oracle9i Enterprise Edition Release 9.2.0.7.0 - 64bit Production
> > > > >>>>> With the Partitioning, Real Application Clusters, OLAP and Oracle Data
> > > > >>>>> Mining options
> > > > >>>>> JServer Release 9.2.0.7.0 - Production
> > > > >>>>> ---Build RAC Database (used on node1 only)
> > > > >>>>> ORACLE_SID=<new_RAC_db>11
> > > > >>>>> export ORACLE_SID
> > > > >>>>> sqlplus /nolog<<EOF
> > > > >>>>> connect sys/<password> as sysdba
> > > > >>>>> Connected to an idle instance.
> > > > >>>>> SQL> startup nomount pfile=$ORACLE_HOME/dbs/init<new_RAC_db>11.ora
> > > > >>>>> create database "<new_RAC_db>"
> > > > >>>>> maxinstances 10
> > > > >>>>> maxlogfiles 20
> > > > >>>>> maxdatafiles 100
> > > > >>>>> maxlogmembers 5
> > > > >>>>> character set "UTF8"
> > > > >>>>> datafile '/opt/app/oracle/oradata/<new_RAC_db>/system/
> > > > >>>>> system01.dbf' size 250M
> > > > >>>>> default temporary tablespace temp tempfile '/opt/app/oracle/oradata/
> > > > >>>>> <new_RAC_db>/tmp/temp01.dbf' SIZE 1000M reuse autoextend on next 640k
> > > > >>>>> maxsize unlimited
> > > > >>>>> undo tablespace "UNDO1" datafile '/opt/app/oracle/oradata/<new_RAC_db>/
> > > > >>>>> undo/undo1_01.dbf' size 500m reuse autoextend on next 5120k maxsize
> > > > >>>>> unlimited extent management local
> > > > >>>>> logfile
> > > > >>>>> '/opt/app/oracle/oradata/<new_RAC_db>/rdo/log01a.dbf'
> > > > >>>>> SIZE 50M,
> > > > >>>>> '/opt/app/oracle/oradata/<new_RAC_db>/rdo/log02a.dbf'
> > > > >>>>> SIZE 50M,
> > > > >>>>> '/opt/app/oracle/oradata/<new_RAC_db>/rdo/log03a.dbf'
> > > > >>>>> SIZE 50M
> > > > >>>>> alter database mount
> > > > >>>>> create undo tablespace "UNDO2" datafile '/opt/app/oracle/oradata/
> > > > >>>>> <new_RAC_db>/undo/undo2_01.dbf' size 500m autoextend on next 5120k
> > > > >>>>> maxsize unlimited extent management local
> > > > >>>>> alter database add logfile thread 2
> > > > >>>>> group 1 SIZE 50m,
> > > > >>>>> group 2 SIZE 50m,
> > > > >>>>> group 3 size 50m
> > > > >>>>> alter database enable public thread 2
> > > > >>>>> spool off
> > > > >>>>>>> EOF
> > > > >>>>> ORACLE instance started.
> > > > >>>>> Total System Global Area 1813482504 bytes
> > > > >>>>> Fixed Size 744456 bytes
> > > > >>>>> Variable Size 788529152 bytes
> > > > >>>>> Database Buffers 1023410176 bytes
> > > > >>>>> Redo Buffers 798720 bytes
> > > > >>>>> SQL> SQL> 2 3 4 5 6 7 8 9 10 11 12 13
> > > > >>>>> 14 15 16 17 18 19 20 21 SQL> 2 SQL> SQL>
> > > > >>>>> SQL> /
> > > > >>>>> create database "<new_RAC_db>"
> > > > >>>>> *
> > > > >>>>> ERROR at line 1:
> > > > >>>>> ORA-01092: ORACLE instance terminated. Disconnection forced
> > > > >>>>> SQL>
> > > > >>>> Try running just the CREATE DATABASE statement alone and see if it is
> > > > >>>> successful. Do you have enough space in /opt? If I were you, I would not
> > > > >>>> put
> > > > >>>> datafiles in /opt.- Hide quoted text -
> > > > >>>> - Show quoted text -
> > > > >>> Thanks for your reply Bob.
> > > > >>> I tried just running the CREATE DATABASE statement earlier.
> > > > >>> Unfortunately it returns with an 'ORA-01034: ORACLE not available'
> > > > >>> error message.
> > > > >>> Point taken regarding /opt howevver, this particular filesystem /opt/
> > > > >>> app/oracle/oradata appears to sit on SAN and not the server HDD.
> > > > >>> Thanks again Bob.
>
> > > > >> STARTUP NOMOUNT
> > > > >> then run CREATE DATABASE.- Hide quoted text -
>
> > > > >> - Show quoted text -
>
> > > > > Silly me!
>
> > > > > Thanks Bob. However, I now get this message:
>
> > > > > create database "<new_RAC_db>"
> > > > > *
> > > > > ERROR at line 1:
> > > > > ORA-01092: ORACLE instance terminated. Disconnection forced
>
> > > > > Thanks again
>
> > > > > tb3101
>
> > > > What I think Bob is missing is that this is RAC. And that the RAC
> > > > install MUST push the binaries to both servers and know what to do with
> > > > a lot of other moving pieces. While you are only creating one database
> > > > you must support multiple instances. Even after the binaries are
> > > > installed it isn't vanilla. What are you using for a file system? The
> > > > correct answer is likely ASM have you built an ASM instance on each
> > > > node? Did you do it without using dbca? Or perhaps OCFS, OCFS2, or
> > > > NFS, again no indication from your script.
>
> > > > Have installed and validated the Oracle clusterware with the cluster
> > > > verify tool?
> > > > You can play with this and inflict as much pain and suffering as you
> > > > wish ... but if you fire up dbca ... the installer handles it all for you.
> > > > In my lab we build clusters in far less time than you've spent trying
> > > > to hurt yourself: From operating system to failover testing for a 2 node
> > > > cluster takes less than 6 hours.
> > > These are very valid points which have crossed my mind whilst since I
> > > started all this.
> > > However, without access to DBCA I fear I have no choice but to plough
> > > on and see how far I
> > > can get with a manual install.
> > > I truly appreciate your contributions to this thread and I am by no
> > > means dismissing your
> > > valuable advice. Being the first time I have had anything to do with
> > > Oracle RAC I was expecting
> > > a steep learning curve.
> > Right - First to get 16k blocks out Solaris would take some pretty
> > specialized work. IOW I dont believe it. That would requite some
> > proof : >
> > Secondly how the HEY can you install RAC "successfully" if you cannot
> > even get dbca to run remotely..?
> > Rethink your approach or better yet, hire a consultant!
> Not clear what you mean. Are you suggesting that we should'nt use 16K
> db_block_size database on Solaris?

Apparently it depends. Normally its 8k. Another poster suggested otherwise
that 16k is better in some situations. Were it me, I'd opt for ASM - then its less of an issue
maybe. But some places get scared off by ASM and wont have it. Amazing.
In any case 8K is what newfs lays down. Thats what I was referring to. So it should certainly be a multiple of that starting at 8K methinks.

If you really want to know then set up some tests.. I wonder if would make a measurable
difference. I have some doubts. Therefore choose the default. O I see DA M says that bit about test and verify already : > Received on Sun Dec 16 2007 - 21:01:11 CST