Re: dbca error
Date: Thu, 29 Jan 2009 10:16:14 -0800 (PST)
On Jan 29, 2:50 am, macmenco <me..._at_bolt.tk> wrote:
> On Jan 28, 7:47 pm, joel garry <joel-ga..._at_home.com> wrote:
> > On Jan 28, 5:46 am, macmenco <me..._at_bolt.tk> wrote:
> > > Hello all,
> > > I've been playing around with an 11g DB. I installed it on OEL 5.2 (no
> > > updates / no subscription) and did a software only install. Then I
> > > patched it to 126.96.36.199.0. After that I wanted to create a starters
> > > database, so I fired up DBCA, selected a General Purpose database and
> > > said "Go !".
> > > But half way through I got a "Disconnection Forced" message. After
> > > doing some research/testing, the problem apears only in the templates.
> > > Yes, also in the Datawarehouse template. The only way to create a
> > > database, is a custom database. I searched metalink and only found
> > > something suggesting I changed the blocksize to something different
> > > then 8K. Well, I didn't. I even tried adding the template datafiles to
> > > an existing DB and dumped the headers with "alter session set events
> > > 'immediate trace name FILE_HDRS level 10';". The trace file tells me
> > > it is 8K. I even tried copying the template again and dump again.
> > > Still 8K.
> > > Am I missing something, or is there a little bug in DBCA
> > > (188.8.131.52.0) ? Anybody got the same experience ?
> > > Kind regards,
> > > Menco
> > Don't know anything about 11, but am LOL given a flamewar not long ago
> > spread over several blogs and forums about, among other things, oracle
> > not testing blocksizes other than 8K. Could you cite the metalink
> > note?
> > jg
> > --
> > _at_home.com is bogus.
> > Man kills family, self, claimed in suicide note an administrator told
> > the couple on Dec. 23: “You should not even had bothered to come to
> > work today you should have blown your brains out.”http://abclocal.go.com/kabc/feature?section=news/local/los_angeles&id...
> The note ID is : 159324.1
OK, that says not to change it from 8K when using seed db's. I misread your original post.
Mismatches between seed db's and patch sets sounds likely to me, wouldn't be the first time. Also, on several versions and platforms I've seen various things in patches not work because files are kept open. One speculation could be the files were open from an aborted dbca session, then not copied over when patching. Not what I would expect with linux (I would expect the inode would be overwritten for the new file), but programs can always override expectations...
Thanks for the cite.
-- _at_home.com is bogus. ... happened to see some code running that had 48 "AND 1=1)" statements in the where clause. First you have to get the optimizer's attention...Received on Thu Jan 29 2009 - 12:16:14 CST