Howard J. Rogers wrote:
> Glen A Stromquist wrote:
>
> snipped
>
>
>
>>"away it goes" meant that it started fine when I again pointed at the
>>spfile, which shouldnt work, right?
>
>
>
> Correct. Because you *can't* point startup directly at an spfile. For
> example, when I try what you go on to report you did (and I have a genuine
> spfile) then I get this:
>
>
> SQL> startup pfile=$ORACLE_HOME/dbs/spfilelx92
> LRM-00123: invalid character 0 found in the input file
> ORA-01078: failure in processing system parameters
>
> It's expecting a text file; you feed it a binary spfile; it vomits to death.
>
>
>>yes, I get it now, the mystery to me was why did it start when I pointed
>>it to the spfile (shouldnt have) but wouldnt start just by issuing
>>"startup" as it would look for this same file.
>
>
> The fact that you said pfile=spfile.ora (or words to that effect) at startup
> means that your spfile is *not* an spfile, but merely a boring old text
> file which happens to have a name which makes it *look* like an spfile.
> (See above)
>
> Let me guess: when you edited the spfile to add in your three parameters,
> you used vi? Or some other text editor? Therefore, when you saved (or
> whatever weird combination of characters it is to save something in vi
> these days!) what had been a *binary* spfile suddenly became a text file of
> the same name.
>
> So *that's* why startup pfile=spfile.ora works, because you now do in fact
> have a regular init.ora. It's just called something rather unusual. But
> that's fine, because you can call your init.ora
> fredandmargaretathomeonthursday.com if you really had a mind to. So long as
> you specify the right filename with the pfile= clause, it will work.
>
> But when you just issue startup, it doesn't work. Why? Because you don't
> *really* have an spfile, and you don't have an init.ora that's called
> initSID.ora either. So you therefore fit none of the default patterns of
> startup behaviour, and the thing keels over.
>
> So, the quick fix to this mess is simply this. Rename what is currently
> called spfileSID.ora to be initSID.ora. Then issue the startup command on
> its own... I'll lay odds it will now startup just fine. With that done,
> just issue the command "create spfile from pfile", and you'll have a new
> file created called spfileSID.ora which is really a genuine, binary spfile.
>
> And it's precisely because of this utter chaos that can ensue that I'm not a
> wild fan of the spfile. But it is the way of the future, and I suppose we
> had better just get on with it.
>
> By the way, Glen: don't take this the wrong way, but this is one of the best
> examples of a stuff-up with the spfile that I've come across. It never
> occurred to me to test what happens when you have a text-version init.ora
> that's named to look like an spfile. So congratulations: you saved me the
> bother of finding out myself the hard way!!
>
> Best of luck fixing it:
> HJR
Hi Howard - glad I could help!
Fixing it was no problem, by the end of the day yesterday I'd figured
out that compatible=9.2 was needed because of my system tablespace and
it was starting ok, so this morning just re-created an spfile from it
and all's ok, although I'm still reading to find out why I want to do
this over the old init file.
I was using Kedit when I edited the original spfile, which must have
changed it to a text file when saved. At least now I know that a "*." in
front of a parameter does not affect the startup. And not only can you
start with a text-version init.ora thats named to look like an spfile,
it'll start fine when the contents "look" like an spfile as well...
Received on Fri Oct 10 2003 - 10:17:51 CDT